Portfolio & BA control room

Choose the right change before it becomes delivery work.

This is GAB's front door for serious initiatives: define the need, justify the business case, confirm benefits, authorize the charter, then promote only approved work into Projects.

G0-G2 pre-project governanceLight / Standard / Full tailoringOperations feedback returns here

Operating rule

No serious initiative enters Projects before G2.

Documents support a decision. The gate decision authorizes movement.

Awaiting Portfolio attention

gate items plus operational feedback

Intake

ideas at G0

Analysis

needs, case, benefits

G2 prep

ready for Projects

in delivery or ops

6 live

Business analysis process groups

Use the process groups to know what work is expected before the next gate.

Defining & Aligning

Is this the right change?

Need, fit, score, tailoring

Initiating

How do we frame the effort?

Stakeholders, scope, assumptions

Planning

How will BA be governed?

BA plan, elicitation plan, method

Executing

What is required and why?

Requirements, options, benefits

Monitoring & Controlling

Is alignment still intact?

Traceability, change, gate checks

620

Releasing

Can this move onward?

Acceptance, transition, review

Gate path

Movement requires a recorded decision.

Intake screen

tailor

Justification

business case

Authorization

charter

Project baseline

plan

Execute, monitor & control

delivery

Acceptance

release

Closure

benefits

Decision queue

Items closest to gate or project movement.

6 items

PSFUApproved For ProjectG2

Project Sankofa — Follow-Up Initiative

Proceed with Option A full 4-WP implementation.

6ae84d40CharterG1

Tree Registration & Tenure Evidence Platform (TreeReg)

Institutional outreach is deprioritized per CEO's direct read of stakeholder readiness -- FC/TCDA/COCOBOD/World Bank will not engage before seeing a working tool. Shift focus to scoping and executing the Aframso pilot (capture workflow: unique tree ID, GPS/polygon, species, photo, owner, witness attestation) as the critical path to G2/charter. Revisit COCOBOD tree-module integration once the pilot produces a working proof point.

eb851306ConvertedG2

CLICOM - Ghana Climate Communication & History Series

Review and record the next gate decision.

VEH-ACCRA-2026ConvertedG2

Company Vehicle Acquisition – Accra Operations

Open delivery project and issue market-scan task.

ECO24ConvertedG2

ECO24 — 24hr Economy GAB Positioning & Revenue

RETIRED 2026-10-11 (CEO): "this project has been retired/closed and [was] part of the scattered agribusiness projects consolidated as the Farm Explorer Live". Consolidated 2026-07-21 into Farm Explorer Live (project 7c70e097): offtake work under Cluster 1, policy channels under Cluster 3. No live next action here. Remaining offtake work: Farm Explorer product action b5350264. Cluster 3 work (24hplus EOI and registration, tax incentive, including pm_task d6d23c80 rehoused here on 2026-09-29): parked in operational feedback fa07c385-dec3-46b8-af87-ff4f8f251562. Shared memory 8233995e.

PBA-EIRAS-GREENCARBONApproved For ProjectG3

Green Carbon / Forest Intelligence Lite

Prepare G3 baseline for forest/carbon source registry and report-pack workflow.

Product and service readiness

Public tools, paid services, platforms, and data backbones.

6 live

7064bb7aInternal Platform

Company coordination and initiative-to-product lifecycle

Data: Not setLaunch: Not set

d2df94c5Internal Platform

GAB OS — Operating Capability Review

Data: PartialLaunch: Not set

b629bb4fEnabling Backbone

GAB Supabase Estate Consolidation — Two Applications, One Database

Data: PartialLaunch: Planning

ff84349dEnabling Backbone

GAB Company Knowledge Layer — ACOI Completion

Data: PartialLaunch: Not set

044bc747Enabling Backbone

GAB Enterprise Architecture Consolidation — ACOI Follow-On

Data: PartialLaunch: Not Started

d6ed5f3eEnabling Backbone

GAB Identity, Naming & Design System Re-architecture

Data: PartialLaunch: Planning

1f8bf7f7Internal Platform

Product Operations Module — live product activation, handoffs and learning

Data: ReadyLaunch: Planning

ae5b00d1Enabling Backbone

Climate Reliability & Local Decision Intelligence

Data: PartialLaunch: Planning

Operations feedback loop

Operations → Portfolio & BA

Capture lessons, benefit gaps, risks, stakeholder signals, scale opportunities and retirement signals from live operations. Portfolio decides whether they become no action, an operational fix, a project change, a benefits review, or a new initiative.

7 feedback items awaiting decision

Capture operational signal

Use this when running work reveals something that should be considered again by BA.

Title

Source moduleType

Linked projectSummaryEvidence

Recommended actionPriority

SystemBenefit GapMediumTriaged

Final telemetry audit: knowledge closeout availability and desktop surface routing gaps

Fresh audit on 12 October confirms operating per-turn receipts across native hosts but governed knowledge closeout remains unavailable or permission-blocked in Codex CLI, Claude Code/Desktop Code/Chat, Kimi Claw, Gemini CLI and Antigravity app/IDE. Claude Code CLI follow-up received an extra Cowork instruction for the same native user UUID; its genuine claude_code receipt completed, and no false Cowork receipt was created. Desktop Chat UUIDs remain agent-declared with no native Stop enforcement. Kimi Work's first audit command approvals expired; a fresh native follow-up actually attested both current audit keys. Keep the knowledge-base initiative closed; do not backfill historical outcomes.

Action: Operational Fix • Process: Defining Aligning • Area: Business Analysis

Final audit finished with 24 fresh completed operating turns across 12 configured surfaces. Cowork Stop blocked both omission probes and obtained distinct receipts; its knowledge tool was unavailable. Kimi Work needed a real native follow-up after first command approval expiry. Two Claw and two in-app UI audit turns persisted under installed normal mode (remaining twenty controlled_qa); treat these four as mechanism tests, not business-benefit evidence, without inventing reviewer outcomes. Consumer Gemini/Perplexity unconfigured, Code Assist login/process blocked and Antigravity agent CLI unavailable. Watchdog recording: no pending delivery/current stale warnings; 104 local missing outcomes and two legacy warnings unknown. All 273 baseline terminal/legacy rows and full parked/stop knowledge initiative unchanged. Scope remaining fixes in operational feedback; do not reopen initiative.

2026-10-12 13:22 UTC — Repaired knowledge-closeout access and Claude Code/Cowork routing. Account plugin 1.0.2 positively gates Cowork entrypoints, requires current knowledge dry/write then operating receipt; fresh CLI and Desktop Code retained claude_code with zero Cowork misroutes. Independently verified 14 fresh native turns with immutable knowledge audit IDs and completed exact-key operating records (13 controlled QA, 1 restored-normal Claw other0/0). Scope: Codex CLI, Claude Code CLI, Desktop Code/Chat/Cowork, Kimi Work, linked/local Claw, Gemini API CLI, Antigravity app/IDE. See docs/operating-model/knowledge-closeout-access-qa-2026-10-12.json. Full parked/stop knowledge initiative and all 298 frozen terminal/legacy rows unchanged. Failed fresh probes remain unknown. Remaining limits: Chat declared identity/instruction-only, Cowork cached exact-key query fallback, Claw RPC final stream, unsupported consumer Gemini/Perplexity, separate Code Assist runtime, unavailable Antigravity agent CLI and OAuth. Watchdog recording, no pending deliveries/current stale warning; legacy unknown warnings retained. This remains operational feedback, not reopened Portfolio work.

Chief Of StaffBenefit GapHighClosed

In-app closeout counters ignore tool traces and knowledge receipts can be skipped

Fresh native turn e4e7a3bd in session f78f1f12 had two successful calls and a failed redundant closeout call but attested zero failures and skipped required knowledge dry-run/write. Server currently trusts tool counts and only gates the operating receipt.

Action: Operational Fix • Process: Defining Aligning • Area: Business Analysis

Fixed and independently verified 2026-10-12 at ae4d84d (production READY; GitHub 38226483630 success). Server derives tool success/failed/denied counters from executed tool records and forces per-turn knowledge dry-run/write with acknowledged current-key receipt before operating completion. Native session 083c5fbf-f6d2-4477-9dc0-80a5fd719214: turn77b82cbb completed with observed/stored4 success1 failure and knowledge813220a1; follow-up8baa7208 completed3/0 and knowledgef89c2354. Both actual knowledge audits checked against current keys. Controlled QA omission951bdd06 completed with mandatory2 knowledge calls. Knowledge-receipt failure21578753 returned503, withheld answer, zero knowledge receipt, zero operating outbox/completed row. Behavioral tests611 passed7 skipped0 failed, typecheck/lint passed. Synthetic QA archive probe was not requested by the model in two attempts; do not claim live unexecuted-write branch coverage (covered by unit test). 265 frozen historical rows unchanged; missing historical business outcomes stay unknown. Knowledge initiativeff84349d remains parked/stop; full initiative and requirements unchanged. Evidence output/inapp-reporting-verification-2026-10-12.json and native screenshot.

SystemBenefit GapMediumClosed

Watchdog falsely degrades on legacy Claude session-wide stale records

Live watchdog heartbeat and service are fresh with zero retry/outbox or identity failures. Two legacy auto:claude_code:<session> records alone trigger degraded status. Their missing outcomes remain unknown.

Action: Operational Fix • Process: Defining Aligning • Area: Business Analysis

Repaired and deployed 94f392d. Live server and independently scheduled collector expose total stale=2, current stale=0, legacy stale=2, retries=0, queue available; wakeup activation recording. Quality/Security and Deploy Migrations both succeeded in GitHub run 38222727862; production READY gab-ixjqn6ixp-bustabus48-7182s-projects.vercel.app. 606 tests passed, 7 skipped, focused lint/typecheck passed. Real-transport fixture tests verify exact-turn Stop gates, next-turn isolation, durable retry, no inferred success, and protected concurrent actual closeout. Two fresh native in-app turns have distinct completed receipts; first preserved a rejected redundant closeout call, second corrected an unsupported backlog narrative and issued empty knowledge receipt c8d054d6. None of 258 frozen historical rows or knowledge initiative requirements changed. Historical unknowns remain unknown; initiative remains parked/stop. Native current Codex scheduled retry proof is recorded separately at final task closeout.

Chief Of StaffImprovementMediumClosed

Successful Portfolio read mislabelled denied by telemetry content matching

Fresh in-app knowledge closeout succeeded, but query_pba_detail was labelled denied because nested requirement text contains an access-denied test criterion. Keep this separate from the wound-down knowledge initiative.

Action: Operational Fix • Process: Defining Aligning • Area: Quality

Repair authorized by Daniel on 2026-10-12. Classified BROKEN/REPAIR with passed Lane B pre-build gate. Commit ca26bd0 limits JSON classification to top-level execution fields and preserves explicit denial/error/legacy approval responses; regression tests reproduce the old false positives and pass after repair. Full suite: 604 passed, 7 skipped; typecheck and focused lint passed. Pushed main; production and native in-app verification are pending. No historical reclassification and no knowledge initiative reopening.
Verified end to end in production: native route 91475c56-d152-4b3f-8241-a9fa9d1f8e0f / assistant message 3d8237df-5993-42ca-9b92-4b4095282695 records successful Portfolio and memory reads despite nested denial language; the absent-source negative control remains denied, with immutable denial audit 18edcd2c. Knowledge closeout 7451e1d2 recorded no durable promotion; native operating receipt acknowledged completed. Production gab-kgcdwrim1 is READY for ca26bd0. GitHub run 38217114798 succeeded in both Quality and Security Gates and Deploy Supabase Migrations. No schema change. All 252 non-active historical session rows and the two earlier tool traces match their baselines; initiative parked/stop and requirements unchanged. Historical missing outcomes remain unknown.

SystemRiskMediumNew

Pre-build gate accepts any text as project-task evidence

npm run gab:preflight requires evidence.project_tasks but only checks that it is filled in. During initiative ff84349d it was given the initiative id and requirement codes ("initiative ff84349d BR-001") for more than twenty Lane B builds while the initiative sat at G1 with no project, so the gate never flagged that no project or task existed. Proposed: accept a real pm_task id (or a recorded CEO no-project approval on the initiative) and verify it live.

Action: Operational Fix • Process: Defining Aligning • Area: Integration

ProjectsImprovementLowNew

Farm Explorer Live project record is stale

Project 7c70e097 (Farm Explorer Live — Agribusiness Flagship) is closed but its description still routes execution through "Farm ERP" (retired 2026-10-06), and its three consolidation milestones (Cluster 1 offtake, Cluster 2 value-addition R&D, Cluster 3 institutional & policy channels) remain pending. Live work sits in Product Operations (farm_explorer_live). Cluster 3 work is parked in feedback fa07c385.

Action: Operational Fix • Process: Defining Aligning • Area: Integration

SystemRiskMediumNew

Scheduled agent runs cannot be confirmed through governed tools

Each cron agent writes a run summary to agents_log, but no governed tool reads it, and the commodity listener prints nothing on success, so a run cannot be confirmed (2026-10-11: the 06:45 UTC commodity run could not be confirmed). Proposed: read-only query_agent_runs over agents_log; listeners print their summary line. Pending check that depends on it: INT-2, the first recorded commodity-listener run should change no owned opportunity fields (no opportunities history rows were seen on 2026-10-11).

Action: Operational Fix • Process: Defining Aligning • Area: Quality

SystemImprovementMediumNew

Record protection remainder from the knowledge-layer initiative

Moved out of initiative ff84349d at its wind-down (CEO 2026-10-11). Already fixed and live: append-only contact and operational notes, opportunity listeners refreshing only source facts, task progress and first-completion evidence kept, archive reasons, before/after history on 104 tables, write-protection CI gate, history attribution for agent tool writes (migration 20261011062515). Remaining: (1) universal rejection of concurrent stale-version updates (only some tools check updated_at or expected text); (2) the write-site scan does not cover SQL functions or Python scripts, and behavioural tests do not cover every class; (3) attribution package 2: about 120 background-job and dashboard routes on the admin client still record database:service_role.

Action: Operational Fix • Process: Defining Aligning • Area: Quality

ProjectsNew NeedLowParked

Parked ECO24 Cluster 3: institutional and policy channels (24hr economy) — 4 items

CEO 2026-10-11: ECO24 "has been retired/closed and [was] part of the scattered agribusiness projects consolidated as the Farm Explorer Live"; Cluster 3 (Institutional & Policy Channels — ECO24 & School Farms) work to be parked. No live home exists: ECO24 and Farm Explorer Live (7c70e097) are both closed, and no Farm Explorer product action covers Cluster 3. Moved out of the closed ECO24 project (task statuses set to superseded) so they are not lost: - Register GAB as 24hr economy participant (pm_task 343532a8-ce15-4a7f-977d-8223d4af6dd6; was todo; due 2026-07-15; project: ECO24 — 24hr Economy GAB Positioning & Revenue) - Submit 24hplus.gov.gh EOI registration (pm_task 654f96c4-9928-41a6-beb3-0673ca0a3b49; was todo; due 2026-07-15; project: ECO24 — 24hr Economy GAB Positioning & Revenue) - Complete tax incentive eligibility assessment (pm_task b656127e-3f17-4d26-b735-5092065f55a3; was todo; due 2026-07-31; project: ECO24 — 24hr Economy GAB Positioning & Revenue) - Register GAB on 24hplus.gov.gh (pm_task d6d23c80; superseded 2026-09-29 into the ECO24 initiative c5ce5b55 next action, which is now retired; project: Value Addition Feasibility & Processing Readiness)

Action: Review In Portfolio • Process: Defining Aligning • Area: Scope

SalesNew NeedMediumParked

Parked commercial and finance decisions from closed projects (6 items)

CEO 2026-09-29: "park all of H for now". Six items needing a CEO decision, moved out of closed projects (task statuses set to superseded) so they are not lost: - Contact Premium Foods Ltd for contract farming discussion (pm_task 7ea928a7-2136-43f0-bb45-27717c93999b; was todo; due —; project: Value Addition Feasibility & Processing Readiness) - Draft standard contract-farming term sheet with legal review (pm_task c6b4d9bd-4500-4588-bc6c-f67e9ac619bc; was todo; due —; project: Value Addition Feasibility & Processing Readiness) - Evaluate 24H+ Value Chain Lending via GIRSAL and DBG pre-qualification (pm_task 7721ed1e-919f-452f-95cc-ad26f5bfd6e9; was todo; due —; project: Value Addition Feasibility & Processing Readiness) - Register GAB on AgroCenta and Complete Farmer (pm_task 257320bb-1888-484b-92b6-74926709a1a9; was todo; due —; project: Value Addition Feasibility & Processing Readiness) - Qualify & Scope MLNR ReportBridge Lead (pm_task 0b64c741-4f4b-464f-ba45-13b401902f5f; was todo; due —; project: Value Addition Feasibility & Processing Readiness) - Deliver first anchor pilot and document case study (pm_task 3c996bb3-8a98-4343-b0f1-37c1059b85a4; was todo; due 2026-08-10; project: AI Operations Integration Service Line) Context at parking: Premium Foods has no supply agreement though its buyer record is active; a standard term sheet would help before the Maame Krobo agreement ends on 2026-11-30; the MLNR ReportBridge lead (SSL-20260713-6894D3) is not in the consultancy pipeline; no signed AIOPS engagement exists for an anchor pilot.

Action: Review In Portfolio • Process: Defining Aligning • Area: Business Analysis

EirasImprovementMediumLinked To Initiative

EIRAS product backlog carried from closed product projects (17 items)

CEO 2026-09-29 confirmed: roadmap items left open in closed product projects become backlog for the EIRAS Product Family Commercialization initiative (919188e4), not tasks. Task statuses set to superseded: - Build paid screening report workflow (pm_task 6b85eb54-5d57-4eb5-b8d5-7654a7777a3d; was todo; due —; project: BioMainstream Lite) - Design biodiversity offset support package (pm_task 04a2814f-fda0-40bf-bdfe-f575bfdb93ca; was todo; due —; project: BioMainstream Lite) - Engage MESTI CBD focal point (pm_task c8144b29-e071-4c7f-a17c-99cc8df273b7; was todo; due —; project: BioMainstream Lite) - Expand source pack to 50+ verified sources (pm_task 953e9586-5c89-4be6-a176-ec8d28daf0c2; was in_progress; due —; project: BioMainstream Lite) - Integrate with ReportBridge MEA templates (pm_task 0620b692-4bc4-4ea5-81e0-274c2b08b5f8; was todo; due —; project: BioMainstream Lite) - Survey 5–10 EIA consultants on checklist utility (pm_task 43fbe1c8-f2ed-409b-9e75-13017191ad19; was todo; due —; project: BioMainstream Lite) - Build carbon stock estimation workflow (pm_task 23a9d814-c9d7-42dc-afa2-d796928be1dd; was todo; due —; project: Green Carbon / Forest Intelligence Lite) - Design restoration pathway template (pm_task 65385d74-69f0-4dc1-afd8-17ce3b57465f; was todo; due —; project: Green Carbon / Forest Intelligence Lite) - Develop Ghana-specific biomass equations and parameters (pm_task 9700bbcd-7dfa-4b9c-86a1-26c1a2efa647; was todo; due —; project: Green Carbon / Forest Intelligence Lite) - Engage FC REDD+ Secretariat for collaboration (pm_task ac59bd2f-9c75-441e-89dd-43d6198beb0a; was todo; due —; project: Green Carbon / Forest Intelligence Lite) - Build institutional partnerships (EPA, FC, Minerals Commission) (pm_task 070eca3a-4a93-4e7c-a0ee-28f047882d98; was todo; due —; project: EIRAS Integrated Environmental Reporting & Advisory System) - Design cross-product bundles (pm_task 394c4b60-81cb-4b02-9b23-21af31c9d8ed; was todo; due —; project: EIRAS Integrated Environmental Reporting & Advisory System) - Launch enterprise advisory retainers (pm_task 469b8fd4-3333-43d5-b5c0-49e27d446f10; was todo; due —; project: EIRAS Integrated Environmental Reporting & Advisory System) - Build cross-initiative dependency dashboard (pm_task e178c125-a863-420f-8d14-1618fbb1385b; was todo; due —; project: Efficiency & PMO / Sustainability Maturity) - Build sustainability maturity scoring framework (pm_task 6b8a8d1d-b2ef-4b39-982f-539a516397f5; was todo; due —; project: Efficiency & PMO / Sustainability Maturity) - Pilot PMO-as-a-service with 1 external client (pm_task bf5532f2-361f-4741-9b32-7ea26660d07a; was todo; due —; project: Efficiency & PMO / Sustainability Maturity) - Optimise mobile performance (<3s load) (pm_task 781262af-865c-4ec2-8af6-4f922dc0c379; was todo; due —; project: SiteSense Lite)

Action: Review In Portfolio • Process: Defining Aligning • Area: Business Analysis

Open linked Portfolio initiative →

CommsImprovementMediumParked

Comms backlog parked for the comms work (3 items from closed projects)

CEO 2026-09-29: park for the separate comms work Daniel will return to later. Moved out of closed projects (task statuses set to superseded): - Build reel/poster content templates for personal-contact tool sharing (pm_task 0b1a123d-efeb-4ec4-bc56-9b3db650d8b8; was in_progress; due —; project: AI Operations Integration Service Line) - Build creative benchmarking — study competitors/ads to raise comms output quality (pm_task fb00b2e1-458f-470e-b297-74a8ad142ef5; was todo; due —; project: Platform Development & Enhancements) - Set up LinkedIn for GAB before public launch (personal-led + company page) (pm_task 45cfd256-d34a-4b07-af2d-addbbac879f2; was todo; due —; project: Platform Development & Enhancements)

Action: Review In Portfolio • Process: Defining Aligning • Area: Business Analysis

SystemNew NeedHighConverted To Initiative

Company knowledge discovery — agents can retrieve known sources but cannot discover what GAB already knows

CEO expectation (Daniel, 2026-09-29): "I always think I have a company knowledge base for agents to tap into, so things already in the OS are available and accessible to agents." Triggering incident: an agent proposed creating a Meta system user the company already had (gabos, ID 61578649817263, documented in docs/comms/comms-ecosystem-reference.md and the pinned Comms memory). CEO directed this be assessed holistically against the recent design and architecture work, not built piecemeal. WHAT THE LINEAGE ALREADY DECIDED (constraints to respect): Baobab established the enterprise evidence base; Canopy the multi-owner dimensions; ACOI (project 1c9864e8) WP4 chose "a retrieval path, not a knowledge repository" — owners keep truth, nothing copied into a second store; the Agent Context Authority Manifest is navigation only; retrieve_governed_knowledge is exact-identity, audited, and explicitly "never a broad search". Architecture Consolidation (044bc747 -> 0f14a718) closed 2026-09-21 with the three ACOI capability questions undispositioned; Design System (ed16b3ff) closed 2026-09-15 and governs any human-facing surface. THE GAP: retrieval works, DISCOVERY does not. An agent can fetch a source it can already name, but cannot ask "what does GAB know about X?". Contributing: (1) recall_context exact-phrase defect — fixed 2026-09-29 (0c57419b), operational only; (2) reference knowledge in repo docs (docs/comms/*.md etc.) is outside the manifest and retrieval allowlist, reachable only by grep; (3) knowledge scattered across chief_of_staff_memory, contracts, reference docs, STATE.json and per-agent local memories with no single discovery path; (4) freshness: pinned summaries hand-kept and drifting; the ACOI monthly maintenance (6d111446) was not performed until 2026-09-29; 8 pinned memories still present Brand v2.2 (outgoing canon-v2). QUESTIONS FOR THE NEEDS ASSESSMENT: Can a discovery index that points to owners (snippet + owner + freshness + hash, rebuilt from sources on deploy, never authoritative) satisfy discovery without becoming the second store WP4 ruled out? Should reference docs join the manifest/allowlist? Word vs semantic search (embeddings) and cost? How pinned memory should be generated from sources rather than hand-kept? How this relates to Ghana Systems Phase 0 knowledge methodology and the three open ACOI capability questions (plugin 76280206, specialised model 98b88d17, GIS reasoning bd5d54b0)? Human surface under the Ground Sample design system, or agent-only? SEQUENCING: not yet decided by the CEO. Per the sequencing principle (4ca41064) this is deferred-by-sequence, not excluded; candidate placement is after or alongside the FloodSense campaign review.

Action: Review In Portfolio • Process: Defining Aligning • Area: Integration

Open linked Portfolio initiative →

ApprovalsImprovementHighClosed

Chat approval path for queued approvals (decide_agent_action) — operational fix

CEO 2026-09-29: HITL means Daniel explicitly decides, not that he must click. Daniel instructed approval of the Sept payroll_run (50b80576, GHS 30,096.27) in chat; it stayed pending because no chat path to a queued card existed. Pre-build gate: PARTIALLY_BUILT / FINISH_OR_WIRE. Repository: no approve/reject tool in gab-os, gab-mcp-server or gab-voice-relay; shared executor executeApprovedAction() exists (dashboard, executive link, WhatsApp); lib/actionPolicy.ts + ACTION_POLICY.md already treat daniel_chat as approver. Contracts: approvals.md said 'there is no approve agent tool' — contradicting ACTION_POLICY. Live records: 113 executed cards = dashboard 76 / executive_link 31 (all issued by ceo-notify email cron, last used 2026-06-27) / WhatsApp 5 / 1 null (2026-05-01 token-refresh retry); chat-created posts (1 & 12 Aug) were approved 2-4 min later under Daniel's dashboard login, i.e. agent clicks were recorded as his own. Project tasks: no task owns this; Platform Development (be2d35e8) is closed, so recorded here per 'projects must end'. Fix: decide_agent_action (preview→fingerprint→verbatim instruction, refuses autonomous callers, reviewed_by daniel:chat, audit-logged, shared executor), approvals.md + ACTION_POLICY.md aligned, Morning Briefing lists chat decisions from last 24h.

Action: Operational Fix • Process: Operations • Area: Integration

SystemImprovementMediumNew

A review that states no correction counts is arithmetically identical to one confirming zero — and the bias runs in the flattering direction

review_agent_operating_session requires session_key and useful_outcome, but leaves its four confirmed-count fields optional: confirmed_corrections, confirmed_deviations, confirmed_wrong_module, confirmed_wrong_source. The aggregate then reads an absent value as zero, so "reviewed, not stated" and "reviewed, confirmed none" are indistinguishable in every rate computed from them. The bias is not symmetric. An unstated count can only pull a rate DOWN, never up — toward fewer corrections, fewer deviations, fewer wrong-module and wrong-source incidents. That is the direction that makes a model look better than the evidence supports, in a measurement whose entire purpose is comparing models honestly. This contradicts the discipline applied everywhere else in the same layer. complete_agent_operating_session makes tasks_attempted and tasks_completed REQUIRED, precisely so task success cannot be inferred. reconcileTurn refuses to treat a host final message as proof of success. readClaudeLifecycle refuses to guess a model or attribute a session spanning two. The review tool is the one place the same class of number may be silently omitted and silently read as a fact.

Action: Change Request • Process: Defining Aligning • Area: Business Analysis

SystemRiskHighTriaged

The 1 October cross-agent measurement cannot reach decision-readiness — the mechanism is fixed, the sample is not recoverable in 10 days

The corrected cross-agent measurement collecting 1-30 September 2026 will not be decision-ready at its 1 October review. Two causes, one now fixed and one not. FIXED 2026-09-20 (commit 3d0d9d0b): automatic enrolment recognised only Codex, so Claude Code, Kimi and Gemini never enrolled. A cross-agent measurement was collecting automatically from one surface — the same shape of failure feedback 06137b87 recorded against ACOI. A Claude Code adapter now enrols with the exact model read from host metadata, and session close binds the telemetry closeout in both AGENT_OPERATING_PROTOCOL.md and 05-session-hygiene.md. NOT FIXABLE BY 30 SEPTEMBER: the sample. Nothing infers completion by design — the watchdog records 'interrupted' with zero counts when an agent ends without closing out, because a final message does not prove the work succeeded. That is correct and must not be relaxed, and it is why 33 of 36 sessions are interrupted.

Action: Benefits Review • Process: Defining Aligning • Area: Business Analysis

CEO DECISION 2026-09-20: extend the collection window to October.

Collection 1-30 September -> 1-31 October 2026. Review 1 October -> 1 November (benefits review 7ef2cd95 on project 1c9864e8, status set to deferred with the full reasoning recorded there). The three ACOI capability questions that depend on this evidence moved 2026-10-15 -> 2026-11-15: 76280206 plugin packaging, bd5d54b0 GIS-specialist capability, 98b88d17 company-specialised model.

The mechanism half is done and deployed (commit 3d0d9d0b): Claude Code now enrols automatically with the exact model read from host metadata, and the telemetry closeout is bound into session close in AGENT_OPERATING_PROTOCOL.md and 05-session-hygiene.md. Kimi and Gemini still have no adapter — if either carries material work in October, the measurement will under-count it the same way it under-counted Claude.

THE OCTOBER WINDOW IS NOT SELF-EXECUTING. What failed in September was not only the missing adapter; it was that nobody looked at the numbers until day twenty. Two things need watching during October, not at the end of it: whether completed sessions accrue across more than one surface, and whether review coverage moves off zero. Review coverage is CEO time and was the threshold furthest from being met — 0 of a required 24.

THIS IS THE SECOND DEFERRAL. If October also fails to reach decision-readiness, the honest response is to decide the three capability questions without the measurement and record that it was never obtained, rather than defer a third time.

ProjectsImprovementMediumClosed

query_pm_tasks ignores its task_ids and search arguments — targeted lookup is silently impossible

query_pm_tasks accepts task_ids and search but honours neither. Both calls return the identical default page of oldest tasks across unrelated projects. It fails SILENTLY — no error, no empty result, just plausible-looking wrong rows. An agent that trusts it will reason about tasks it did not ask for, and a targeted check ("is task X still open?") will return whatever happens to be oldest.

Action: Change Request • Process: Defining Aligning • Area: Business Analysis

RESOLVED 2026-09-21, commit 9cf17d8b, deployed and verified against production.

THIS RECORD'S ORIGINAL DIAGNOSIS WAS WRONG, AND THE CORRECTION IS THE POINT. It claimed query_pm_tasks "ignores its task_ids and search arguments". The tool never HAD those parameters. I called it with arguments I had invented; validateInput passed unknown fields through as "handler-tolerant"; the handler ignored them; and I received the unfiltered default page of oldest tasks from unrelated projects as normal JSON with no warning — and concluded the tool was broken. The tool did exactly what it advertised.

THE REAL DEFECT IS THE SILENCE, and it is worse than the one reported, because it applied to all 231 tools rather than one. A wrong answer delivered confidently is worse than an error — the same reasoning that retired the eiras contract alias the previous day.

THREE PARTS, in the order that made each safe:
1. query_pm_tasks gained task_ids and search. The underlying need was real: fetching known tasks by id was impossible, so "is task X still open?" could not be answered without already knowing its project. Search terms are stripped of PostgREST filter punctuation rather than escaped, since a substring search over titles has no honest use for it.
2. EIGHTEEN TOOLS were found reading parameters their schemas never declared — query_projects/division, query_prices/crop, query_alerts/include_snoozed, set_division_budget/dry_run, create_alert's three source_* fields and more. Capability that existed and no caller could discover. All declared. This had to come first: enabling strict rejection while they were still hidden would have refused calls that worked.
3. Unknown parameters are now rejected in validateInput, the single choke point, with the accepted parameter list in the error so a wrong guess corrects itself instead of guessing again.

A METHOD NOTE THAT ALMOST COST ME. My first scan reported 21 tools with undeclared parameters. The regex missed properties sharing a line, so complete_agent_operating_session appeared to be missing tasks_completed, which it plainly declares. Verified one case by hand before reporting; the true figure is 18. A probe that produces a plausible number is not a verified number.

VERIFIED AGAINST PRODUCTION, not tests alone: task_ids returned exactly the two requested tasks; search 'retirement discipline' returned the two matching rows; an unknown parameter was refused naming all six accepted ones.

OPERATIONAL FACT WORTH KNOWING: the local MCP stdio bridge caches the tool catalogue at session start. The same calls through an already-open session still behaved as before while the deployed endpoint was already correct. New tools and new parameters need a fresh agent session, not just a green deploy.

A regression test re-runs the handler-reads-versus-schema-declares comparison on every build, so the hidden-parameter gap cannot reopen. 374 tests pass; MCP parity 233 handlers / 231 definitions / 231 external tools.

WebsiteImprovementMediumClosed

36 React Compiler lint findings carried as warnings after the Next 16 upgrade — setState in effects across 39 files

Action: Review In Portfolio • Process: Defining Aligning • Area: Business Analysis

EirasImprovementMediumClosed

react-leaflet throws "Map container is already initialized" in DEV under React 19 — map pages unusable with npm run dev

Action: Review In Portfolio • Process: Defining Aligning • Area: Business Analysis

WebsiteImprovementMediumClosed

Next.js 14 → 16 migration required: the only fix for two critical advisories, and CI carries a dated exception until it happens

Action: Review In Portfolio • Process: Defining Aligning • Area: Business Analysis

EirasImprovementMediumClosed

Efficiency Lite renders GHG tonnage with a bare toLocaleString() — the figure changes meaning with the reader's browser locale

Action: Review In Portfolio • Process: Defining Aligning • Area: Business Analysis

EirasImprovementMediumClosed

FloodSense Lite result contradicts itself on wetland distance, and formats population as a decimal

Action: Review In Portfolio • Process: Defining Aligning • Area: Business Analysis

EirasImprovementMediumClosed

No governed way to mark an eiras_sites row as a test — is_test is only ever set by the paid-review lead notifier

Action: Review In Portfolio • Process: Defining Aligning • Area: Business Analysis

EirasImprovementMediumClosed

SiteSense Lite: district is required by the API but not by the form — users get a bare "Validation failed"

Action: Review In Portfolio • Process: Defining Aligning • Area: Business Analysis

SystemRiskCriticalClosed

Next.js critical security advisories block the migration pipeline — major upgrade required

GAB OS runs a Next.js version carrying 23 published security advisories, one rated CRITICAL, including UNAUTHENTICATED REMOTE CODE EXECUTION on Windows-hosted servers and REMOTE CODE EXECUTION in the Image Optimization API when AVIF files are used. Others include server-side request forgery in rewrites and in Server Actions, HTTP request smuggling in rewrites, cache poisoning of React Server Component responses, middleware/proxy bypass, cross-site scripting in App Router applications using CSP nonces, and unauthenticated disclosure of internal Server Function endpoints. The affected range is next 9.3.4-canary.0 through 16.3.0-preview.10. The only remediation npm offers is next@16.3.5, flagged isSemVerMajor true — a MAJOR VERSION UPGRADE of the framework the entire OS and every public tool runs on. SECOND, SEPARATE CONSEQUENCE: this has silently broken the migration deploy pipeline. The GitHub Actions workflow "Deploy GAB OS" gates job deploy-supabase on needs: quality, and the quality job runs `npm audit --omit=dev --audit-level=critical`. That step now fails on this advisory, so quality fails, so NO MIGRATION CAN DEPLOY THROUGH CI. Verified 2026-09-12. NOT CAUSED BY ANY COMMIT. Deploy GAB OS last succeeded 2026-09-06 (run 34013123643). The two pushes since, both on 2026-09-12, failed. Nothing was pushed between those dates. `npm audit` evaluates against a live advisory feed, so the pipeline went red on its own when the advisories published. A CI gate that can fail with no code change, silently, is itself worth addressing. WHY IT WENT UNNOTICED: Vercel deploys on push through its own GitHub integration, independently of the Actions workflow. Vercel production stayed green throughout while CI was red, so `npm run gab:verify-deploy` reported success for commits whose Actions run had failed. VERCEL GREEN AND CI GREEN ARE DIFFERENT THINGS and the wakeup deploy check only observes the former.

Action: New Initiative • Process: Defining Aligning • Area: Risk

New initiative intake

Capture the idea first. Tailoring decides whether it stays light, becomes standard, or needs full governance.

7064bb7aConvertedStandardG3Internal Platform

Company coordination and initiative-to-product lifecycle

Implement the existing coordination proposal docs/proposals/2026-10-19_briefing-decisions-to-work.md at 7b1c35c4, including company-wide reconciliation/routing/verified follow-through and the integrated Portfolio & BA → Projects → Products → operational owners → feedback lifecycle. Daniel authorised implementation on 20 October 2026: 'Implement the coordination proposal. i will be here to provide any clarification and guidance when needed'. Existing initiative inventory and operational feedback were checked; the Operating Capability Review project is scope-limited and the Knowledge Layer initiative is wound down, so neither supplies an implementation owner for this proposal. Reuse existing module records and workflows; no duplicate inbox or business ledger.

Owner: Daniel Nsowah — Chief of Staff coordinationCategory: Company operationsPriority score: Not scoredLatest gate: G0 go Drive doc

Open 6-domain workspace →Open project

Analysis record

StatusStage gateGovernanceDecision

PriorityTailoringFitValueRisk

Needs assessmentBusiness caseBenefits planCharter basisRecommendationNext action

Productization and commercial readiness

TypeData readinessLaunch readiness

Public URL

Target usersPaying customers

Free vs paid boundaryPricing hypothesisService delivery model

Success metricsGo / no-go criteria

Productization plan

d2df94c5ConvertedNot setG3Internal PlatformData: Partial

GAB OS — Operating Capability Review

Daniel approved an OS-wide assessment on 6 October 2026 following discussion of DevDay, Your dot and other AI capability improvements. Review every operating module and cross-module workflow against efficiency, responsiveness, security and management relevance. Consultancy was only an example. Preserve demonstrated improvements and existing governance. Initial assessment followed by explicit authorization for bounded price safeguards, financial-label repairs and existing-owner Briefing connections; these are now deployed and verified. No model migration, new subscriptions or delivery project authorized. Existing Company Knowledge Layer initiative ff84349d-f0ed-4a13-84b3-f10a83b44036 remains the owner of shared knowledge work; consume October evidence and the 1 November benefits review rather than duplicate them.

Owner: Daniel NsowahCategory: Company operations / AIOPSPriority score: Not scoredLatest gate: G2 go Drive doc

Open 6-domain workspace →Open project

Analysis record

StatusStage gateGovernanceDecision

PriorityTailoringFitValueRisk

Needs assessment

# OS-wide operating capability assessment — 6 October 2026

## Decision context and evidence limits

Daniel says GAB OS has improved significantly. The objective is to exploit stronger AI capabilities where they improve company management, including efficiency, responsiveness, security and relevance. This is an operating-capability review across the company, not a consultancy-only upgrade or a defect-only audit.

This assessment covers the 31 modules returned by the live module catalogue plus Overview, Masterclass, CSR Partners, Investors and Notifications. It combines contract/workflow review, repository inspection of agent schedules and selected implementation paths, current Data Trust and alert records, existing initiative decisions, and official provider documentation. It is an initial architecture and opportunity assessment, NOT a completed production acceptance test of all 36 modules/surfaces. A module's existence does not prove reliable operation; a proposed opportunity does not mean the capability is wholly absent. No paid inference experiment, live external communication, deployment or model migration was performed.

Evidence levels below: L = live governed record queried; R = current repository implementation inspected; C = documented contract behavior reviewed; H = improvement hypothesis requiring a like-for-like test. Repository source is not a fresh execution trace of a deployed job.

## Findings with direct implementation evidence

1. **The timing of the management briefing deserves redesign (R).** vercel.json schedules daily-digest at 05:30 UTC, market-intel at 06:00, opportunity-listener at 06:30 and commodity-listener at 06:45. The daily briefing therefore precedes those scheduled results. Moving a clock alone would not handle failed/late inputs; compare an initial briefing with an update after relevant inputs finish, carrying source timestamps and explicit missing-input state. It may remain useful to brief early; do not silently change Daniel's routine.

2. **The briefing prompt encourages a fixed volume of work (R).** app/api/agents/daily-digest/route.ts requests 3–7 actions every day and directs every overdue task, critical alert and stale buyer follow-up to produce an action item. The existing approval executor already has deduplication and owner routing, so retain and test those. Compare with a change-aware briefing that may recommend no new action, explains what changed, respects recorded CEO deferrals, and links to existing work. Existing exclusion of repetitive GIS findings was a CEO decision; do not reintroduce them merely because a stronger model can summarise them.

3. **Automatic finance narration is narrower than the Finance module (R).** finance-weekly reads seven days of transactions, computes totals deterministically, and passes at most 20 transaction descriptions to a model for a 3–5 sentence narrative. This does not establish that Finance lacks forecasts, receivables or compliance tools. It shows that this scheduled workflow is retrospective. Compare adding evidence-backed upcoming obligations, receivables and cash scenarios through existing owner queries, with recorded cash clearly separated from forecasts. Missing transactions must remain unknown, not “healthy finances”.

4. **Changing the agent in this chat does not upgrade OS jobs (R).** Multiple routes explicitly configure gemini-2.5-flash; the orchestrator and task executor contain claude-sonnet-4-6 defaults. lib/research/grounded.ts also has a model default. This establishes configuration, not that the models are unavailable or inadequate. Evaluate by task class before selecting a replacement. Separate scheduling, prompts, retrieval, deterministic computation and model choice.

5. **Masterclass is a concrete candidate for contextual learning (R).** app/dashboard/briefing/course/page.tsx loads docs/gab_os_masterclass_course.md and parses modules/quizzes; CoursePlayer stores completed modules in browser localStorage. Retain the course and sandboxes. A candidate extension is short explanations and exercises linked to the decision Daniel is currently making, with authoritative source links and explicit learning progress. Synced progress and an adaptive tutor require verification/build; they are not already demonstrated.

6. **Aggregate trust needs a decision-specific view (L).** query_data_trust at 18:11 UTC returned 88% trusted overall, while Scientific/GIS was 28/125 trusted and the file-mirror family was 891/891. Counts measure different kinds of things; the headline is not a probability that a particular recommendation is correct. Show the trust of evidence used for the current decision. Do not infer that all scientific results are wrong or treat archived source items as missing without examining their lifecycle.

7. **Creative automation already has a recovery loop (C/R/L).** Weekly drafting, Studio Floor and bounded Studio Repair exist; the live alert register includes a graphic blocked after two repairs for a headline/brief mismatch. One alert cannot diagnose the entire creative system. A useful test is alternative rendering and semantic/visual review on the same brief, with the current approval boundary preserved. Avoid proposing “add automatic QA” as though none exists.

## Module-by-module disposition

These are recommended directions, not implementation approvals. KEEP means retain the operating owner and proven machinery; IMPROVE means test a bounded extension; SIMPLIFY means reduce duplicated/noisy work. No whole-module retirement or replacement is justified by this review.

| Module / surface | Current operating basis | Recommended disposition and candidate benefit | First verification |
|---|---|---|---|
| Overview | C/R: aggregates live owner records, sales dashboard and last successful digest | IMPROVE: show material changes, evidence freshness and links to the management decision | Reconcile displayed totals to owners; compare normal, missing-input and degraded states |
| Briefing | C/R: scheduled synthesis, approval and internal routing | SIMPLIFY + IMPROVE: change-aware priorities, fewer repeated actions, input-aware updates | Same-day replay with zero changes, a new risk, and an already deferred item |
| Chief of Staff | C/R: tool-based orchestrator, memory, context routing | IMPROVE: compare tool discovery and sustained task execution against current routing | Same cross-module requests; check sources, handoffs, latency and instruction adherence |
| Comms | C/R: campaigns, drafting, approved publishing and metrics | IMPROVE: connect campaign objective, audience evidence, creative choices and measured response | Follow one existing campaign from brief through results; retain retired-theme decisions |
| Production | C/R/L: assets, specialist tools, QA, floor and bounded repair | IMPROVE: better craft and recovery on difficult formats | Same brief through current and candidate workflow; human review of quality and effort |
| Brand | C: versioned identity and audit | KEEP: ground model outputs in the existing brand owner | Check generated assets against the selected brand version; do not rewrite tokens with AI |
| Website | C: routing/deployment ownership and inquiry workflows | KEEP + IMPROVE: faster evidence-backed content/inquiry handoffs | Trace one inquiry through response; compare content brief-to-approved page workflow |
| Market Intel | C/R: grounded research, price feed, valuation guards | IMPROVE: deeper investigation of material price changes and contradictory sources | Same commodity/date/geography; compare evidence, coverage, usefulness and runtime |
| Research | C/R: dossiers, freshness and counterparty verification | IMPROVE: multi-step research using existing dossiers and source controls | Refresh one dossier without duplicates or loss of history |
| Consultancy | C/R: verified opportunity intake; separate proposal/delivery/payment funnel | IMPROVE: requirements-level qualification and evidence-backed pursue/partner/decline advice | One actual notice traced to company capabilities and missing eligibility evidence |
| Sales | C: buyer records, agreements, inventory, follow-ups and receivables | IMPROVE: prioritise viable buyer actions and prepare evidence-backed next steps | Reuse existing suppression decisions; distinguish recorded supply from forecasts |
| Voice Agent | C: shared contact identity, messaging, private voice intake and governed replies | IMPROVE selectively: continuity and response quality | Same scripted enquiry across channels; no new calling capability assumed |
| Farm | C: OS-owned seasonal management and GIS context | KEEP + IMPROVE: management interpretation with minimal input burden | Test an actual seasonal decision without requiring daily logs or ERP data |
| GIS | C/R: spatial readings, mapped geometry, analysis and owned findings | IMPROVE: explain spatial changes with uncertainty and provenance | Time series plus cloud/freshness limits; ground-truth selected anomalies |
| Land | C: tenure/evidence/valuation history | KEEP + IMPROVE: document extraction and decision evidence packs | Extract a sample document with human verification; no automatic legal conclusions |
| Assets | C: asset register, photos, condition and valuation history | KEEP + IMPROVE: easier evidence capture and maintenance planning | Link source image/document to the correct asset; preserve valuation authority |
| Finance / compliance | C/R: ledger, receipts, invoices, forecasts and compliance; narrow weekly digest | IMPROVE: decision-aware finance brief and document-assisted preparation | Reconcile to ledger; unknown inputs stay unknown; no automatic ledger posting from a guess |
| Staff & HR | C: relationships, people costs and events | KEEP; selective document/admin assistance | Verify classification and Finance linkage; do not infer performance from sparse records |
| Portfolio & BA | C/L: intake, analysis, decisions and governance | KEEP + IMPROVE: better option analysis and source-linked decision packs | Compare alternatives with explicit assumptions; AI cannot manufacture a gate decision |
| Projects | C: tasks, milestones, controls, output packs and closure | IMPROVE: dependency-aware follow-through and evidence collection | Trace a real task to accepted output and operating handover |
| Tasks | C/R: orphan inbox and execution receipts | SIMPLIFY + IMPROVE: owner-aware intake and better bounded execution | Check duplicate prevention, correct ownership and honest completion evidence |
| Product Operations | C: ongoing product actions and owner links | IMPROVE: connect usage/feedback signals to specific owner actions | Trace one product signal through owner action to result |
| EIRAS | C: shared product family, screening and review | KEEP + IMPROVE: better source-supported interpretation and reviewer preparation | Fixed benchmark sites and current authoritative sources; no stronger unsupported claims |
| SiteSense | C/R: intake, screening worker and lead conversion | IMPROVE: clearer reasons, missing-evidence questions and conversion handoff | Replay approved sample sites; preserve preliminary-screening boundary |
| Reports | C/R: registered review packs, PDF/DOCX and Drive mirror | IMPROVE: reusable source-linked outputs and change-aware refresh | Regenerate a report after one source changes; verify provenance and numerical fidelity |
| Data Trust | C/L: source freshness and claim-readiness families | IMPROVE: decision-specific trust and targeted refresh | Trace each recommendation to its relevant source family, date and validation state |
| Audit Trail | C: append-only change evidence; completeness requires checking | KEEP + IMPROVE: before/after coverage and explainable action chronology | Sample actual mutations; absence of a trace remains unverified, not a pass |
| Approvals | C: explicit Daniel decisions through governed channels | SIMPLIFY presentation: concise action, evidence, consequence and exact scope | Verify displayed proposal matches executed payload and retry cannot duplicate action |
| Alerts | C/L: source-linked exceptions, suppression and reconciliation | SIMPLIFY: relevance, change detection, ownership and resolution evidence | Replay resolved/deferred/repeated alerts against source state |
| System | C/R: job/readiness/repository/deployment checks | IMPROVE: demonstrate useful outputs, not only successful HTTP/run status | Sample recent job output, freshness and downstream consumption; recovery exercise separately |
| Settings | C: controlled configuration | KEEP; make selected model/capability/version and cost limits reviewable where useful | Check configuration changes have provenance and controlled rollback |
| Personal Money | C: owner-only records, planning and evidence; isolated from company agents | KEEP isolation; evaluate owner-scoped assistance separately | Verify owner access and approved evidence handling before any new AI path |
| Masterclass | R: Markdown course, quizzes, local progress and sandboxes | IMPROVE: contextual explanations and practice tied to real OS decisions | Daniel can explain a decision and complete a relevant exercise; retain source-linked curriculum |

Contract evidence: docs/modules/<module>.md (hyphenated names), docs/modules/MCP_TOOL_MAP.md. Additional implementation evidence: vercel.json; app/dashboard/page.tsx; app/api/agents/{daily-digest,finance-weekly,weekly-comms-drafter,orchestrator,task-executor}/route.ts; lib/research/grounded.ts; lib/opportunities/listener.ts; app/dashboard/briefing/course/{page,CoursePlayer}.tsx.

## AI capability options and boundaries

Official documentation reviewed 6 October 2026. These options are not all newly released at DevDay; they are currently documented capabilities relevant to this review. No provider account eligibility or paid availability has been verified for GAB.

- **Ongoing responsibility:** Dots can work between conversations across connected apps; rollout is gradual. Candidate: a bounded follow-through role using GAB owner tools and returning results to existing records. Dot memory must not become a competing company ledger. Source: https://learn.chatgpt.com/docs/dots
- **Event-triggered work:** MCP Events supports matching server events in eligible Work/dot surfaces; requires MCP 2.0 and a webhook/subscription implementation. Candidate: respond to an inquiry, material market update or completed report instead of waiting for a fixed scan. GAB compatibility is untested. Source: https://developers.openai.com/plugins/build/mcp-events
- **Tool discovery:** OpenAI and Anthropic support loading relevant tools on demand. Candidate: compare against ACOI's existing progressive routing, not build a second tool catalogue. Sources: https://developers.openai.com/api/docs/guides/tools-tool-search and https://platform.claude.com/docs/en/agents-and-tools/tool-use/tool-search-tool
- **Background work:** Responses background mode supports longer asynchronous responses; it is not by itself a durable company workflow with ownership, retries or approvals. Candidate: research/report preparation with a visible job record and bounded completion callback. Source: https://developers.openai.com/api/docs/guides/background
- **Research:** Gemini Deep Research plans and synthesises multi-step research, supports document inputs/MCP and requires the Interactions API in background mode; currently preview. Candidate: use on consequential questions after a low-cost scan, rather than every price refresh. Source: https://ai.google.dev/gemini-api/docs/deep-research?hl=en
- **Security tooling:** DevDay documents Security Cloud in research preview for repository scans and reviewable fixes. This could complement current checks; no installation or repository authorization has been performed. Source: https://learn.chatgpt.com/docs/whats-new/devday-2026

Technical provider research used official documentation because GAB's exposed business-research tools specialise in counterparties and Ghana-market research; they do not provide a provider API documentation lookup. Claims here describe documented capability, not verified deployment suitability. Confidence high in cited interface descriptions, provisional in GAB benefit.

## Recommended comparison sequence

1. **Management view:** Overview + Briefing + Finance. Highest direct exposure to Daniel; compare existing outputs with change-aware, source-linked drafts on the same captured records. Start by testing scheduling/input freshness and recommendation relevance, not simply a model swap.
2. **Signal-to-action:** Market Intel + Research + Sales/Consultancy. Use cheap discovery and selective deeper investigation; compare evidence quality and accepted next actions, not raw lead volume.
3. **Creative delivery:** Comms + Production + Brand + Website. Compare one complete campaign workflow, including actual approved output and performance interpretation.
4. **Follow-through and learning:** Projects/Tasks/Product Operations, Chief of Staff and Masterclass. Test bounded continuing work and contextual learning; later consider events/dot where access and controls are verified.
5. **Specialist and sensitive paths:** Farm/GIS/EIRAS, finance document intake, people/land/assets and Personal Money need domain-specific validation and preserved boundaries. Supporting security, trust and audit checks apply across every comparison.

These are proposed priorities, not delivery dates or approved purchases. Low-risk operating improvements can use existing module paths; substantial changes follow their existing governance. No estimated cash savings, revenue uplift or performance percentage is invented.

## Dependencies and what remains open

- Company Knowledge Layer ff84349d-f0ed-4a13-84b3-f10a83b44036 owns discovery, currency, truth and history. Reuse it.
- Existing benefits review 7ef2cd95-c798-43b0-9c76-e9d2f25d40ab collects October evidence and reviews on 1 November. Do not move its dates or inflate past completion/review counts.
- Shared plugin/model/GIS capability decisions already sequenced for 15 November must not be silently decided by this review.
- Farm ERP stays retired; OS Farm/GIS own management data. Website remains a public app backed by OS-owned records. Personal Money retains isolation.
- An open backup alert still describes the retired ERP. This is evidence of record drift, not authorization to restore ERP, reactivate an obsolete backup workflow, or claim current recovery capability. Follow the earlier CEO-directed backup-purpose review.
- Full per-module live workflow walkthroughs, account access, runtime/cost baselines, model compatibility, input/privacy policies, vendor retention and negative-path controls remain unverified. Representative tests should resolve these before implementation approval.
- G0 intake/tailoring and any later gate decision remain distinct from Daniel's authorization to assess. This document is a draft for review, not a charter or a blanket implementation mandate.

## Additional navigation surfaces verified during filing

The authenticated sidebar exposes several workflows outside the 31-module catalogue; these are included rather than silently omitted. They do not automatically require new module ownership.

| Surface | Current implementation | Recommendation / first test |
|---|---|---|
| CSR Partners | R: app/dashboard/sales/csr-partners/page.tsx reads csr_partners and CSR opportunity candidates, with qualification and pipeline states | IMPROVE within existing commercial ownership: partner-fit analysis grounded in actual programmes and GAB capabilities; trace one candidate through eligibility, outreach preparation and partnership result |
| Investors | R: app/dashboard/investors/page.tsx aggregates land, assets, blocks, projects, invoices, transactions and harvest records | KEEP + IMPROVE: source-linked investor narrative and scenarios, separating recorded cash, invoiced amounts, valuations and projections; test missing data and approval of outward claims |
| Notifications | R: app/dashboard/notifications/page.tsx aggregates alerts, approvals, agent runs, audit and task executions with per-user read state | SIMPLIFY: meaningful-change grouping and consistent severity/ownership across Briefing, Alerts and Notifications; test that suppressing duplicate presentation never hides a pending approval |

The separate Consultancy Pipeline page is a view of the Consultancy workflow already assessed; Chat is the Chief of Staff surface. No automatic duplication of modules is proposed.

Business caseBenefits planCharter basisRecommendationNext action

Productization and commercial readiness

TypeData readinessLaunch readiness

Public URL

Target usersPaying customers

Free vs paid boundaryPricing hypothesisService delivery model

Success metricsGo / no-go criteria

Productization plan

b629bb4fConvertedStandardG3Enabling BackboneLaunch: PlanningData: Partial

GAB Supabase Estate Consolidation — Two Applications, One Database

CEO direction 2026-10-05: retain GAB website and GAB OS as separate applications/deployments, migrate website-backed records and functions to the GAB OS Supabase project, make Farm Agent independent of the unused Farm ERP database without treating demo entries as real farm history, and archive/retire Ramsar, Blue Carbon, Personal Banking, the unnamed original and source Supabase projects after verified backups, cutovers and explicit exact-target retirement approval. Execution plan: docs/projects/2026-10-05_supabase-estate-consolidation-plan.md. The completed Enterprise Architecture Consolidation project is an input, not reopened.

Owner: Daniel NsowahCategory: AIOPS / System / Infrastructure CostPriority score: Not scoredLatest gate: G1 go Drive doc

Open 6-domain workspace →Open project

Analysis record

StatusStage gateGovernanceDecision

PriorityTailoringFitValueRisk

Needs assessmentBusiness caseBenefits planCharter basisRecommendationNext action

Productization and commercial readiness

TypeData readinessLaunch readiness

Public URL

Target usersPaying customers

Free vs paid boundaryPricing hypothesisService delivery model

Success metricsGo / no-go criteria

Productization plan

ff84349dParkedFullG1Enabling BackboneData: Partial

GAB Company Knowledge Layer — ACOI Completion

SOURCE SIGNAL (CEO direct, 2026-09-29). Daniel Nsowah: "I always think I have a company knowledge base for agents to tap into, so things already in the OS are available and accessible to agents." The future being discussed is an operating system with a company knowledge base and a good harness — what GAB OS was meant to be. CEO directed a holistic assessment against the recent design and architecture work, not a piecemeal build. Promoted from operational feedback 9ae49538 on explicit CEO instruction: "promote it, four properties, agent-only for now". LINEAGE. Baobab established the enterprise evidence base; Canopy the multi-owner dimensions and Atlas; ACOI (1c9864e8) the context manifest and governed exact-identity retrieval, deciding "a retrieval path, not a knowledge repository"; Architecture Consolidation (0f14a718) one-declaration-per-bound-value and retirement discipline; Design System (ed16b3ff) brand as a governed module with one versioned source and a drift gate — the pattern to reuse. Closed projects are read as inputs, never reopened. THE GAP. GAB OS has a strong harness; its knowledge is plentiful but organised for governance, not for use at the point of action. Retrieval works when an agent can name the source. Discovery, currency, verification and integrity were never designed. DRAFT: docs/projects/2026-09-29_company-knowledge-layer-G0-draft.md (9fd00e99).

Owner: Daniel NsowahCategory: AIOPS / System / Agent OperationsPriority score: Not scoredLatest gate: G0 go Drive doc

Open 6-domain workspace →

Analysis record

StatusStage gateGovernanceDecision

PriorityTailoringFitValueRisk

Needs assessment

# Company Knowledge Layer — dashboard, scheduled jobs and seed history assessment

Date: 9 October 2026. Initiative: `ff84349d-f0ed-4a13-84b3-f10a83b44036`.
Elicitation: `1dcf52f9-d14f-47d1-b356-e93abfbb6ab8`.
Author: Codex. Inspected source: `faada22983f278cb8b2720034b6262c904221ce4`.
Status: completed bounded G1 assessment and proposed implementation scope; not gate or build approval.
Canonical decisions, requirements and assessment remain in Portfolio & BA. This file is dated repository evidence.

## Conclusion

Protecting orchestrator tools alone will not protect company history. Dashboard actions, scheduled jobs and seeds reach the same records through separate writers. Some preserve useful event history; others only stamp the latest actor, log an event without the old content, or overwrite/delete directly. The correct implementation boundary is the business record and all its writers, not the agent interface.

Two isolated behaviour checks reproduced actual payload loss: the dashboard clears task progress on a blocked transition, and its assessment editor removes sibling JSON keys. No production mutations were used to test these cases. No claim is made that a particular live record has already lost data.

## Method and coverage

- Enumerated 129 TypeScript/TSX files under `app/dashboard`, all 27 scheduled routes declared in `vercel.json`, and JavaScript/TypeScript scripts.
- Followed static local imports from dashboard and scheduled roots. Parsed call expressions with the installed TypeScript parser, recording chained `from(...).update/delete/upsert` and RPC calls. The inventory is [the dated JSON evidence](2026-10-09_company-knowledge-layer-write-path-inventory.json).
- Scanned 500 unique files, yielding 330 mutation/RPC candidate sites: 159 directly in dashboard files, 16 directly in scheduled route files, and the remainder in imported helpers, agent handlers and scripts. Nineteen candidates are RPC calls and include reads. Origin tags overlap.
- Import closure is an over-approximation, not proof that every function in an imported file runs from a cron. For example, the screening worker imports an EIRAS action file containing many unrelated actions. Do not report the 153 scheduled-origin candidates as 153 live cron mutations.
- Manually traced the concrete findings below, including adjacent helpers and repository triggers. Also inspected raw REST memory seeding, which the chained-call scan does not detect.
- This is not an exhaustive call-graph, access-control or live database trigger audit. Dynamic imports, raw SQL/REST, external workers and jobs outside `vercel.json` require explicit coverage during implementation. No destructive production tests, live schema changes or cron invocations were performed.

## Findings and disposition

| Finding | Evidence at inspected revision | Disposition |
| --- | --- | --- |
| Dashboard still clears recorded task progress | `app/dashboard/projects/actions.ts:105–143`, especially line 113, sets non-done progress to null. The MCP handler was repaired, but the dashboard was not. | INT-5 is only partially satisfied across surfaces. Share transition logic and preserve 60% when moving in-progress to blocked; define reopened completed work consistently. |
| Assessment editor removes supporting JSON keys | `app/dashboard/portfolio/actions.ts:515–529` replaces the entire field with `{body}` or `{}`. The AI document generator also replaces it at line 735. | INT-2 must cover manual and generated dashboard documents as well as MCP. Patch body while retaining evidence and decision keys; require explicit replacement/removal. |
| Hard deletion can remove financial/project history | `finance/actions.ts:38–44`, `assets/actions.ts:30–35`, `land/actions.ts:34–39`; Projects `deleteProject`, `deleteTask`, `deleteMilestone` at lines 212–258. Project deletion removes updates/tasks/milestones in separate calls. | Prioritise a retained archive/tombstone or explicit governed correction path. Preserve content and relations before permitted deletion. Multi-row operations need atomic failure behaviour. Existing source code alone does not establish historical deletion incidence. |
| A logged action is not a retained record version | `app/dashboard/tasks/actions.ts:42–59` deletes then logs empty metadata. Status changes log the new status, not the old record. PBA requirement/stakeholder edits and Project risk/issue edits carry latest provenance but do not retain prior text/state in these handlers. | INT-1/7: distinguish event-only logging from reconstructable change history; fail safely if history cannot commit. |
| Scheduled valuation writes can erase earlier valuation notes | `app/api/agents/market-intel/route.ts:685–734` directly updates current values and notes after a research-grounding check. The dashboard already inserts `asset_valuations` / `land_valuations` before updating current values. | Reuse valuation history tables for cron and tool paths; keep valuation commentary separate from general asset/tenure notes. Preserve grounding and source evidence. Do not equate search grounding with an approved valuation. No live valuation was changed in this audit. |
| Memory supersession bypasses the repaired archive tool | Daily digest line 405, GIS line 520 and Farm Intelligence line 307 directly set `archived_at`, then separately insert a replacement. `scripts/seed-os-memory.mjs:45–81` does the same through raw REST and does not check the archive response. | KL-03/INT-4: old bodies survive, so this is not body destruction, but reason/successor linkage and atomic replacement are missing. Preserve an active valid source if replacement fails; concurrent runs must not produce conflicting current entries. |
| Source refresh is not necessarily source version preservation | `lib/eiras/document-downloader.ts:130–217` writes a destination file and upserts hash/metadata by path/citation; repeated same-path content may replace evidence. | For changed content at the same URL/path, preserve immutable prior bytes and source version before updating the current pointer. Unchanged-hash refresh can remain a cache update. Actual affected files were not quantified. |
| Seed replay can rewrite governance history | `scripts/seed-eiras-portfolio-retrofit.mjs:577–613` upserts initiatives, then deletes source-tagged gate/stakeholder/elicitation/requirement children before re-inserting. It is exposed as `portfolio:seed-eiras` in package scripts. | Inventory administrative writers, distinguish one-time historical migration from normal operation, and prevent replay from replacing reviewed records. It was not executed. |
| Some multi-record money operations lack one atomic commit | `lib/sales/buyerPayments.ts:179–189` changes ledger dates then payment date; payment receipt uses a compensating delete after failure at lines 220–249. | Preserve the existing ledger evidence, but test partial failures, concurrency and retry identity. Compensation is not proof of transaction-wide atomicity. Do not describe every receivable update as missing history: the payment/ledger row is valuable evidence. |
| Existing partial protections should be extended, not duplicated | `lib/production/videoWorkflow.ts:29–38` checks version/metadata before writing, but writes its audit afterwards and reports “Job saved but audit recording failed”. | Retain its conflict guard; bring mutation/history under the same commit guarantee. Apply by record and caller, including workers. |

The session snapshot writer has the same bypass: `scripts/write-state-snapshot.mjs:113–140` archives and inserts session memories with raw REST when passed `--session-note`. Include that generator in KL-04/INT-4; do not consider the closeout procedure exempt. This review records its substantive outcome through Portfolio tools and uses any snapshot refresh only as a derived read/export.

## Protections already present and false positives

- `017_approval_events.sql` records approval status transitions with a new-state snapshot. It is not a general before/after history of every payload edit or deletion.
- `20260724190000_production_handoff_integrity.sql` records Production status transitions and rejects handoff without an output. It does not retain every previous brief, URL, selected-media list or QA note.
- These repository triggers contradict any broad reading of the earlier triage as “no history triggers anywhere”. Their existence in migrations is repository evidence; installation and behaviour in the live database were not reverified here.
- `system_state` key `current` is explicitly a generated snapshot. Regenerating it need not create another canonical company history store; preserve source references, expiry and owner authority instead.
- Project `progress_pct` is derived from tasks. Latest-value GIS/block caches should not be counted as lost decision history merely because they update. Where historical measurements support a claim, separately verify a retained observation/version source; this audit does not assume every cache has one.
- Asset/land valuation rows, payment ledger entries, append-only project updates and existing approval events are foundations to reuse.

## Reconciliation of the three reported repairs

Existing `tests/contracts/knowledge-layer-intact-fixes.test.mjs` checks the source shape of the MCP fixes; these are structural assertions, not database transaction tests.

- INT-4: MCP archive reason/before-state logging exists and refuses archive when that log fails. Scheduled and seed writers still bypass that path; two calls remain non-atomic.
- INT-5: MCP ordinary non-done moves preserve progress. Dashboard still nulls it. Global completion cannot be claimed.
- INT-6: MCP reopening stores the old completion date first and refuses on audit error. Repeating a completed status still assigns a new timestamp without that reopen branch. Shared reconciliation also derives milestone state. Test repeat completion, reopen/re-complete and competing writers before claiming first-completion preservation across paths.

Requirements remain proposed and unverified; this assessment does not silently approve or certify them.

## Proposed implementation scope

1. **First repair package — prevent known loss.** Unify task progress transitions; make PBA document edits key-preserving; preserve repeated completion dates; route memory supersession through one governed operation; prevent hard deletion of finance/project evidence without retained history and an explicit correction/archive policy. Cover UI and MCP together. Add executable fixture and database failure tests.
2. **Shared history boundary — protect money and decisions.** Use atomic database-backed mutation/history protection for explicit governed tables, with verified actor/source, request identity and concurrency guards. Cover finance/receipts, valuations/tenure, PBA documents/requirements/elicitation, project controls and contact notes. Reuse existing audit/event tables where their schema and retention are adequate; do not create another knowledge store. Choose per-table transaction/RPC/trigger design after verifying schema, access and caller inventory, not a public arbitrary-table write endpoint.
3. **Scheduled and administrative writers.** Apply the same protections to valuation jobs, memory generators, source refreshes and seeds. Keep derived snapshots/caches lightweight. Prove retry safety and source-version retention; do not run dormant seed scripts against live records to test them.
4. **Discovery and operating proof.** Deliver KL-01/02/03/04 against preserved sources; reuse the existing FINDABLE set and real held-out work. Source correctness, stale exclusions and durable session decisions are agent-evidenced checks, not another CEO survey.

Acceptance for each package: same rules across registered writers; no unrelated key loss; durable before/after or lifecycle evidence; audit failure cannot leave an unaudited successful change; concurrent stale writes rejected; repeat request does not duplicate money/history; approved outward payloads remain frozen; historical evidence remains available with clear labels. Follow the existing gate/owning-module repair boundaries. No new spending, publication, project creation or blanket build is authorised here.

## Verification and limits

Two in-memory checks executed extracted dashboard functions against fixture databases:

- `{status: in_progress, percent_complete: 60}` → blocked produced `percent_complete: null`.
- Editing a document containing body/evidence_refs/decision produced only the new `{body}`.

These confirm the checked-out write payload behaviour; they do not reproduce a production incident or establish loss counts. Source-only tests of the three earlier repairs passed. The full suite initially found a documentation wording regression from the preceding benefits-review update; the evidence classifications and non-subjective safeguards were restored to the current review guidance without restoring repeated CEO scoring. Final suite: 469 tests, 467 passed, 2 skipped, 0 failed. The two behavioural reproductions also passed as expected demonstrations of the defects.

Live Portfolio was read before update: 13 proposed requirements, G1, existing assessment still planned. This assessment marks that bounded work conducted and records the implementation scope. Live impact attribution remains unverified because current rows alone cannot reconstruct missing historical states. Any broader deployment assurance requires checking installed triggers and exercising rollback/concurrency in an isolated database.

## First repair package implemented — 9 October 2026

Implementation c167a3e74a9158b8c6ae4e4351a91eb8c3a6021f; migration 20261009191508; docs/projects/2026-10-09_company-knowledge-layer-first-repair.md. Local build/typecheck/lint/auth/bound/parity pass; main suite 467 pass, 3 skips, 0 failures; isolated Postgres 17 suite 10 pass. Production deployment READY and two new governed tools visible.

Task progress, milestone lifecycle, portfolio document keys, scoped deletion history, valuation observations and generated-memory replacement now have database protection shared across their writers. Full report: docs/projects/2026-10-09_company-knowledge-layer-first-repair.md. This supersedes the earlier assessment statement that no runtime repairs had been made; that assessment remains historical evidence. The initiative remains G1. Remaining work includes broader write-path coverage, contact-note preservation and source discovery/currency. No repeat CEO questionnaire or blanket November 1 wait.

Business caseBenefits planCharter basisRecommendationNext action

Productization and commercial readiness

TypeData readinessLaunch readiness

Public URL

Target usersPaying customers

Free vs paid boundaryPricing hypothesisService delivery model

Success metricsGo / no-go criteria

Productization plan

044bc747ConvertedFullG3Enabling BackboneLaunch: Not StartedData: Partial

GAB Enterprise Architecture Consolidation — ACOI Follow-On

SOURCE SIGNAL (CEO direct, 2026-09-08). Daniel Nsowah: architecture items should lead to a new initiative following up on ACOI, where architecture is looked at HOLISTICALLY — removing the old mistakes and doubles that have been built on top of, instead of addressing them once and for all. Stated dependency: the concurrent design initiative "will not be meaningful if the architecture is confusing and overlayered." LINEAGE. Baobab discovered and revealed the enterprise architecture as it exists. Canopy turned that into a maintained Enterprise Map and Architecture Atlas — but Canopy's own boundary explicitly excluded "product renaming" and "restructuring project", so it MAPPED the architecture and was barred from streamlining it. ACOI delivered the embedded operating-context layer and then produced a set of architecture questions that were recorded and never triaged. This initiative is the consolidation step that was never opened. ACOI OUTPUTS STILL UNTRIAGED (all created 2026-07-14 from initiative 28fb7f45, all due 2026-09-30, all still status "new"): plugin packaging for the GAB OS operating context layer (76280206-c529-4671-8ea1-5d7d3d10903d — "No plugin benefit was demonstrated by this project"); GIS-specialist reasoning capability (bd5d54b0-99a3-4b84-9d96-976a9497a16c); company-specialised model or training strategy (98b88d17-8731-47dd-b3a7-a9fc5030f75f). Also outstanding from ACOI: enforce enum membership in the central tool input validator (09256d55-7da1-41fc-a196-924182ba8c61, change_request, was due 2026-08-31, status new) and maintain the ACOI context layer after closure (6d111446-0951-4915-8352-f8bedd564025, was due 2026-08-13, status new). MEASUREMENT DEPENDENCY. The CEO expected the 13 August telemetry and measurement under ACOI to have informed any architecture change and the plugin question. It did not, for a specific reason: feedback 06137b87 established on 2026-08-31 that the closed ACOI project measured only the in-app Chief of Staff, while the CEO's real operating surfaces had moved to Codex, Claude, Kimi and Gemini — zero eligible normal routes before or after the 9 August tool-surface decision. The corrected cross-agent measurement is collecting 1-30 September 2026 with benefits review 1 October 2026 (review pack dae7aa72-61c2-41e6-b8db-b8c78c86a2b5). This initiative's evidence base should therefore consume the 1 October result rather than pre-empt it. EVIDENCED EXAMPLES OF ACCUMULATED DOUBLES AND UNOWNED BINDINGS (verified 2026-09-06 to 2026-09-08, illustrative not exhaustive). 1. WEBSITE AND ROUTING HAVE NO CONTRACTED OWNER. Farm Explorer Live task 1ec5ff33 recorded in its own pre-build evidence: "docs/modules/comms.md owns company website management; docs/modules/website.md is missing; MCP tool map has no website deployment tool." Comms owns the website by contract while no website module contract exists and no deployment tool is mapped. That task repaired a live split where canonical gabclimatesmart.com/farm-explorer redirected to an older /partnerships/explore experience while the READY deployment served the current catalog; e9a3e7f9 then added gab-website CI and five-route canonical verification. The specific defects were fixed; the ownership gap that produced them was not. 2. PRODUCT RENAMING DONE AT THE REGISTER, NEVER BOUND TO ROUTING. The 17-product register carries the renamed EIRAS family. Two products have already dropped the Lite suffix in the register (Regulatory Roadmap & ESG Pack; Efficiency & PMO / Sustainability Maturity) while their subdomains still read regulatory-lite and efficiency-lite. Subdomains ramsar and wetland-flood are not products at all — they are legacy names predating the rename, now serving SiteSense and FloodSense respectively. flood and wetland-flood serve the SAME FloodSense application. Tool products carry no URL in the register (FloodSense shows "delivered by: not recorded" and no url) while Farm Explorer Live does, so nothing reconciles a subdomain to a product. 3. DIVISION TAXONOMY EXISTS IN AT LEAST FOUR VARIANTS IN CODE. Migration 20260625060000 established six divisions (farm, agribusiness, consultancy, land, management, shared) and documented that a stale five-item list had been "baked in four application-code copies", silently blocking management and shared — real divisions at roughly GHS 15k/month — from ever receiving a budget row. Live code still carries: the six-division list; a seven-item variant adding ai_operations (lib/comms/receipt-tagger.ts); a five-item variant with Carbon (lib/voice/persona.ts, which is what the voice agent tells callers); and a seven-item category enum with carbon, corporate and platform (orchestrator route.ts). FRAMING DISCIPLINE INHERITED FROM THE DESIGN INITIATIVE. Accumulation is the normal condition of a continuous OS and is not evidence of organisational failure. This initiative exists to consolidate deliberately, once, rather than to catalogue faults. Closed projects (Baobab, Canopy, ACOI, Farm Explorer Live) remain closed and are read as inputs, never reopened.

Owner: Daniel NsowahCategory: AIOPS / System / Enterprise ArchitecturePriority score: Not scoredLatest gate: G2 go Drive doc

Open 6-domain workspace →Open project

Analysis record

StatusStage gateGovernanceDecision

PriorityTailoringFitValueRisk

Needs assessmentBusiness caseBenefits planCharter basis

# GAB Enterprise Architecture Consolidation — ACOI Follow-On
## Charter basis v1.0 — draft for Daniel's G2 review
Prepared 8 September 2026. Portfolio initiative 044bc747-1ee6-424d-9c0d-14990d53f572. Full governance. G0 and G1 are approved; this document does not record G2 approval, authorize implementation or spending, or start any code change.

## 1. Purpose and authority
Produce an architecture GAB can ADOPT — settled enough that the Design and Visual initiative (project ed16b3ff) builds on it rather than inheriting accumulated layers. The CEO's standard, stated 2026-09-08: the design work "will not be meaningful if the architecture is confusing and overlayered."
This is the consolidation step the lineage never authorized. Baobab discovered and revealed the architecture. Canopy turned it into a maintained Atlas but its own boundary explicitly excluded "product renaming" and "restructuring project" — it mapped, and was barred from streamlining. ACOI delivered the operating-context layer and produced questions that were recorded and never triaged. Work has continued to be layered on top of an architecture nobody was authorized to change.
Governing principle, CEO 2026-09-08, recorded to strategic memory 4ca41064: "we are not excluding anything, we just need to sequence them in a way that all the initiaves provide their best inputs for the final outcome." Nothing here is dropped; Strand B is sequenced behind this work.

## 2. Scope and outcomes
Strand A only. Four outcomes, each of which must change OWNERSHIP or BINDING rather than correct a current value:
- Every live public surface resolves to a named product in the register, with its URL recorded and a stated rule binding subdomain to product name.
- Module contract coverage complete for surfaces that have none: website first, eiras second, and the modules registry updated to match.
- One division taxonomy, consumed from a single source by every code path that currently declares its own.
- Duplicate surfaces disposed of explicitly — resolved, or retained with a recorded reason and owner.
EXCLUDED. Reopening Baobab, Canopy, ACOI or Farm Explorer Live; all remain closed and are read as inputs, never re-derived. Re-baselining the enterprise architecture Canopy already mapped. Redefining the company or its divisions, which are settled. Visual and design decisions, which belong to ed16b3ff. Plugin packaging, fine tuning and GIS specialisation, deferred by CEO decision to after this work delivers. Any proposal that changes operating ownership beyond the four outcomes above returns for an explicit scope decision.

## 3. Evidence and known limits
Baseline verified 2026-09-08 and recorded in the G1 needs assessment. Module contracts: 27 on disk in docs/modules for 28 registered modules; eiras has no contract document; website has neither a registered module nor a contract, while docs/modules/comms.md owns company website management and the MCP tool map exposes no website deployment tool. Product register: 17 products carrying the completed EIRAS renaming; tool products record no URL (FloodSense shows "delivered by: not recorded"); two register names have dropped the Lite suffix while their subdomains have not; ramsar and wetland-flood are not products at all; flood and wetland-flood serve the same FloodSense application. Division taxonomy: migration 20260625060000 established six divisions and documented a stale list baked into four application-code copies; four variants remain live.
KNOWN LIMITS, to close in WP1 before estimates become commitments. The public surface list is NOT yet complete: three subdomains were inspected live and eight Lite tools were identified by repository directory, but no exhaustive enumeration of live public routes and subdomains exists. Inspection to date covered gab-os and three subdomains; gab-website is a separate repository and was not audited for surface coverage. Contract-coverage counts are from file presence, not from reading each contract against CONTRACT_STANDARD.

## 4. Delivery work packages and review points
### WP1 — Surface enumeration and baseline
Enumerate every live public surface across both repositories and all subdomains. Establish current owner, serving application, product-register match and version for each. Close the coverage limits in section 3. Confirm capacity and produce the integrated G3 scope/schedule/quality baseline. Exit: Daniel accepts the surface register and the plan; no discovery restart afterwards.
### WP2 — Product-to-surface binding
Bind each enumerated surface to a named product with its URL recorded in the register. Propose the subdomain-to-product rule and the disposition of each duplicate — including whether flood or wetland-flood is retired, and what becomes of the legacy ramsar name now serving SiteSense. Naming proposals require Daniel's selection before use; this work package proposes, it does not decide. Exit: Daniel selects the binding rule and each duplicate disposition.
### WP3 — Contract coverage
Author docs/modules/website.md against CONTRACT_STANDARD, defining ownership of domains, routing, deployment and public surface release, and reconciling it with comms.md which currently holds company website management. Author the eiras contract. Register website in the modules registry. Identify any MCP tool gap the website contract implies and flag it as a Lane B gap rather than building it here. Exit: contracts reviewed and merged; modules registry matches contracts on disk.
### WP4 — Single-source taxonomy
Establish one division taxonomy consumed from a single source, and retire the superseded declarations rather than leaving them in place. All code change under this work package passes the Lane B pre-build gate with recorded classification and evidence. Exit: one declaration remains, demonstrated by a repeatable check.
### WP5 — Retirement discipline
The mechanism, and the reason this initiative is not another point fix. Establish how a superseded copy is retired rather than supplemented: the contract rule that assigns the duty, and an automated check that fails when a second declaration of a bound value appears. Without this, the four outcomes above decay exactly as their predecessors did. Exit: the check runs in CI and demonstrably fails on a seeded duplicate.
### WP6 — Handoff, G5 acceptance and G6 closure
Deliver the naming and product-relationship answers to project ed16b3ff in a form it can consume without deciding anything. Hand over the surface register, contracts, taxonomy source and retirement check with their owners. Schedule benefits review after accepted delivery. Exit: Daniel accepts; the design initiative confirms it can proceed on the delivered architecture.

## 5. Responsibilities
Daniel Nsowah: sponsor, naming and disposition selector, gate and acceptance authority, capacity prioritization.
Portfolio & BA: charter and G2 decision record. Projects: G3–G6 delivery and control owner after promotion.
Engineering role: contracts, taxonomy source, retirement check, CI integration and deployment evidence. Comms: currently holds website ownership by contract and must agree the reconciliation in WP3. Product Operations: owns the product register and the surface bindings recorded in WP2.
Roles are proposed allocations, not claims that anyone has been booked.

## 6. Effort, duration, budget and capacity
Initial planning estimate, not a quote or approved baseline: WP1 2–3 focused working days; WP2 2–3; WP3 3–4; WP4 2–3; WP5 2–3; WP6 1–2. Total 12–18 focused working days, plus a 25% allowance for integration uncertainty (approximately 15–23 days). These are estimator judgments; no measured historical productivity exists, and agent runtime is not equivalent to this effort.
CAPACITY IS THE BINDING CONSTRAINT AND IT IS ALREADY ALLOCATED ELSEWHERE. The CEO allocated first call on his review time to the design initiative on 2026-09-06, with all five other active projects continuing unpaused. This initiative has four CEO decision points (WP1 baseline, WP2 bindings and dispositions, WP6 acceptance, plus gate decisions). It must be sized to land alongside the design initiative's WP1–WP3 rather than compete with them. Before promotion the CEO must state what review allocation this work receives — an explicit workable allocation, not a cleared backlog.
No monetary budget is sought. No external purchase is implied.

## 7. Transition and release constraints
Any subdomain retirement or renaming requires link and reference migration, and release through applicable approval paths. Public URL changes affect live tools with real users; no surface is retired without a redirect plan and a recorded rollback.
The design initiative must not be blocked while this runs. Where a naming answer is not yet settled, the design work proceeds on the visual system and defers only the specific naming-dependent items.
All code change passes the Lane B pre-build gate per standing CEO decision. Contract and test updates ship in the same change as behaviour change.
Closed projects are not reopened. Where evidence originates in a closed project it is cited, not re-derived.

## 8. Acceptance and benefits
Primary acceptance measure, the CEO's own test: the design initiative can proceed on the delivered architecture without raising architecture questions of its own.
Supporting checks, all countable against the G1 baseline:
1. Every enumerated public surface has a recorded product binding and URL.
2. Module contract coverage complete for website and eiras; registry matches disk.
3. Division taxonomy declarations in code reduced to one.
4. Every duplicate surface resolved or retained with a recorded reason and owner.
5. The retirement check runs in CI and fails on a seeded duplicate.
6. Project ed16b3ff confirms it received consumable naming and product-relationship answers.
Benefits review after accepted delivery, repeating the baseline comparison. No revenue, conversion or engagement benefit is claimed — this is enabling-backbone work. The 1 October cross-agent benefits review measures agent operating behaviour and must not be substituted as evidence for this initiative.

## 9. Decisions needed at G2 and boundaries
For Daniel's review: accept or revise this scope and work-package sequence; confirm the review allocation this initiative receives given the design initiative holds first call; accept or revise the provisional effort assumptions.
G2 authorization is separate from approval to prepare this document. G3 approves the detailed delivery baseline. Naming and disposition selections are made at WP2. Acceptance at G5, handover at G6.
Current state remains charter/G2 awaiting review, with no Project created and no code, contract or surface change made.

RecommendationNext action

Productization and commercial readiness

TypeData readinessLaunch readiness

Public URL

Target usersPaying customers

Free vs paid boundaryPricing hypothesisService delivery model

Success metricsGo / no-go criteria

Productization plan

d6ed5f3eConvertedFullG3Enabling BackboneLaunch: PlanningData: Partial

GAB Identity, Naming & Design System Re-architecture

FRAMING SET BY CEO 2026-09-04 — READ THIS BEFORE ANY OTHER FIELD. WHAT THIS INITIATIVE IS. GAB has run a deliberate lineage of enterprise-improvement initiatives aimed at the OS itself: Project Baobab (discover and reveal the enterprise architecture exactly as it exists), Project Canopy (turn that discovered knowledge into a maintained Enterprise Map and Architecture Atlas), and the follow-on agent context and operating-telemetry work. That lineage overhauls and improves the OS in terms of ARCHITECTURE AND CODE. The Design Initiative is the counterpart arc aimed at REDESIGN. It is the same kind of undertaking, applied to the design estate rather than the architectural one. WHAT THIS INITIATIVE IS NOT. It is not a defect-repair project and must not be recorded as one. Type, logo and naming all have documented canon; divergence from canon is not the justification. The OS is continuous and things are built on things — accumulation is the normal condition of a working system, not a failure. The CEO's stated reason is forward-looking: the old design does not serve what he wants the company to BE and to GO. Agents drafting against this initiative must stop producing defect narratives; three were produced and rejected on 2026-09-04. METHOD DISCIPLINE — CEO INSTRUCTION 2026-09-04. Do not ask the CEO to restate things already established in the OS. Where an answer exists in a prior initiative, project, pack or contract, read that artifact and cite it. Asking the CEO to re-express a settled position produces a near-identical restatement, and those restatements ARE the duplicates the enterprise work keeps finding. This is a standing rule for this initiative, not a one-off correction. The CEO pointed to Baobab and Canopy for exactly this reason. AUTHORITATIVE SOURCES FOR COMPANY DEFINITION AND DIVISIONS. Project Canopy (pm_project 0f8556af-dc41-43dc-98d2-0a0afe7c445f, closed 2026-07-12, nine approved packs) and Project Baobab (8780744b-2ef4-4ff7-ab00-1be3043d4089). Specifically WP2 Enterprise Entity and Capability Map (pack 27681357-aa05-4008-83d1-78e8db700c1f), WP3 Source-of-Truth and Governance Atlas (e68817e9-8751-4c37-ae12-f6ad4b24de81), WP4 Product-Service-Module Relationship Map (bd5bdf40-f072-41c1-9c63-31cf3de3c6b6), WP5 Agent Navigation and Operating Map (bf205228-1dac-4432-929f-8cdfe7f62f88), and the Pre-G5 Independent Strategic Review and Future Direction (1967c19b-dcd7-41f9-b824-aea72cf91568). Company definition and division taxonomy are settled there. Code-level lists (lib/company.ts, scripts/gab-wakeup.mjs strategic_canon, lib/voice/persona.ts, finance division arrays) are downstream copies of varying age and are NOT the definitional source; an earlier draft of this analysis wrongly treated them as such and the resulting claim that GAB has no single authoritative company definition is WITHDRAWN as incorrect. TARGET SHAPE OF THE OUTCOME. Canopy is the model for what good looks like: not a document but a maintained, canonical, navigable map that both humans and agents operate from. The design outcome should take the same form — a maintained design source that every surface derives from — rather than a static guide that surfaces copy from by hand. ORIGINAL INTAKE SIGNAL (CEO direct, 2026-09-01) retained for record. Daniel Nsowah raised that the GAB palette has become common and overused, that website, media outputs and the OS interface all derive from the same green/gold system, and that design capability has advanced enough to warrant an uplift. Stated intent is company-wide rebrand and reorganisation at rebrand level, explicitly including decomposition of how GAB OS itself is organised, so the company meets emerging standards.

Owner: Daniel NsowahCategory: Enterprise Identity / Brand Architecture / Design SystemPriority score: Not scoredLatest gate: G1 go Drive doc

Open 6-domain workspace →Open project

Analysis record

StatusStage gateGovernanceDecision

PriorityTailoringFitValueRisk

Needs assessmentBusiness case

Business case: CEO-confirmed 2026-09-06: fresh company-wide design and visuals because current colours and much of the aesthetic have become too commonplace. Use the documented brief; do not restart discovery or require proof of business harm. Spatial specificity and restrained/accent palette are prior agent suggestions, not approved creative requirements. G1 is approved; prepare G2 charter. Full original scope and eight success metrics remain.

Registered legal name fixed; trading identity and expression in scope. CLICOM outgoing-version pinning and synchronized generation/visual-QA migration are required before token changes ship. Blue Carbon handover must match its delivered report and be checked with its owning project. Capacity checked 2026-09-06: 5 active projects, 69 open tasks, 29 overdue across per-project queries; actual allocation remains unresolved. No Project promotion before explicit capacity reconciliation and G2 approval. No implementation or visual direction selected by charter preparation.

See charter v1.0 for deliverables, planning estimates, assumptions and decision points.

Benefits planCharter basis

# GAB Identity, Naming & Design System Re-architecture
## Charter basis v1.0 — draft for Daniel's G2 review
Prepared 6 September 2026. Portfolio initiative d6ed5f3e-bbc3-4b83-91aa-8fd201f4d85d. Full governance. G1 is approved; this document does not record G2 approval, select a visual direction, authorize spending or start implementation.

## 1. Purpose and authority
Give GAB a fresh company-wide design and visual identity because its current colours and much of its aesthetic have become too commonplace to express the distinctiveness Daniel wants. The outcome should express the company as established and where Daniel wants it to go, across external surfaces and his everyday experience of GAB OS.
Daniel confirmed this interpretation and authorized charter preparation and record reconciliation on 6 September 2026. This follows the recorded G1 go decision 3b00aef2-b813-43ca-aecb-d3d386e7c1d2. The justification is forward-looking renewal. Existing accumulation is not evidence of organizational failure and a fresh discovery interview is not required.
Baobab and Canopy provide company definition, relationships and source ownership. This initiative delivers their design counterpart: a maintained source humans and agents operate from. No particular palette, restrained-accent formula, spatial theme, logo treatment or admired-brand reference is predetermined.

## 2. Scope and outcomes
- Company identity and brand relationships: company, GAB Farms, services, sub-brands, EIRAS family, public tools and internal OS; define which name and expression appear where using established enterprise relationships.
- Naming as presentation: coherent product naming convention, explicit Lite-suffix decision and canonical EIRAS expansion; proposals require Daniel's selection before use.
- Visual system: palette, typography, mark treatment, imagery, composition, iconography and data presentation, with coherent variation rather than one repeated layout.
- Motion: timing, transitions, pacing, lower thirds, end cards, captions and reduced-motion treatment. Series may retain legitimate narrative and format conventions within versioned identity rules.
- Application: website, OS interface (including its visual hierarchy and presentation), public tools, reports, documents, decks, letterhead/signature, social formats and CLICOM.
- Maintained design capability: canonical versioned assets/rules/tokens, human navigation, agent consumption, generation/QA alignment, audit, ownership and a module contract.
Registered legal name remains fixed. Company definition/division taxonomy, enterprise re-baselining, Canopy reconstruction, agent-routing redesign, new product functionality and unrelated copy/lead-capture repairs are excluded. Any proposal changing operating ownership or workflow semantics returns for an explicit scope decision.

## 3. Evidence and known limits
Use the existing E1–E6 records and the CEO amendments without re-deriving settled answers. Approved Canopy packs read for this charter:
- WP2 Enterprise Entity and Capability Map: 27681357-aa05-4008-83d1-78e8db700c1f.
- WP3 Source-of-Truth and Governance Atlas: e68817e9-8751-4c37-ae12-f6ad4b24de81.
- WP4 Product-Service-Module Relationship Map: bd5bdf40-f072-41c1-9c63-31cf3de3c6b6.
These establish a company spanning agriculture, environmental/NRM services, climate/GIS capability, consultancy and AIOPS; no single product theme should silently define the whole identity. Historical pack gate terminology yields to the current Projects contract.
E1 identifies the website, OS, eight tools sharing LiteToolShell, three separate public tools, report/document and social generators, corporate stationery and three CLICOM bibles. It was a source inspection, not a rendered-surface review; flood, ramsar and wetland-flood were not inspected. Close these bounded coverage gaps in G3 and baseline the full surface list before implementation estimates become commitments.
E3 market conclusions remain recorded prior analysis, not newly verified facts or binding design directions. E5 supplies no adequate external conversion baseline; no revenue or conversion uplift is promised.
Child elicitation rows still say conducted/unconfirmed, while G1 reviewed them. Gate acceptance is sufficient to proceed to charter; it does not establish independent verification of every research claim. Preserve this distinction and historical gate records.

## 4. Delivery work packages and review points
### WP1 — Brief, traceability and G3 baseline
Reuse the approved brief; map all eight original success metrics and BR-001–BR-009 to outputs and checks. Close E1 coverage limits, resolve current asset/repository owners and make the surface/version register. Confirm capacity, source references, tooling feasibility and estimates. Deliver an integrated G3 scope/schedule/cost/quality baseline. Exit: Daniel accepts the plan; no discovery restart.
### WP2 — Visual direction comparison
After G3 authorization, develop three materially different candidate systems from the documented brief. Use comparable real content and show each on an OS overview and dense working screen, website, a public-tool input/result, report/deck sample, social composition and short motion sample. Include farming and institutional/advisory contexts so product imagery does not narrow the company. Explain distinctive choices, limitations and naming/brand implications. Exit: Daniel selects one direction, requests refinement or rejects all. This is a distinct decision after charter approval.
### WP3 — Selected identity and design system
Develop the chosen direction into complete identity, naming/relationship rules, component and composition families, document/media/motion conventions and responsive/accessible treatments. Preserve established product and service semantics. Deliver editable masters and executable specifications with variation examples. Exit: Daniel approves system specification and key prototypes; verify coverage against WP1.
### WP4 — Maintained source and governed integration
Select and implement the canonical design source and delivery mechanism under the approved G3 plan; do not preselect a vendor in this charter. Define versioned exports/adapters for code, documents and generation prompts; machine-readable QA derives from the same version. Add the brand capability/module contract, operating tools where justified, audit and change/release procedure. All machinery work requires its own evidence-backed Lane B pre-build classification. Exit: representative consumers and generation/QA demonstrate the same approved version, including pinning and rollback.
### WP5 — Surface rollout and G5 acceptance
Roll out in controlled batches: internal/staging pilots first, then website/tools, document/deck/stationery and media consumers in the agreed sequence. Every inventoried surface has an owner, accepted version and tested disposition. Verify representative real workflows and outputs, naming, responsive states, legibility, accessibility, generation/QA and regressions. Release outward changes only through applicable approval paths. Exit: Daniel accepts the delivered design and all mandatory checks pass or a documented exception explicitly limits release.
### WP6 — G6 handover and benefits review
Hand over masters, canonical source, consumer/version register, maintenance contract, release/rollback instructions and QA evidence. Register the delivered capability/product through the existing governed path where appropriate; link operational feedback and benefits ownership. Schedule benefits review 30 days after accepted rollout. Close the finite Project after accepted handover; subsequent upkeep belongs in operations.

## 5. Responsibilities
Daniel Nsowah: sponsor, creative direction selector, gate/acceptance authority and capacity prioritization.
Portfolio & BA: charter and G2 decision record; Projects: G3–G6 delivery/control owner after promotion.
Production/design role: candidate systems, selected design, masters and visual QA; Comms: existing words/story ownership and public rollout coordination.
Engineering role: source/adapters, UI integration, tests and deployment evidence; Reports: reviewable outputs; Data Trust: factual/public claims where applicable.
Brand capability owner: Daniel accountable; operating responsibilities and named executor/reviewer assignments settled at G3. Roles are proposed allocations, not claims that people or external suppliers have been booked.

## 6. Effort, duration, budget and capacity
Initial planning estimate, not a quote or approved baseline: WP1 2–3 focused working days; WP2 4–6; WP3 5–8; WP4 4–7; WP5 5–8; WP6 1–2. Total 21–34 focused working days of combined delivery effort, plus a proposed 25% allowance (approximately 27–43 days total) for integration uncertainty and design revisions. These are estimator judgments based on the E1 surface families, three candidates, representative prototypes, one selected-system refinement cycle and staged integration; no measured historical productivity is available. Agent runtime and elapsed calendar time are not equivalent to this effort.
Provisional scheduling scenario: 6–10 calendar weeks after G3 only if a delivery lane and prompt review slots are explicitly allocated. Rebaseline if selected-direction complexity, source inspection or available capacity differs. No start date is committed.
Live query on 6 September 2026: five active projects, 69 open tasks, 29 overdue. Counts include todo/in_progress/review/blocked, exclude done/superseded/cancelled/archived, and use due_date before 2026-09-06. Queried each active project separately at limit 200; returned counts were below the limit:
- Sankofa follow-up: 12 open / 6 overdue.
- Ghana Systems: 22 open / 1 overdue.
- Vehicle acquisition: 10 open / 10 overdue.
- CLICOM: 18 open / 12 overdue.
- Blue Carbon final report: 7 open / 0 overdue.
Task counts signal competing commitments, not measured available hours. Capacity is checked but NOT reconciled. Before promotion Daniel must decide which delivery/review allocation this initiative receives without assuming existing work is paused. Do not require every overdue task to close; require an explicit workable allocation.
No monetary budget is approved. Existing tooling is an assumption to validate, not a zero-cost claim. G3 must itemize any incremental generation, font/asset licensing, subscriptions and external support using actual costs before purchase; optional purchases require separate authorization.

## 7. Transition and release constraints
Pin the active CLICOM 20-theme series to the outgoing identity where required for coherence; demonstrate pinning before the first token change ships. Move generation and visual judgment together. Preserve series-specific storytelling conventions and natural depicted colours; shared motion rules do not require identical episodes.
Keep Blue Carbon handover/onboarding consistent with the report delivered to the client. Existing records do not prove that handover is settled; confirm with the owning project before dependent release. Do not recolour delivered client packs automatically.
For each surface record outgoing/target version, owner, dependencies, test evidence, approval, release order and rollback. Temporary differences are only accepted when explicitly version-pinned and documented. Prototype work does not alter production.
Changes to naming/public URLs must include link/reference migration and approval; no naming choice is made by this charter.

## 8. Acceptance and benefits
Daniel's judgment of freshness, distinctiveness and authenticity against the G1 baseline is the primary creative acceptance measure. Supporting checks:
1. All eight original success metrics have delivered evidence and disposition.
2. Naming and brand relationships are decided explicitly without redefining the company.
3. Every inventoried surface is covered or has a Daniel-approved, versioned exception; no unexplained drift within a version.
4. Different subject/format samples demonstrate useful variation within one recognisable identity.
5. Static, motion, interface and document outputs are legible, accessible and fit for their real use; G3 sets measurable test thresholds.
6. A controlled source-change demonstration propagates correctly; pinned legacy outputs remain stable; QA and generation agree; rollback works.
7. The OS's representative working journeys remain functional and Daniel accepts its everyday visual experience.
Benefits review repeats the baseline comparison and checks actual use, propagation and maintenance. Do not substitute template counts for creative success or claim increased sales without a valid measurement basis.

## 9. Decisions needed at G2 and boundaries
For Daniel's review: accept/revise this charter scope and sequence, confirm the capacity allocation required before promotion, and accept/revise the provisional effort assumptions. G2 authorization is separate from approval to prepare this document. G3 then approves the detailed delivery baseline; visual direction is selected at WP2; system approval at WP3; release acceptance at G5; handover at G6.
Current state remains charter/G2 awaiting review, with no Project created and no visual or code changes made.

RecommendationNext action

Productization and commercial readiness

TypeData readinessLaunch readiness

Public URL

Target usersPaying customers

Free vs paid boundaryPricing hypothesisService delivery model

Success metricsGo / no-go criteria

AMENDED 2026-09-04 BY CEO DIRECTION (supersedes the G0 wording "evidence beyond founder sentiment"). The CEO's own judgment is the governing standard of authenticity for this initiative. Daniel Nsowah stated: the rule of his sentiment or feeling is the most important thing in authenticity, and the required outcome is a design that is not generic and not common. Elicitation therefore does NOT exist to justify the decision to advance; it exists to characterise what is currently generic so that the new system can verifiably avoid it. GO to G1 on the strength of E6 root-cause separation plus a documented map of the current commonplace. STOP or REDIRECT no longer triggers on "surface coverage rather than identity system" — the CEO answered that test directly at E2 on 2026-09-04 and rejected the consistency-only remedy; anything short of a new system does not meet the stated intent. STANDING OPERATIONAL CONSTRAINTS UNCHANGED AND STILL BINDING: no token change ships until CLICOM's active 20-theme series can be version-pinned to the outgoing system and the visionQA judgment path is migrated in the same change as generation; the Blue Carbon client handover and onboarding materials must be settled against the delivered report; promotion to a Project is held until capacity is reconciled against 5 active projects and 28 overdue tasks. LEGAL CONSTRAINT UNCHANGED: registered legal name fixed at the Registrar; trading identity and expression are in scope, the registered name is not.

Productization plan

1f8bf7f7ConvertedStandardG3Internal PlatformLaunch: PlanningData: Ready

Product Operations Module — live product activation, handoffs and learning

Design a shared operating home for products delivered by completed projects. The capability must let Daniel or an agent open a live product, see its owner, purpose, claim boundaries, operating lanes, product health and next action, and start a governed follow-up such as Public Activation. It must create accountable handoffs to the existing owning modules—Comms, Production, Approvals, Sales, Farm and Finance—and return status, outcome and feedback to the product without duplicating their records or reopening a delivery project.

Owner: Daniel NsowahCategory: System / Cross-module OperationsPriority score: Not scoredLatest gate: G1 go Drive doc

Open 6-domain workspace →Open project

Analysis record

StatusStage gateGovernanceDecision

PriorityTailoringFitValueRisk

Needs assessment

## Problem
GAB now has 17 live products delivered by completed projects, but the Product Registry is a catalogue rather than a working operating home. The follow-up actions that make a product valuable—public activation, commercial conversion, service readiness, evidence maintenance and benefits review—are distributed across valid owner modules. A CEO or agent cannot open a product, identify the next action, begin the right governed workflow and see the returned outcome in one traceable place.

## Root cause
The registry, operational-feedback loop and module workflows exist independently. The current OS has no reusable product-level action/request object and no product-to-campaign linkage. Farm Explorer Live demonstrates the gap: its Public Activation intent remains visible only as a historical WP8 row in a closed project, although Comms and Production are the valid execution homes.

## Desired future state
Every live product has a Product Operations workspace with purpose, operating owner, claim boundaries, current operating lanes, product health, measures and next best action. A bounded request starts a handoff in the owning module and returns only linked status/outcome. The product view never duplicates field truth, campaign records, creative assets, leads, agreements, invoices or payments.

## Validation cases
1. Farm Explorer Live: Public Activation → Comms campaign → Production → Approvals → canonical Farm Explorer enquiries → Sales/Farm/Finance fulfilment → product outcomes.
2. SiteSense Lite: a future activation or service-readiness request must traverse the same reusable product action model while retaining EIRAS/Data Trust claim and HITL boundaries. No activation has been created for SiteSense; it is a design validation case, not a fabricated operating result.

Business case

## Feasibility assessment — 2026-08-13

### Existing usable spine
- `products` is the enduring product registry and is already linked to `operational_feedback`, operational transitions and benefits reviews by `product_id`.
- `comms_campaigns` is already the stable Comms parent for strategy, posts, Production grouping, approvals through posts, and engagement.
- `production_requests` already retain `campaign_id` in metadata; Production can therefore inherit product context through a campaign rather than receive a duplicate product field.
- Sales, Farm and Finance already own buyers, supply agreements, farm truth, invoices and buyer payments. Invoice/payment chain can remain derived from an agreement.
- EIRAS public contributions already support `campaign_id`, which validates the pattern beyond Farm Explorer.

### Verified gaps
1. There is no Product Action Request record or Product Operations dashboard.
2. `comms_campaigns` has no `product_id`; a campaign cannot currently be reliably attributed to a live product.
3. Canonical Farm Explorer website enquiries sit in the separate website database and currently carry no product/campaign attribution. The OS sweep creates the owned response task correctly but cannot prove which product/campaign originated the enquiry.
4. Buyer/agreement/payment records do not carry product attribution. This is acceptable only if the Product Action Request retains the canonical linked agreement and outcome references; do not copy product IDs into every commercial table without a demonstrated need.

### Smallest safe architecture
Create one new Product Operations module with two additive records: `product_operation_actions` (the bounded request) and `product_operation_links` (typed links to canonical records and outcomes). Add `product_id` to `comms_campaigns`. Add optional, privacy-safe product/campaign attribution to website enquiries and carry it through the existing intake reconciliation source context. Do not add product fields to Production, Sales, Farm or Finance where a stable campaign/action/agreement path already expresses the relationship.

### Public Activation proof chain
Product Action (Farm Explorer Live) → linked Comms Campaign (product_id) → existing Comms Posts → existing Production Request (campaign_id) → existing Approval Action → canonical website enquiry (product/campaign attribution) → existing owned response task/contact → Sales buyer/agreement if qualified → existing Farm fulfilment and Finance invoice/payment. The Product Action reads the linked statuses and records no duplicate operational facts.

Benefits planCharter basis

## Product Operations Module — First Release Charter

### Sponsor and owner
Sponsor: Daniel Nsowah. Delivery owner: GAB OS / Product Operations.

### Objective
Create a reusable internal Product Operations capability so every live product has a discoverable operating home, an explicit next action, governed cross-module follow-up and product-level learning—without duplicating the operating modules that execute the work.

### Scope in
1. Product index and product detail/workbench for live registry products.
2. Additive Product Action Request and Product Action Link records.
3. Reusable action types: Public Activation, commercial follow-up, service readiness, evidence/data-quality review, benefits review and retirement review.
4. Additive product attribution on Comms campaigns and explicit product/campaign attribution on canonical website enquiries.
5. Read-only linked-status/outcome aggregation for the Farm Explorer Live Public Activation proof chain and one non-Farm validation product (SiteSense Lite).
6. Contracts, MCP surfaces, provenance, tests and deployment verification.

### Scope out
Direct publishing, creative production, CEO approvals, lead messaging, buyer creation, agreements, Farm ERP changes, invoices/payments, a second CRM/task system, bulk migration/backfill, or a general workflow engine.

### Success criteria
- A CEO or agent can open Farm Explorer Live and start a draft Public Activation request.
- The request creates/links a Comms campaign without bypassing Comms, Production or Approvals.
- Canonical source links make the campaign → Production/Approval → website enquiry → Sales/Farm/Finance outcome traceable.
- SiteSense Lite can use the same action model without weakening EIRAS/Data Trust/HITL controls.
- No source facts are copied into Product Operations and no existing module authority changes.

### Risks and controls
Cross-module link ambiguity: additive foreign keys/typed links only. Authority drift: destination modules remain the sole writers of their records. Public-claim drift: Comms/Production/Approvals controls remain unchanged. Scope creep: six action types only; no generic task/CRM replacement.

RecommendationNext action

Productization and commercial readiness

TypeData readinessLaunch readiness

Public URL

Target usersPaying customers

Free vs paid boundaryPricing hypothesisService delivery model

Success metricsGo / no-go criteria

Productization plan

PSFUApproved For ProjectStandardG2

Project Sankofa — Follow-Up Initiative

Operationalize Sankofa P0-P4 evidence: heat/dry-spell protocols, BR-005 auto-tasking, community engagement packs, crop diversification, and EIRAS consultancy service lines.

Owner: Chief of StaffCategory: FarmPriority score: Not scoredLatest gate: G0 go Drive doc

Open 6-domain workspace →Open project

ae5b00d1Needs AssessmentStandardG1Enabling BackboneLaunch: PlanningData: Partial

Climate Reliability & Local Decision Intelligence

A post-Sankofa capability maturation initiative. Extend Sankofa’s shared GIS/Data Trust foundation into a reusable, uncertainty-aware Climate Reliability Engine for EIRAS, stakeholder/community awareness and GAB operations. Apply the engine first at the Aframso/Sekyere Central farm boundary: use coarse-scale climate data for shared seasonal context, terrain/hydrology and structured farm observations for local exposure, and crop/management state for actionable decisions. Do not claim exact block-level rainfall or temperature where source resolution cannot support it. Address the unresolved P1/BR-002 operational question: onset, false onset, cessation, dry spells, heavy-rain clustering, heat extremes and compound heat-dry conditions; convert material findings into owned action and learning loops.

Owner: Daniel NsowahCategory: AIOPS / EIRAS / FarmPriority score: Not scoredLatest gate: G0 go Drive doc

Open 6-domain workspace →

Analysis record

StatusStage gateGovernanceDecision

PriorityTailoringFitValueRisk

Needs assessmentBusiness caseBenefits planCharter basisRecommendationNext action

Productization and commercial readiness

TypeData readinessLaunch readiness

Public URL

Target usersPaying customers

Free vs paid boundaryPricing hypothesisService delivery model

Success metricsGo / no-go criteria

Productization plan

1816cb4fConvertedStandardG3HybridLaunch: Not StartedData: Partial

Farm Explorer Live: Block-Level Produce Forward Sale

Enrich the existing Farm Explorer tool on gabclimatesmart.com with a public, block-level live data layer that acts as a lead converter for produce sales — turning the buyer information packs GAB already produces manually into an always-on surface. Each production block shows crop, planting month, growth stage, NDVI, GIS layers, and photos/videos from the Farm Manager. Commercial mechanism is commitment-first, not payment-first: buyers express interest and bid on blocks against a published stage-based floor price (deepest discount near planting, rising toward spot as harvest approaches). Payment or part-payment is the reservation mechanism — it locks a block exclusively; without it the block remains open to competing bids. "Auction" means open competitive interest on a block, not formal auction mechanics. Strategic logic: (1) GAB already validates data-pack-driven selling manually — this scales it from one-buyer-at-a-time to public; (2) nobody in Ghana does single-estate block-level radical transparency (Complete Farmer, Grow For Me, BuyFromFarmers prove pre-harvest demand but aggregate smallholders they don't control); (3) the same enriched block layer improves conversion of existing Farm Explorer funnels (land lease, contract growing) — one layer, three funnels; (4) public transparency enforces agronomic discipline; (5) GAB OS already holds the back office (buyers, supply agreements, buyer payments, GIS readings, farm media). Commitment-first structure with no pooled prepayments keeps the model strictly commercial offtake with minimal investment-scheme exposure. Delivery path: lean block layer shipped on Farm Explorer for the 3rd/4th RD cassava blocks (harvest Aug–Oct 2026) as the live pilot; 5th RD blocks (harvest Feb–Apr 2027) provide the full planting-to-harvest cycle as launch inventory for the complete experience.

Owner: Daniel NsowahCategory: FarmPriority score: Not scoredLatest gate: G0 go Drive doc

Open 6-domain workspace →Open project

Analysis record

StatusStage gateGovernanceDecision

PriorityTailoringFitValueRisk

Needs assessmentBusiness caseBenefits planCharter basisRecommendationNext action

Productization and commercial readiness

TypeData readinessLaunch readiness

Public URL

Target usersPaying customers

Free vs paid boundaryPricing hypothesisService delivery model

Success metricsGo / no-go criteria

Productization plan

6ae84d40CharterStandardG1HybridLaunch: Not StartedData: Partial

Tree Registration & Tenure Evidence Platform (TreeReg)

Tree registration to support tree tenure and benefit sharing has persistently failed in Ghana's forestry sector due to manpower limits, operational cost, accessibility, and paper-based processes — prior software attempts (including during GFIP) never produced a working field model. Opportunity: a field-first digital tool that lets a farmer/tree owner (or a trained registration agent) document each planted tree or stand with unique ID, GPS coordinates or polygon, species, photo evidence, owner identity and witness attestation — structured to feed the Forestry Commission's official registration process rather than replace it. Strategic timing: MLNR is finalising tree tenure & benefit-sharing policy reform (Feb 2026: proposal to grant farmers benefit rights to naturally occurring trees + announced national digital tree registration system; EU/EFI-funded stakeholder platform launched). Demand drivers beyond government: EUDR geolocation/polygon compliance for cocoa and timber supply chains, and extension to the Tree Crop Development Authority (cashew, shea, rubber, mango, oil palm). Positioning: GAB is the field capture + evidence-chain + GIS layer (SiteSense/EIRAS DNA), not a competing national registry. Founder advantage: Daniel served as M&E Manager at the GFIP PMU and holds direct institutional knowledge of prior registration attempts, the Timber Tenure Benefit-sharing mechanism, and the implementing stakeholder network.

Owner: Daniel NsowahCategory: ConsultancyPriority score: Not scoredLatest gate: G0 go Drive doc

Open 6-domain workspace →

Analysis record

StatusStage gateGovernanceDecision

PriorityTailoringFitValueRisk

Needs assessmentBusiness caseBenefits planCharter basisRecommendationNext action

Productization and commercial readiness

TypeData readinessLaunch readiness

Public URL

Target usersPaying customers

Free vs paid boundaryPricing hypothesisService delivery model

Success metricsGo / no-go criteria

GO if: (1) [resolved] third-party collection + FC/RMSC sample validation is the Ministry/UNDP-backed policy direction, notwithstanding RMSC's own institutional resistance -- confirmed by CEO's direct DPO/GFIP knowledge. (2) Anchor demand: no further outreach to World Bank/FC/TCDA/COCOBOD right now -- per CEO, none will commit before seeing a working tool; LBCs remain the most plausible funder given prior IDH/WCF-adjacent budget, but even that is best approached with a working proof point in hand. COCOBOD's existing all-cocoa-farm data system (currently missing a tree module) is the single highest-leverage distribution opportunity to revisit once a pilot exists. (3) Aframso pilot proving the capture workflow end-to-end is now the primary near-term gate -- elevated above institutional outreach in sequencing. HOLD if RMSC's revenue-motivated resistance escalates into an actual policy-level block (not just implementer friction). STOP condition (legal non-recognition) remains very unlikely given the reaffirmed Tree Tenure & Benefit-Sharing policy direction.

Productization plan

28fb7f45ConvertedFullG3Enabling BackboneLaunch: PlanningData: Partial

GAB Agent Context & Operating Intelligence

G1 confirmed that GAB's material need is reliable activation of authoritative enterprise context across agent surfaces and module transitions. G2 has defined a bounded phased response: establish baseline evidence, deliver direct integrity repairs, then proceed through authority/manifest, governed retrieval, multi-label transition and cross-agent parity tranches only when evidence supports continuation. Existing modules remain source owners. Plugin packaging, model training and GIS specialization are deferred separate decisions.

Owner: Daniel NsowahCategory: Enterprise Architecture / Agent Operations / GAB OSPriority score: 86Latest gate: G2 go Drive doc

Open 6-domain workspace →Open project

Analysis record

StatusStage gateGovernanceDecision

PriorityTailoringFitValueRisk

Needs assessment

# GAB Agent Context & Operating Intelligence
## G1 Forensic Needs Assessment

**Initiative:** GAB Agent Context & Operating Intelligence  
**Initiative ID:** 28fb7f45-9141-4655-9854-5cf06e29499c  
**Gate under review:** G1 — Justification  
**Governance level:** Full  
**Assessment date:** 12 July 2026  
**Decision authority:** Daniel Nsowah

## 1. G1 Decision Question

Is there sufficient evidence that GAB has a material enterprise need concerning agent context activation, module transitions and operating intelligence, such that bounded G2 charter analysis is justified?

G1 does not authorize implementation. It does not choose a plugin, context broker, retrieval architecture, model training, fine-tuning, GIS specialist, module rewrite or public product.

## 2. Executive Assessment

The evidence supports a genuine enterprise-level need.

Project Baobab established that GAB has mature but distributed institutional knowledge. Project Canopy converted that knowledge into an accepted Enterprise Map and Architecture Atlas. The remaining operating problem is not primarily missing knowledge. It is inconsistent activation of authoritative knowledge when an agent changes module, workflow, evidence source or governance lane.

All eight defects carried into the G0 intake were confirmed against current repository or live operating surfaces. Two additional conditions were identified: the in-app agent duplicates mutable enterprise facts in a static base prompt, and the UI also retrieves the oldest 100 messages for long sessions. The defects do not all share one local code cause, but they share one enterprise cause: each agent surface assembles context independently and GAB has no shared agent-neutral context authority, transition and evaluation contract.

Counter-evidence is material. Wakeup, preflight, Tier 2 contracts, live MCP tools and manual repository access already let a disciplined coding agent operate well. Some symptoms could be resolved by narrow repairs. No current telemetry quantifies the total loss, and no evidence proves that a plugin, training programme or new platform is required. Therefore G1 justifies continued analysis, not a predetermined build.

## 3. Evidence Standard

The assessment separates:

- **Verified live:** current MCP inventory, governed PBA/Project records and report-pack query behavior.
- **Repository-current:** current code, contracts, policies, schemas and local agent skill surfaces.
- **Founder-confirmed operating evidence:** Daniel's repeated experience and operational feedback record.
- **Prior approved evidence:** Baobab and Canopy packs, used where corroborated by current evidence.
- **Unverified:** vendor-hosted behavior not inspectable from this repository, future plugin behavior, model-training value and post-change benefit levels.

An independent reviewer can challenge every finding using the evidence identifiers below.

## 4. Evidence Register

| ID | Evidence | Status |
|---|---|---|
| ACOI-EV-001 | Operational feedback c245c143-b551-44ed-b043-9e1f511336ba and Daniel's G0 approval | Verified live / founder-confirmed |
| ACOI-EV-002 | app/api/agents/orchestrator/route.ts lines 3398-3405: ascending history query with limit 40 | Repository-current |
| ACOI-EV-003 | components/chat/ChatInterface.tsx and app/dashboard/layout.tsx: page context is sent every turn and injected as a focus instruction | Repository-current |
| ACOI-EV-004 | Orchestrator live context reads legacy projects; Projects contract defines pm_projects as canonical | Repository-current contradiction |
| ACOI-EV-005 | In-app prompt says strategic memory is ground truth; Agent Operating Protocol says memory cannot override live state or contracts | Repository-current contradiction |
| ACOI-EV-006 | Live MCP inventory: 177 tools; no contract, Atlas or Enterprise Map retrieval tool | Verified live |
| ACOI-EV-007 | Live query_project_report_packs omits markdown_body; handler select confirms omission although the table stores it | Verified live + repository-current |
| ACOI-EV-008 | Live external tool schemas total about 149 KB; in-app calls pass the complete TOOLS list | Verified live + repository-current |
| ACOI-EV-009 | Skill audit: Codex 44 GAB skills, Claude 13, Kimi 13, repository 12; broken Kimi operating-protocol target; no discovered Gemini GAB skill path; one shared skill differs | Repository-current local environment |
| ACOI-EV-010 | Chat telemetry records tool calls but no module intent, transition, context source, correction event, routing outcome or time-to-useful-action measure | Repository-current |
| ACOI-EV-011 | Static in-app base prompt duplicates mutable company framing, project, product and pricing information rather than deriving it from canonical strategy and records | Repository-current |
| ACOI-EV-012 | Wakeup, preflight, contracts, live context, PBA detail and MCP operation already work and reduce risk when manually followed | Verified strength / counter-evidence |

## 5. Current-State Strengths

G1 does not describe a failing system. GAB already has:

- a canonical Agent Operating Protocol;
- 26 Tier 2 module contracts;
- explicit Lane A and Lane B governance;
- Portfolio-to-Projects lifecycle gates;
- action provenance, HITL approval and Data Trust rules;
- live wakeup and module preflight;
- an operating MCP bridge with 177 externally visible tools;
- approved Baobab and Canopy evidence, maps and maintenance method;
- Reports packs and Drive mirrors;
- current live context queries and strategic memory;
- a repeatable method for separating observations, implications, recommendations and decisions.

The initiative should preserve these strengths and must not replace them with a competing truth store.

## 6. Confirmed Findings

### F-01 — Long-session context is selected incorrectly

The server orders messages oldest-first and limits to 40 while describing them as the last 40. The UI similarly displays the oldest 100. For long conversations, the most recent decisions and topic transitions are excluded.

**Immediate cause:** incorrect pagination/order logic.  
**Business implication:** agents can appear to forget current work and require reorientation.  
**Evidence:** ACOI-EV-002.

### F-02 — Page context can override current conversational intent

The current dashboard page is sent with every message and the system prompt instructs the model to focus on that module. There is no comparable instruction to detect a deliberate topic change or preserve multiple relevant modules.

**Immediate cause:** page location is treated as a dominant context signal rather than one signal among intent, history and enterprise relationships.  
**Business implication:** abrupt shifts can be misrouted or framed too narrowly.  
**Evidence:** ACOI-EV-003.

### F-03 — Live project context violates the canonical source boundary

The in-app live snapshot queries the legacy projects table and limits results to five. The Projects contract says pm_projects is canonical and legacy projects are compatibility records only.

**Immediate cause:** stale context query that was not migrated with the Projects architecture.  
**Business implication:** active and recently closed governed projects can be absent while old records are presented as current.  
**Evidence:** ACOI-EV-004.

### F-04 — Context authority is contradictory

The in-app prompt says strategic memory is ground truth. The canonical protocol says memory, snapshots and handovers are starting points and cannot override live state, contracts or the repository.

**Immediate cause:** authority rules evolved in contracts without being reconciled into the runtime prompt.  
**Business implication:** an agent can comply with one instruction only by violating another.  
**Evidence:** ACOI-EV-005.

### F-05 — Approved enterprise knowledge is not available through governed retrieval

No live contract or Atlas retrieval tool exists. Project pack queries omit the stored Markdown body. Coding agents can manually read the repository, while the in-app agent cannot retrieve equivalent content through its governed tools.

**Immediate cause:** Reports and contracts were designed for human review and repository use before agent retrieval became a requirement.  
**Business implication:** agents have unequal access to approved knowledge and may rely on summaries or memory.  
**Evidence:** ACOI-EV-006 and ACOI-EV-007.

### F-06 — Tool context is broad rather than task-scoped

The live external inventory contains 177 tools and about 149 KB of schema metadata. The in-app agent sends its full TOOLS collection on every model call.

**Immediate cause:** tool parity and availability were prioritized before task-level relevance and transition routing.  
**Business implication:** higher context cost, ambiguous tool choice and greater probability of wrong-module actions.  
**Evidence:** ACOI-EV-008.

### F-07 — Agent skill distribution is not reliably canonical

Skill counts and coverage differ across agents. The Kimi operating-protocol pointer targets a missing repository file. Claude and Codex hold separate operating-protocol copies. One shared creative skill differs between repository and Codex, and Gemini has no discovered GAB custom-skill entry point.

**Immediate cause:** manual, vendor-specific skill installation without a governed manifest and parity control.  
**Business implication:** the same request can activate different institutional knowledge depending on agent.  
**Evidence:** ACOI-EV-009.

### F-08 — Benefits and routing performance cannot currently be measured

Existing telemetry records tool names, success and duration but not the intended module, context source, transition, correction or outcome needed to calculate the approved benefit measures.

**Immediate cause:** observability was designed for tool execution rather than context quality.  
**Business implication:** GAB cannot objectively distinguish improvement from perception or compare small repairs with larger options.  
**Evidence:** ACOI-EV-010.

### F-09 — Mutable enterprise knowledge is duplicated in the in-app system prompt

The base prompt contains a fixed division model, operating claims, named projects, service tiers and pricing. These can drift from the Strategic Canon, company sources and live records.

**Immediate cause:** the prompt predates the mature contract, strategy and source-governance architecture.  
**Business implication:** stale knowledge can be presented with system-level authority.  
**Evidence:** ACOI-EV-011.

## 7. Root-Cause Analysis

### Primary enterprise root cause

GAB lacks a shared, governed and agent-neutral context activation contract connecting canonical knowledge to each agent interaction.

Baobab discovered the enterprise. Canopy mapped it. Neither project was authorized to change runtime context assembly. As a result, the architecture is known but not consistently activated.

### Contributing causes

1. GAB OS grew organically by module and agent surface before the Enterprise Atlas existed.
2. Each agent runtime has a different bootstrap path: static in-app prompt, local skills, wakeup/preflight, MCP inventory or manual repository access.
3. Authority precedence is documented but not uniformly enforced in runtime context.
4. Context retrieval was not designed as a governed cross-module capability.
5. Tool availability was optimized for parity rather than relevance.
6. Skill distribution is manual and vendor-specific.
7. No evaluation layer measures context transitions and correction burden.
8. Several direct implementation defects remain inside the in-app context path.

### What is not yet proven

- that all experienced friction comes from GAB OS rather than model/session limitations;
- that a single shared service is preferable to targeted repairs;
- that plugin packaging provides material value;
- that model training or fine-tuning is necessary;
- that a GIS specialist belongs inside this initiative rather than a separate capability;
- the financial value or time saved because no baseline exists.

## 8. Business Need

GAB needs a governed way to ensure that every supported agent receives the correct, current and appropriately scoped enterprise context when work begins or changes direction, while preserving module ownership, evidence traceability, HITL governance and multi-dimensional relationships.

The business need is context reliability and operating continuity. It is not a requirement to build a plugin or train a model.

## 9. Strategic And Policy Alignment

- **GOV:** strengthens operational discipline through evidence, PBA, Projects and controlled change.
- **GIST:** improves use of GIS, Data Trust and technical intelligence by making source authority and uncertainty available at the point of work.
- **Near-term value:** reduces CEO correction and operating friction.
- **Medium-term value:** strengthens multi-agent operations and safer product/service execution.
- **AI-DATA:** requires source authority, HITL and traceability.
- **RECORDS:** preserves approved packs, contracts and Drive records as retrievable evidence.
- **SECURITY:** requires authenticated, scoped access and prevents context from becoming an information-leak path.

## 10. Risks Of Maintaining The Current State

- repeated CEO time spent re-establishing context;
- wrong-module tools or records used with apparent confidence;
- stale memory or static prompt facts overriding current sources;
- inconsistent behavior across agent vendors;
- cross-module governance handoffs missed;
- future plugin or training work encoding partial architecture;
- Atlas value remaining largely documentary rather than operational;
- inability to prove whether corrective investment creates benefits.

## 11. Evidence Against The Initiative

- Wakeup and preflight already provide strong manual orientation.
- Tier 2 contracts and the MCP Tool Map are mature.
- The in-app agent does fetch live context and strategic memory every request.
- Page context is beneficial for stable within-module work.
- Codex can work reliably when the protocol is followed.
- Claude and Kimi already have parity for a 13-skill core set despite broader Codex coverage.
- Narrow repairs may solve a meaningful portion of the problem.
- No quantitative baseline proves frequency, duration or financial impact.

This evidence argues against assuming a large platform or model programme. It does not remove the need to resolve verified authority, retrieval, transition and parity defects.

## 12. Open Questions For G2

1. What minimum intervention resolves the material need?
2. Which defects should be corrected before options are compared?
3. Which agent surfaces are formally supported?
4. What context is stable, dynamic, confidential or vendor-specific?
5. What is the authoritative ordering among live records, contracts, approved packs, Atlas, memory and conversation history?
6. How should multi-module intent and transitions be evaluated?
7. What tool/context scoping preserves necessary cross-module work?
8. What baseline and target thresholds justify investment?
9. Is plugin packaging useful after the capability is proven?
10. Should GIS specialization be separate from the shared operating context capability?

## 13. G1 Recommendation

**Recommendation for Daniel's review: GO to G2 charter-basis and options analysis.**

The need is confirmed and strategically material. G2 should define a bounded charter, supported agent surfaces, authority rules, baseline plan, option comparison, acceptance criteria, risks and exclusions. It should compare targeted repairs with broader capability options and may conclude that the minimum justified response is smaller than originally imagined.

G1 does not authorize implementation.

## 14. Standing Gate Questions

- **Did G1 achieve its objectives?** Yes. The need, strengths, defects, root cause, counter-evidence, business implications and outcome requirements are documented.
- **Is the evidence sufficient?** Sufficient to justify G2 analysis; not sufficient to choose or build a solution.
- **Are architectural decisions justified?** No implementation architecture has been selected. The authority and governance requirements follow existing contracts.
- **Are there unresolved risks?** Yes. Centralization, scope, source authority, security, vendor coupling, measurement and capacity risks must be carried into G2.
- **Is the initiative ready to progress?** Ready for Daniel's G1 decision. If Daniel does not accept the evidence or scope boundary, hold at G1.

Business case

# GAB Agent Context & Operating Intelligence
## G2 Updated Business Case

## 1. Confirmed Need

G1 confirmed nine current findings and one enterprise root cause: GAB's institutional knowledge is mature but each agent surface assembles context independently, without a shared agent-neutral authority, transition and evaluation contract. The resulting business exposure is repeated CEO recontexting, stale or wrong-source context, uneven agent access, broad tool noise and inability to measure improvement.

## 2. Options Considered

G2 compared current/manual operation, targeted repair, a phased foundation, plugin-first packaging, model training first and a combined GIS specialist. The phased foundation scored highest because it begins with the minimum targeted repair, preserves stop points, covers all requirements and keeps future packaging choices open.

## 3. Recommended Investment Posture

Authorize a bounded internal Project for planning. Do not commit to a broad platform. G3 should estimate internal effort and any infrastructure/subscription cost by tranche. WP1 and WP2 provide the first economic decision point: if baseline and direct repairs resolve most material friction, later tranches can be limited or stopped.

## 4. Expected Benefits

- lower Daniel correction and context-repetition burden;
- faster useful action after transitions;
- fewer stale-source, wrong-module and governance-path errors;
- consistent retrievability of approved enterprise knowledge;
- more proportionate tool/context exposure;
- honest cross-agent activation and parity;
- measurable evidence for future plugin, model and GIS decisions;
- reduced institutional-memory and onboarding risk.

## 5. Benefit Conditions

Benefits count only when source accuracy, evidence quality, security, approvals and module ownership are maintained. Thresholds and cost estimates must be baselined in G3.

## 6. Risks Of Investment

The initiative can over-centralize knowledge, increase blast radius, expose restricted sources, block tools through over-scoping, create vendor coupling or consume disproportionate capacity. The phased architecture, explicit source ownership, access classification, fallback, tests and stop/go reviews are required controls.

## 7. Alternative Of No Investment

Manual controls remain usable and should be retained as fallback, but no-investment leaves confirmed defects unresolved and continues dependence on Daniel's repeated intervention.

## 8. Recommendation

Approve G2 for Project promotion and controlled G3 planning. Do not begin implementation until G3 approves scope, schedule, cost, quality, resource, risk and evaluation baselines.

Benefits plan

# GAB Agent Context & Operating Intelligence
## G1 Benefits Management Plan

## 1. Benefits Principle

Benefits must be demonstrated through operating behavior, not inferred from completed code, installed plugins or model claims. Project completion will not equal benefit realization.

## 2. Expected Benefits

### B-01 Reduced CEO recontexting burden

**Beneficiary:** Daniel Nsowah  
**Measure:** median and distribution of recontexting turns before useful action.  
**Current baseline:** not instrumented; founder-confirmed recurring burden.  
**Target:** calibrate after baseline; material reduction required for any future solution to pass.

### B-02 Improved first-turn module and workflow orientation

**Beneficiary:** Daniel and agent workforce  
**Measure:** independently labelled first-turn routing accuracy across representative single- and multi-module scenarios.  
**Current baseline:** not instrumented.  
**Target:** set at G2 after baseline; must include cross-module cases, not only page-matched prompts.

### B-03 Fewer wrong-source and stale-context incidents

**Beneficiary:** GAB governance and operating modules  
**Measure:** wrong canonical source selections, stale memory/static-prompt conflicts and use of legacy records.  
**Current baseline:** nine confirmed current-state findings.  
**Target:** zero critical authority violations; remaining thresholds set at G2.

### B-04 Safer tool and governance routing

**Beneficiary:** GAB, clients and future operators  
**Measure:** wrong-module tool calls, governance-bypass attempts, required approval/HITL adherence and successful cross-module handoffs.  
**Current baseline:** tool calls are logged, but intended routing and bypass outcomes are not labelled.  
**Target:** no accepted workflow may bypass module ownership or approval controls.

### B-05 Faster time to useful action

**Beneficiary:** Daniel and future operating teams  
**Measure:** elapsed time and turns from request to first correct, evidence-backed action or decision-ready response.  
**Current baseline:** not instrumented.  
**Target:** establish after baseline without sacrificing evidence quality.

### B-06 Cross-agent operating consistency

**Beneficiary:** Daniel and agent workforce  
**Measure:** protocol/skill parity, repeated scenario outcomes, valid canonical pointers and intentional vendor exceptions.  
**Current baseline:** Codex 44 GAB skills, Claude 13, Kimi 13, repository 12, no discovered Gemini GAB custom path; known broken and differing files.  
**Target:** every formally supported agent passes the agreed protocol and scenario checks.

### B-07 Reuse of Baobab and Canopy institutional knowledge

**Beneficiary:** future initiatives and implementation teams  
**Measure:** new initiatives retrieve and cite current Atlas/contracts/approved packs instead of repeating enterprise discovery.  
**Current baseline:** Canopy exists but is not loaded or retrievable in the in-app runtime.  
**Target:** every relevant future initiative demonstrates traceable use of the canonical Atlas and current contracts.

## 3. Guardrails

Benefits are invalid if achieved by:

- bypassing HITL or module tools;
- hiding errors or corrections;
- replacing canonical sources with duplicated summaries;
- exposing confidential report or client content;
- narrowing tools so aggressively that valid cross-module work fails;
- prioritizing speed over evidence quality;
- embedding mutable company facts into unversioned model weights or prompts;
- excluding an agent surface without recording it as unsupported.

## 4. Measurement Plan

G2 should define a governed baseline sample using representative workflows:

1. Comms to Production to Approvals.
2. SiteSense/EIRAS to Data Trust, Consultancy, Projects and Reports.
3. Portfolio & BA to Projects and benefits review.
4. Farm/GIS evidence to operational and client-facing analysis.
5. Finance/Sales/Projects questions where source boundaries matter.
6. Abrupt topic changes from an unrelated dashboard page.

For each scenario, record requested intent, page context, retrieved sources, tools offered, tools used, corrections, governance outcome, turns and elapsed time. The same evidence set must be usable by an independent reviewer.

## 5. Ownership And Timing

- **Benefit owner:** Daniel Nsowah.
- **Governance owner:** Portfolio & BA through G2; Projects after promotion if authorized.
- **Operating evidence custodians:** System, Reports, Data Trust and relevant module owners.
- **Initial baseline:** required before any delivery baseline is approved.
- **Canopy-linked review:** 15 August 2026 benefits review should check whether this initiative is using Baobab/Canopy and whether baseline evidence exists.
- **Post-change reviews:** timing to be defined at G2 only if implementation is authorized.

## 6. Benefits Risks

- current pain may be real but infrequent;
- model/vendor changes may affect results independently of GAB changes;
- subjective correction counts may be inconsistently labelled;
- productivity gains may be offset by maintenance cost;
- a technically successful capability may not reduce Daniel's burden;
- benefit claims may be made before sufficient operating exposure.

## 7. G1 Benefits Conclusion

The benefits are strategically relevant and measurable in principle, but current telemetry is insufficient to establish baseline values. G2 must make baseline creation and independent evaluation prerequisites for implementation authorization.

Charter basis

# GAB Agent Context & Operating Intelligence
## G2 Charter Basis

**Initiative ID:** 28fb7f45-9141-4655-9854-5cf06e29499c
**Governance:** Full
**Gate under review:** G2 - Authorization
**Sponsor and decision authority:** Daniel Nsowah
**Prepared:** 13 July 2026
**Recommended delivery type:** Internal enabling backbone
**Current authorization:** Charter analysis only. No implementation or Project promotion has occurred.

## 1. G2 Decision Question

Should GAB authorize a controlled Project to establish a measurable, governed and vendor-neutral foundation for agent context integrity and operating intelligence, beginning with baseline evidence and direct repairs before any broader capability is expanded?

G2 does not approve a plugin, model training, fine-tuning, a GIS specialist, public productization or an unrestricted central context platform.

## 2. Purpose

Project Baobab established repository-backed enterprise knowledge. Project Canopy converted that knowledge into an accepted Enterprise Map and Architecture Atlas. G1 confirmed that GAB's remaining material need is not primarily missing knowledge: it is inconsistent activation of current, authoritative and appropriately scoped knowledge when agents begin work or move among modules and dimensions.

The proposed Project will make that operating foundation testable and usable across supported agent surfaces while preserving existing source owners, module contracts, human approvals and evidence traceability.

## 3. Objectives

1. Establish a repeatable baseline for context-switching and routing performance before changing behavior.
2. Correct the verified context-integrity defects that can be resolved locally.
3. Define and enforce one agent-neutral authority order for live records, contracts, approved evidence, Atlas indexes, memory and session history.
4. Provide governed retrieval of current contracts, approved Atlas evidence and approved report bodies.
5. Support multi-label module intent and explicit transitions without forcing work into one page category.
6. Reduce irrelevant tool and context exposure without removing necessary cross-module escalation.
7. Establish honest, testable activation and parity controls across supported agents.
8. Evaluate benefits, risks and residual gaps before transition or any later packaging decision.

## 4. Governing Principles

- Existing owning modules remain the source of truth.
- The capability may index and retrieve; it must not become a competing truth store.
- Live canonical records and current contracts outrank approved navigation indexes; approved evidence outranks memory, snapshots and handovers.
- Context is classified by authority, ownership, stability, freshness, confidentiality and access.
- Work can belong to multiple enterprise dimensions at once.
- Tool and context disclosure is progressive and reversible.
- Evidence and evaluation precede scale.
- HITL, Data Trust, action provenance and Lane A/Lane B governance remain binding.
- Vendor-specific limitations are recorded as exceptions rather than hidden behind false parity.
- Every conclusion and acceptance decision must be independently challengeable using recorded evidence.

## 5. Scope In

1. A representative evaluation corpus covering stable within-module work, abrupt module changes and multi-dimensional workflows.
2. Baseline instrumentation and measures for first-turn routing, recontexting turns, wrong-module tool calls, stale-context incidents, governance-bypass attempts, Daniel corrections and time to useful action.
3. Direct integrity repairs confirmed at G1, including newest-message history selection, page-context bias, canonical PM project context, memory authority wording, approved report-body retrieval, static-prompt drift and broken required-skill activation paths.
4. An agent-neutral context authority and source manifest referencing current module owners rather than copying their data.
5. Governed retrieval for current module contracts, approved Baobab/Canopy Atlas evidence and approved report bodies, with source identity, status and freshness.
6. Multi-label intent and transition handling for representative cross-module workflows.
7. Task-scoped progressive tool disclosure with an explicit escalation and fallback path.
8. A supported-agent matrix for the in-app Chief of Staff, Codex, Claude Code, Kimi and Gemini/Antigravity.
9. Pilot evaluation, quality controls, security checks, benefits measurement, transition and maintenance ownership.
10. Contract, MCP map, test, release and Drive/report updates required by each authorized implementation tranche.

## 6. Scope Out

- Replacing Supabase, module tables, Data Trust, Git, Drive or module contracts as source owners.
- Rewriting all modules or redesigning the entire GAB OS.
- Building or launching a public or paid product.
- Plugin packaging before the operating capability is proven.
- Training, fine-tuning or hosting a proprietary foundation model.
- Building a GIS-specialist model or combining GIS analysis capability with this enterprise context Project.
- Changing public claims, client reports or irreversible external actions without existing approvals.
- Guaranteeing identical vendor behavior where the vendor does not expose equivalent extension points.
- Migrating historical records merely to simplify the new capability.
- Creating an autonomous action layer that bypasses module tools or CEO approval.

## 7. Work Packages And Deliverables

### WP1 - Baseline And Evaluation Corpus

Deliverables:
- representative transition and multi-module scenario register;
- baseline instrumentation specification;
- pre-change benchmark and evidence pack;
- metric definitions, guardrails and known measurement limitations.

Exit condition: GAB can reproduce and measure the current friction without relying only on anecdote.

### WP2 - Direct Context-Integrity Repairs

Deliverables:
- corrected newest-message selection on server and user interface;
- page context treated as one routing signal rather than an overriding instruction;
- current project context sourced from canonical pm_projects;
- runtime authority wording reconciled with the Agent Operating Protocol;
- approved report-body retrieval corrected;
- duplicated mutable prompt facts removed or retrieved from canonical sources;
- broken required-skill pointers corrected and checked.

Exit condition: focused tests, contract updates and same-surface verification pass. Benefits are remeasured before proceeding.

### WP3 - Authority And Context Manifest

Deliverables:
- machine-readable enterprise/module map referencing Canopy;
- source-authority hierarchy;
- classification for stable, dynamic, confidential and vendor-specific context;
- ownership, freshness, access and fallback rules;
- contradiction and stale-source handling.

Exit condition: an independent reviewer can identify why a context item was selected and which source owns it.

### WP4 - Governed Knowledge Retrieval

Deliverables:
- governed retrieval path for current module contracts;
- governed retrieval path for approved Atlas/Enterprise Map evidence;
- approved report-pack body retrieval with status and provenance;
- authentication, authorization, freshness and citation metadata;
- retrieval tests for allowed and restricted content.

Exit condition: supported agents can retrieve the same approved knowledge through governed paths without a parallel content store.

### WP5 - Intent, Transition And Tool Scoping Pilot

Deliverables:
- multi-label intent and transition classifier/rules;
- transition context packet preserving active objective, relevant modules, evidence and governance path;
- progressive tool disclosure and explicit expansion fallback;
- representative Comms-Production, SiteSense-EIRAS/Data Trust/Consultancy/Projects, and Portfolio-Projects tests.

Exit condition: routing and context exposure improve against baseline without reducing governed workflow completion.

### WP6 - Cross-Agent Activation And Parity

Deliverables:
- supported-agent matrix;
- canonical activation manifest and installation/check procedure;
- parity validator for required protocols, skills, retrieval and tools;
- documented vendor exceptions and fallback behavior;
- agent onboarding and recovery test.

Exit condition: no supported agent silently operates with a missing or stale required protocol.

### WP7 - Pilot, Quality And Benefit Evaluation

Deliverables:
- controlled pilot results by agent and scenario;
- source-authority, security, governance and regression results;
- benefit comparison against WP1 baseline;
- defects, limitations and residual-risk register;
- recommendation to accept, revise, limit or stop.

Exit condition: evidence shows whether the capability produces material benefit without weakening governance or evidence quality.

### WP8 - Transition, Maintenance And Learning

Deliverables:
- operational owner and review cadence;
- manifest, contract, skill and evaluation maintenance procedure;
- monitoring and incident response;
- final accepted architecture and operating documentation;
- lessons learned and future-opportunity register;
- separate decision briefs, only if evidence supports plugin packaging, model specialization or GIS specialization.

Exit condition: accepted capability is maintainable in normal GAB OS operations and unproven opportunities remain explicitly separate.

## 8. Requirements Baseline

Validated and approved at G1:
- BR-001 reduce recontexting effort and make it measurable;
- BR-002 apply a consistent source-authority hierarchy;
- BR-003 preserve current intent and multi-dimensional classification;
- BR-004 provide governed contract, Atlas and report retrieval;
- BR-005 scope tools/context while preserving cross-module paths;
- BR-006 make required protocols and skills current across supported agents;
- BR-007 measure context and routing performance;
- BR-008 preserve approvals, provenance, Data Trust and module ownership.

G2 requirements proposed for Daniel's validation:
- BR-009 phased tranches and stop/go controls;
- BR-010 explicit supported-agent matrix;
- BR-011 source classification by authority, freshness, confidentiality and access;
- BR-012 same-change contracts, MCP parity, tests and live-surface verification.

## 9. Supported Agent Boundary

- **In-app Chief of Staff:** target full runtime integration, governed retrieval, transition handling and telemetry.
- **Codex:** target full repository/MCP activation, manifest use and parity verification.
- **Claude Code:** target required protocol/skill activation, governed retrieval and parity verification within its extension model.
- **Kimi:** target required protocol/skill activation, governed retrieval and parity verification within its extension model.
- **Gemini/Antigravity:** target documented direct-repository fallback and governed retrieval; local custom-skill parity is conditional on a supported vendor extension point.
- **Future agents:** not automatically supported. Entry requires a documented activation, authority, access and evaluation profile.

## 10. Stakeholders And Accountability

- Daniel Nsowah: sponsor, product owner, decision authority and principal operating beneficiary.
- Supported agent workforce: users and controlled execution surfaces.
- Module/source-truth custodians: owners of contracts, records, evidence and access boundaries.
- Future GAB staff/partners: downstream onboarding beneficiaries; not present implementation decision-makers.
- Projects module: delivery and control owner after G2 promotion.
- Portfolio & BA: business need, requirements, benefits and gate traceability.
- Data Trust and Reports: evidence health, approved outputs and retrieval boundaries.
- System and Approvals: release health, access control and irreversible-action governance.

## 11. Constraints

- GAB is a one-person company using multiple agents; delivery must respect limited CEO review capacity.
- Existing modules, tables and approved evidence remain authoritative.
- Security and authenticated access cannot be weakened to simplify retrieval.
- Current telemetry does not provide a quantified baseline; targets must be baselined before G3 approval or established as G3-controlled thresholds.
- Vendor extension mechanisms are unequal.
- Implementation changes require Lane B governance, contract/tool-map parity, focused tests and deployed verification.
- No committed external budget exists at G2.
- Public claims and client outputs remain outside this Project unless separately approved.

## 12. Assumptions To Test

- The representative scenario corpus reflects Daniel's real transition burden.
- Correcting local defects will produce a measurable initial improvement.
- A thin manifest plus governed retrieval can improve consistency without centralizing source ownership.
- Progressive tool disclosure can reduce noise without hiding required actions.
- Agent-neutral protocols can produce useful parity despite vendor differences.
- GAB can maintain the capability at a proportionate cost after transition.

## 13. Dependencies

- Project Baobab evidence and methodology.
- Project Canopy Enterprise Map, Atlas and maintenance model.
- Agent Operating Protocol and 26 Tier 2 module contracts.
- MCP bridge, orchestrator and tool parity.
- pm_projects and Projects controls after promotion.
- Reports approved pack storage and retrieval.
- Data Trust source/freshness and Drive mirror posture.
- Agent-local activation paths and vendor capabilities.
- Vercel, Supabase, Git and authenticated deployment/release controls.

## 14. Principal Risks And Controls

| Risk | Control |
|---|---|
| Competing source of truth | Thin reference/manifest architecture; owning modules remain authoritative |
| Central blast radius | Phased deployment, feature isolation, fallbacks and tranche stop/go reviews |
| Scope expansion into plugin/training/GIS | Explicit exclusions and separate future PBA decisions |
| Restricted information exposure | Authentication, source classification, least-access tests and HITL |
| Wrong routing through over-scoping | Multi-label classification, progressive expansion and cross-module scenarios |
| Tool removal blocks valid work | Explicit fallback to full/adjacent tool discovery with telemetry |
| Vendor coupling or false parity | Agent-neutral core plus documented vendor-specific adapters/exceptions |
| Metrics reward speed over correctness | Governance, evidence quality and source accuracy as non-negotiable guardrails |
| CEO review capacity becomes bottleneck | Bounded work packages, concise gate evidence and defined review points |
| Documentation and implementation drift | BR-012 same-change contract, MCP map, tests and live verification |
| Benefits remain anecdotal | WP1 baseline and WP7 repeatable comparison |
| Existing operations regress | Focused regression suite, deployment checks and rollback/fallback plan |

## 15. Cost And Resource Basis

G2 authorizes no spending. The expected cost is primarily internal engineering/agent time plus Daniel's review time. G3 must estimate effort by work package, identify any subscription or infrastructure cost, set resource assignments and establish cost controls. Any external vendor, model-hosting or material subscription commitment requires separate approval.

## 16. Delivery Governance

1. Daniel records G2 GO, REVISE, HOLD or KILL.
2. Only G2 GO allows promotion into the Projects module.
3. Promotion seeds G3-G5 lifecycle controls, requirements, benefits, stakeholders and risks.
4. G3 must approve the integrated delivery baseline before Lane B implementation starts.
5. Each work package has evidence, quality and stop/go acceptance.
6. Scope changes return through Project change control and, where material, Portfolio & BA.
7. G4 accepts only demonstrated capability and documented limitations.
8. G5 closes only after transition, lessons and benefits-review ownership are established.

## 17. Success And Acceptance

G2 deliberately does not invent numerical improvement targets before a baseline exists. G3 must set thresholds using WP1 or a controlled pre-implementation baseline.

Minimum acceptance dimensions:
- first-turn module and multi-module routing accuracy;
- recontexting turns and Daniel corrections;
- time to first useful, governed action;
- wrong-module tool calls and stale-context incidents;
- governance-bypass attempts;
- current-source retrieval accuracy and citation;
- relevant versus irrelevant tool/context exposure;
- cross-agent activation/parity coverage;
- regression, security and evidence-quality performance.

The Project is not successful if it becomes faster by hiding uncertainty, omitting evidence or bypassing governance.

## 18. Architectural Decisions Justified At G2

G2 may justify constraints and direction, not detailed implementation:
- choose a phased foundation over plugin-first or training-first;
- begin with baseline and direct repair;
- keep source ownership distributed;
- use an agent-neutral manifest and governed retrieval;
- preserve multi-dimensional classification;
- use progressive disclosure with fallback;
- defer plugin, model and GIS specialization decisions.

Detailed component design, schemas, interfaces, sequence, effort and release controls belong in G3 planning and controlled implementation.

## 19. G2 Recommendation

**GO to promotion and G3 planning, subject to Daniel's approval of BR-009 through BR-012 and this charter boundary.**

The recommended option addresses all validated requirements, absorbs the minimum targeted-repair option as its first tranche, preserves stop points and leaves unproven future opportunities outside scope.

G2 GO would authorize Project establishment and planning. It would not authorize implementation before G3.

## 20. Standing Gate Questions

- **Did G2 achieve its objectives?** Yes. Purpose, scope, requirements, options, deliverables, supported agents, dependencies, risks, governance and acceptance approach are defined.
- **Is the evidence sufficient?** Sufficient to authorize a bounded Project and G3 planning; not sufficient to approve detailed architecture, investment targets or implementation.
- **Are architectural decisions justified?** The direction and constraints follow confirmed G1 evidence and current contracts. Detailed design remains open.
- **Are there unresolved risks?** Yes. Baseline magnitude, vendor constraints, security, central blast radius, tool-scoping failure and maintenance cost require Project controls.
- **Is the initiative ready to progress?** Ready for Daniel's G2 decision. If the charter boundary or proposed requirements are not accepted, hold or revise at G2.

RecommendationNext action

Productization and commercial readiness

TypeData readinessLaunch readiness

Public URL

Target usersPaying customers

Free vs paid boundaryPricing hypothesisService delivery model

Success metricsGo / no-go criteria

Productization plan

# Productization Position

The authorized subject is an **internal enabling backbone**, not a public product or paid service.

## Current Position

- Primary users: Daniel and explicitly supported GAB agents.
- Paying customers: none.
- Public URL: none.
- Pricing: not applicable at this gate.
- Data readiness: partial; canonical sources exist, but retrieval, classification and telemetry are incomplete.
- Launch readiness: planning only.
- Service model: repository-governed internal capability operated through existing GAB OS modules and agent surfaces.

## Future Packaging Rule

A plugin may later be useful as a distribution and activation package, but only after the underlying capability:
- is demonstrated through controlled evaluation;
- has a stable authority and access model;
- has clear vendor support boundaries;
- can be maintained without duplicating source truth;
- has passed security and governance review.

Plugin packaging, commercial licensing, a trained GAB model and GIS specialization require separate business cases or explicit later authorization.

2ef4122fConvertedStandardG6Internal PlatformLaunch: LiveData: Partial

Project Sankofa — Aframso Enclave Intelligence Programme

Two linked deliverables: (A) an integrated, desktop-only interdisciplinary assessment of the wider Aframso enclave (climate baseline + downscaled futures, land/soil/water at best available resolution, crop suitability now vs 2040+, plus a desk-researched human layer: local economy, tenure, social/cultural cropping behaviour) producing evidence-based recommendations for GAB's own crop focus and a community resilience engagement programme; and (B) a permanent GIS Intelligence Engine upgrade in GAB OS (time-series and projection data, curated enclave-resolution layers, repeatable referenced pipelines, decision-grade dashboard output, fixing documented Farm/GIS module defects). Principle: build the factory by making the first product. CEO decisions at intake: wider-enclave boundary with farm/block drill-down; cassava primacy is a hypothesis the Study must scientifically validate, not dogma; purely desktop method (no soil lab or fieldwork), confidence levels must reflect desktop provenance. Concept note: docs/proposals/2026-07-12_aframso-enclave-intelligence-concept.md

Owner: Daniel NsowahCategory: FarmPriority score: Not scoredLatest gate: G0 go Drive doc

Open 6-domain workspace →Open project

Analysis record

StatusStage gateGovernanceDecision

PriorityTailoringFitValueRisk

Needs assessment

AUDIT FINDINGS (verified 2026-07-12, evidence inline):

1. SPATIAL BASE PACK IS REAL AND SUBSTANTIAL: ~11.5 GB on disk at data-trust/07-spatial-ghana-base across 18 categories (terrain 8.1G, landcover 1.0G, forest 785M, soils 704M, population 365M, infrastructure 301M, climate 238M, plus boundaries/hydro/wetlands/protected/hazard/governance/constraints/manifests). Built for Ghana-wide EIRAS screening resolution. Housekeeping drift noted: empty duplicate hazards/ dir beside hazard/.

2. GIS AGENT IS ALIVE AND PRODUCING DECISION-GRADE OUTPUT: gis-analysis agent runs every 3 days; verified readings 2026-06-28 through 2026-07-10 contain per-run weather (Open-Meteo), soil moisture at 3 depths, ET0, VPD, and per-block NDVI (avg/min/max across 18 active blocks) with specific AI recommendations and alert levels. The 'not properly built' assumption is PARTLY WRONG: production of intelligence works.

3. THE REAL GAP IS UTILISATION, NOT PRODUCTION: the same critical finding ('35 fallow blocks missed the April-May 2026 planting window, plant now while conditions favor establishment') has repeated in readings since at least 2026-06-28 with the same recommended actions and no recorded closure. Intelligence is produced but does not reliably become decisions/tasks/actions. This is an organisational loop problem, in scope for this programme (see BR-005).

4. PROVENANCE GAP: readings are labelled source=open-meteo, but Open-Meteo does not supply NDVI; the actual NDVI source must be traced and documented in Phase 0 before any scientific claim rests on it.

5. COVERAGE GAPS: readings cover 18 active blocks, not the full 53-block estate, not the other landbank holdings (Bobin 501ac, Kandaha 32ac, Tepa 12ac), and not the wider enclave. Climate data is current-conditions only: no 30+ year historical series in use, no seasonal forecasts, no downscaled projections. No crop suitability modelling exists. The human layer (enclave economy, tenure, social/cultural patterns) is entirely absent from the OS.

6. DOCUMENTED DEFECTS CONFIRMED (from the GIS module contract itself): dashboard analysis renders globally instead of clipped to farm boundary; UI not decision-grade; outputs not consistently reflected into Briefing/Reports.

NEED STATEMENT: GAB cannot currently answer, with referenced evidence, (a) what Aframso's climate is doing over 30 years and where it is heading to 2050, (b) whether cassava primacy survives that trajectory and what alternatives fit block-by-block, or (c) what the enclave's human/economic reality means for adaptation. The capability to answer these desktop-only is buildable on the verified existing foundation.

Business case

OPTIONS CONSIDERED:
A. Fix defects only (clipping/UI) - cheap but answers none of the climate/crop/human questions.
B. Commission an external consultant study - one-off report, no standing capability, external spend, contradicts GAB's AIOPS identity (we sell exactly this kind of capability).
C. In-house desktop programme that builds permanent capability by delivering the Study (RECOMMENDED) - honours the CEO's desktop-only constraint, compounds into consultancy/EIRAS capability.
D. Option C plus EOSDA paid data augmentation - decide via the open EOSDA elicitation before G2; EOSDA is the main way to buy data depth without fieldwork.

COSTS (Option C): agent/CEO time; compute/storage (base pack already ~11.5GB local; new time-series data adds volume); possibly EOSDA subscription (Option D); zero fieldwork spend by CEO decision.

BENEFITS: (1) de-risked crop and investment decisions across the ~1,092-acre landbank, with cassava primacy scientifically tested rather than assumed; (2) permanent GIS intelligence capability that serves consultancy (billable), EIRAS/SiteSense (shared spatial engine), and carbon work; (3) community resilience positioning: GAB as the credible local interpreter of climate reality at Aframso; (4) closes the intelligence-to-action loop so existing agent output stops going unactioned; (5) evidence backbone (Data Trust registered layers) for all future public/client claims.

RISKS: desktop-only credibility ceiling (mitigate: explicit confidence levels on every finding, iSDAsoil limitations stated, possible academic partner per open elicitation); NDVI provenance currently untraced (Phase 0 blocker); local compute/storage constraints for time-series rasters; scope creep toward fieldwork (guard: G-gated, CEO decision); utilisation problem is organisational, not technical, so tooling alone will not fix it (guard: BR-005 makes the decision loop a first-class requirement).

Benefits plan

BENEFITS PLAN

B1 EVIDENCE-BASED FARM DECISIONS: any block decision (plant/fallow/switch/invest) answerable with referenced evidence in minutes from the OS. Measure: decision-support queries answerable per block; baseline today ~0 (NDVI estimated, no climate futures). Owner: CEO/Farm. Timing: from P1 onward, full at P3.

B2 CASSAVA HYPOTHESIS SCIENTIFICALLY TESTED: sourced, confidence-rated statement of whether cassava primacy survives the 2050 climate trajectory, and what alternatives rank where. Measure: published suitability model + rankings. Timing: P3.

B3 HONEST DATA FOUNDATION: zero conflation of estimated vs measured data in briefings; ndvi_source visible everywhere; every layer Data Trust-registered with provenance/freshness. Measure: high alert aframso-programme:ndvi-provenance resolved; Data Trust coverage of programme layers 100%. Timing: P0-P1.

B4 INTELLIGENCE-TO-ACTION LOOP CLOSED: no material agent finding older than one cycle without an owner. Measure: repeat-unactioned findings count (baseline: planting-window finding repeated unactioned since 2026-06-28). Timing: P0 design, enforced from P1.

B5 CAPABILITY SPILLOVER: same engine serves consultancy (billable GIS), EIRAS/SiteSense (shared spatial base), carbon work. Measure: first consultancy/EIRAS use of a Sankofa-built pipeline. Timing: opportunistic from P1.

B6 COMMUNITY POSITIONING: GAB as credible local interpreter of climate reality in Sekyere Central; engagement outputs produced and delivered. Measure: engagement materials produced; community sessions held (delivery mode at CEO discretion). Timing: P4.

Benefits review at G5 closure per PMI pattern.

Charter basis

PROJECT SANKOFA CHARTER (basis for G2)

PURPOSE: Scientifically understand the Sekyere Central enclave (climate past/present/future, land, people) to evidence-base GAB's crop and investment decisions and prepare the community for climate change; build permanent GIS intelligence capability in GAB OS by delivering that Study. Principle: build the factory by making the first product.

SCOPE IN: Sekyere Central District (GH0239) boundary with drill-down to farm/blocks; 30+yr climate baseline + downscaled projections to 2050 (2+ scenarios); desktop soil/terrain/water layers; crop suitability modelling testing the cassava-primacy hypothesis vs 4+ alternatives; desk-researched human layer (census/GSS/market/literature); intelligence-to-action loop closure (BR-005); NDVI honest-sourcing fix (EOSDA free tier for farm + in-house Sentinel-2 pipeline for enclave; retire GEE dependency); Data Trust registration of every layer; community engagement outputs (desk-produced).

SCOPE OUT (WONT, CEO 2026-07-12): all fieldwork - soil sampling, drone flights, ground truthing, enumerator surveys. Formal institutional partnerships (CSIR-CRI/Ghana Met deferred to outreach campaign letters 9-10 at CEO discretion).

PHASES: P0 foundation (defect fixes, NDVI source swap, boundary polygon extraction, layer QA) -> P1 climate baseline + futures -> P2 human layer desk research -> P3 integration + crop scenarios -> P4 recommendations + community engagement outputs. Capability hardening continuous.

DELIVERABLES: enclave climate story report; per-block crop suitability model + rankings under baseline and 2050 scenarios; socio-economic enclave profile; integrated recommendations (GAB strategy + community resilience); upgraded GIS module (clipped, decision-grade, honest sourcing); all layers Data Trust-registered.

GOVERNANCE: Daniel Nsowah is gate authority (G3-G5 after promotion). Delivery via Projects module with work packages and output packs per the Baobab/Canopy pattern. Module contracts (farm.md, gis.md) updated with every behavior change.

CONSTRAINTS: desktop-only; every finding carries explicit confidence level reflecting provenance (estimated vs modelled vs measured); no scientific claim on untraced data.

KEY RISKS: desktop credibility ceiling; local compute for time-series rasters; organisational non-adoption (mitigated by BR-005 as MUST); GEE licensing (mitigated by retiring dependency).

RecommendationNext action

Productization and commercial readiness

TypeData readinessLaunch readiness

Public URL

Target usersPaying customers

Free vs paid boundaryPricing hypothesisService delivery model

Success metricsGo / no-go criteria

Productization plan

226546b8ConvertedFullG3Enabling BackboneLaunch: PlanningData: Partial

Project Canopy — GAB Enterprise Map & Architecture Atlas

Follow-on initiative from Project Baobab. Baobab established that GAB has a rich but distributed enterprise architecture across modules, contracts, MCP tools, source-of-truth systems, products, service lines, evidence records, governance workflows, and agent operating rules. The next highest-value initiative is to turn that discovered knowledge into a maintained Enterprise Map / Architecture Atlas so humans and agents can navigate GAB consistently before any plugin, restructuring, or implementation work is considered. Problem/opportunity: GAB's architecture is now evidenced, but it is not yet packaged as one canonical operating map. Without that map, future plugin packaging, product architecture, onboarding, agent reasoning, and governance decisions risk encoding partial or implicit architecture. Boundary: this initiative is not a plugin build, code change, module rewrite, product renaming, public positioning exercise, or restructuring project. It begins as Portfolio & BA analysis and must pass G0-G2 before any Project delivery work.

Owner: Daniel NsowahCategory: Enterprise Architecture / GAB OSPriority score: 86Latest gate: G2 go Drive doc

Open 6-domain workspace →Open project

Analysis record

StatusStage gateGovernanceDecision

PriorityTailoringFitValueRisk

Needs assessment

# Project Canopy G1 Needs Assessment

## 1. Assessment Question

Why is Project Canopy necessary now, after Project Baobab?

G1 is not a design phase. It does not decide the final Atlas format, build a plugin, change code, restructure modules, rename products, or create public positioning. Its purpose is to determine whether there is sufficient evidence that GAB needs a canonical Enterprise Map / Architecture Atlas.

## 2. Evidence Basis

Primary evidence comes from Project Baobab, now closed at G5:

- Baobab project status: completed, gate state closed, progress 100%.
- Baobab WP5 Gate Review Pack and Lessons v1.0, approved. Pack ID: c0eb0ba3-73b1-45e2-91cc-21bba8956193.
- Baobab WP4 Findings Synthesis Closeout v1.0, approved. Pack ID: f68742c3-3653-4b0d-a041-3b4c45a05fd3.
- Baobab WP1-WP3 evidence, classification, and contract/module review packs.
- Project Canopy G0 approval by Daniel Nsowah.
- Project Canopy initial requirements BR-001, BR-002, and BR-003.

## 3. Current State Observations

### Observation 1: The enterprise architecture is evidenced but distributed.

Baobab demonstrated that GAB architecture exists across module contracts, source-of-truth boundaries, MCP tools, Reports packs, Projects/Portfolio governance, EIRAS/SiteSense product surfaces, Data Trust, Comms, Production, Approvals, and operating protocols. It is not yet packaged as one canonical navigation artifact.

### Observation 2: GAB entities are multi-dimensional.

Baobab showed that entities such as EIRAS/SiteSense cannot be safely classified in one category only. They can simultaneously be products, modules, public tools, data workflows, service capabilities, governance surfaces, and possible future plugin components.

### Observation 3: The OS already governs work through contracts and gate records.

Baobab itself followed the Portfolio-to-Projects lifecycle, produced Reports packs, recorded gate decisions, preserved constraints, and closed through G5. This confirms the governance machinery can support Canopy if the business need is justified.

### Observation 4: The plugin idea remains a future opportunity, not a next action.

Baobab established that a plugin may eventually be valuable, but it also established that packaging knowledge before the canonical map exists risks encoding partial architecture.

## 4. Business Need

GAB needs Project Canopy because Baobab turned implicit enterprise architecture into evidence, but the evidence still needs a maintained navigational structure. Without the Atlas, humans and agents must continue reconstructing the enterprise from contracts, reports, tool maps, project records, and repository evidence each time a strategic question arises.

The business need is therefore:

Create a canonical Enterprise Map / Architecture Atlas that lets GAB humans and agents understand what GAB is, how it works, which systems own which facts, how modules and products relate, and where future initiatives should begin.

## 5. Root Cause / Underlying Cause

The root cause is not failure or disorder. It is successful organic growth. GAB OS matured through modules, contracts, tools, products, service lines, governance records, and evidence practices faster than a single enterprise navigation layer could emerge. Baobab made that growth visible; Canopy is needed to make it navigable.

## 6. Business Implications If No Action Is Taken

- Agent reasoning may remain dependent on long contract reads and prior chat context.
- Future plugin work may package incomplete or stale enterprise assumptions.
- Product/service boundaries, especially around EIRAS/SiteSense, may remain hard to explain consistently.
- New initiatives may repeat Baobab-style discovery instead of starting from a maintained map.
- Source-of-truth ownership may remain correct in contracts but difficult to navigate quickly.
- Enterprise knowledge may remain durable but scattered.

## 7. Evidence Against The Initiative

There is evidence that GAB already has strong architecture discipline: module contracts are mature, MCP tools are mapped, Portfolio and Projects gates work, Reports packs preserve outputs, and Baobab closed cleanly.

However, that evidence argues against panic or redesign, not against Canopy. It shows that Canopy should organize and navigate the existing architecture rather than replace it.

## 8. Scope Boundary

Canopy G1 does not authorize:

- plugin build;
- code changes;
- database/schema changes;
- module rewrites;
- public positioning;
- product renaming;
- folder restructuring;
- constitutional principle promotion;
- delivery execution in Projects.

## 9. G1 Recommendation

Recommendation: proceed to G2 charter-basis analysis for Project Canopy.

The evidence supports a genuine enterprise-level need for a canonical Enterprise Map / Architecture Atlas. G2 should define scope, deliverables, evidence standards, governance controls, and promotion criteria. It should not authorize implementation until Daniel approves G2 and promotion into Projects.

Business case

# Project Canopy Initial Business Case

## Strategic Rationale

Project Canopy converts Project Baobab's discovered enterprise knowledge into a navigable operating artifact. This directly supports GAB's one-man-company multi-agent operating model by giving humans and agents a shared map of modules, products, governance surfaces, source-of-truth systems, evidence records, and operating pathways.

## Value Hypothesis

The highest value is reduced enterprise reasoning friction. Canopy should reduce repeated rediscovery, improve agent handoffs, strengthen product/service clarity, protect governance boundaries, and create the foundation required before any plugin or agent enablement package is considered.

## Alternatives Considered

1. Do nothing: preserves current contracts and Baobab packs but leaves enterprise navigation distributed.
2. Build plugin now: premature because the canonical map of what the plugin should package is not yet formalized.
3. Rewrite module contracts: unnecessary and risky; Baobab found the contracts are a strength.
4. Create Enterprise Map / Architecture Atlas: best aligned with evidence because it organizes existing architecture without changing it.

## Preferred Option

Proceed to G2 charter-basis analysis for an Enterprise Map / Architecture Atlas.

Benefits planCharter basis

# Project Canopy G2 Charter Basis

## 1. Purpose

Project Canopy exists to create a canonical Enterprise Map / Architecture Atlas for GAB. It is the governed follow-on to Project Baobab. Baobab discovered and evidenced the enterprise architecture; Canopy will define the scope and delivery basis for making that architecture navigable.

This G2 charter basis authorizes only structured analysis and potential promotion into Projects if Daniel approves G2. It does not authorize implementation until G2 is explicitly approved and the initiative is promoted into Projects.

## 2. Business Need

GAB has a mature but distributed enterprise architecture across module contracts, MCP tools, source-of-truth systems, governance gates, Reports packs, Data Trust, EIRAS/SiteSense, Comms, Production, Approvals, Portfolio, and Projects. Baobab established this through evidence. Canopy is needed because GAB now requires a maintained navigation layer that humans and agents can use without rediscovering the enterprise each time.

## 3. Scope

In scope for the Project phase, if G2 is approved:

- Enterprise entity map: company, divisions, modules, products, service lines, systems.
- Capability map: what GAB can do operationally, commercially, and institutionally.
- Source-of-truth map: which module or system owns which facts.
- Governance map: Portfolio, Projects, Approvals, Reports, Data Trust, Comms, Production, and related gates.
- Product/service map: especially EIRAS, SiteSense, public tools, consultancy, and reports.
- Agent operating map: contracts, MCP tools, skills, hooks, Lane A/Lane B boundaries, and module handoffs.
- Evidence traceability model: every major Atlas claim traceable to Baobab evidence, current repository evidence, or a recorded Daniel decision.
- Maintenance model: owner, update triggers, review cadence, and relationship to module contracts.

## 4. Out Of Scope

G2 does not authorize:

- plugin build;
- code changes;
- database or schema changes;
- module rewrites;
- public positioning;
- product or service renaming;
- folder restructuring;
- publishing external claims;
- constitutional principle promotion;
- replacing module contracts, Data Trust, Reports, Portfolio, Projects, or other source-of-truth systems.

## 5. Deliverables If Promoted To Projects

### Deliverable 1: Enterprise Map / Architecture Atlas v1
A structured map showing GAB's enterprise dimensions and relationships.

### Deliverable 2: Source-of-Truth and Governance Atlas
A map of ownership boundaries, gate paths, approval paths, Reports outputs, Data Trust roles, and cross-module handoffs.

### Deliverable 3: Product-Service-Module Relationship Map
A specific map of how products, services, modules, public tools, data workflows, and governance surfaces relate, with EIRAS/SiteSense treated as multi-dimensional entities.

### Deliverable 4: Agent Navigation Guide
A practical map for agents: which contract to read, which MCP tool to use, which module owns the work, which guardrails apply, and when Lane B is allowed.

### Deliverable 5: Atlas Maintenance Model
A lightweight governance model for keeping the Atlas current without duplicating or replacing module contracts.

### Deliverable 6: Plugin Feasibility Input
A bounded input pack for future plugin feasibility. This deliverable identifies what a plugin might package, but does not build or authorize a plugin.

## 6. Evidence Standard

Every Atlas claim must be classified as one of:

- Established knowledge: supported by Baobab or current repository/contract/tool evidence.
- Open question: requiring further investigation or Daniel clarification.
- Future opportunity: requiring strategic decision.
- Recommendation: proposed next action, not authorization.
- Decision: approved by Daniel at the appropriate gate.

Founder clarification may resolve ambiguity or set future intent, but must not silently replace repository evidence.

## 7. Governance

- Portfolio & BA owns G0-G2.
- Projects owns delivery only after G2 approval and promotion.
- Daniel remains gate authority.
- Reports should hold substantial review outputs.
- No public-facing claims leave the OS without Comms, Production, Reports, Data Trust, and Approvals discipline where applicable.

## 8. Initial Work Packages For Project Planning

If promoted, the Project phase should include:

1. WP1: Atlas Evidence Refresh and Baobab Trace Review.
2. WP2: Enterprise Entity and Capability Map.
3. WP3: Source-of-Truth and Governance Map.
4. WP4: Product-Service-Module Relationship Map.
5. WP5: Agent Navigation and Operating Map.
6. WP6: Maintenance Model, Risks, and Final Review Pack.

These are planning candidates only. Final WBS belongs to G3 in Projects.

## 9. Risks

- Atlas duplicates contracts without adding navigation value.
- Plugin/build scope creep.
- Oversimplifying multi-dimensional entities.
- Promoting Baobab evidence candidates into principles too early.
- Atlas becomes too large for fast agent use.
- Future agents treat the Atlas as replacing module contracts instead of guiding navigation to them.

## 10. Success Measures

- Daniel accepts the Atlas scope and delivery boundary.
- Every major map claim is evidence-traceable.
- GAB humans and agents can identify the correct module, contract, tool, source-of-truth, and governance path faster.
- Plugin feasibility can be assessed from a canonical map rather than reconstructed enterprise memory.
- Future enterprise initiatives begin from Canopy/Baobab foundation rather than rediscovery.

## 11. G2 Recommendation

Recommendation: approve G2 and promote Project Canopy into Projects for controlled delivery of the Enterprise Map / Architecture Atlas.

This recommendation does not authorize plugin build, code/schema change, restructuring, product renaming, public positioning, or external publication. Those remain future decisions requiring separate approval.

RecommendationNext action

Productization and commercial readiness

TypeData readinessLaunch readiness

Public URL

Target usersPaying customers

Free vs paid boundaryPricing hypothesisService delivery model

Success metricsGo / no-go criteria

Productization plan

d5b8c3a8ConvertedFullG3Enabling BackboneLaunch: PlanningData: Partial

The Ghana Systems Project — Strategic Research Programme

A long-horizon, interdisciplinary GAB research programme whose permanent purpose is to build an evidence-based, integrated and continuously revisable understanding of Ghana as a living national system so that GAB and other legitimate users can make more meaningful decisions. Its governed methodology—Evidence, Knowledge, optional Models, decision support and Living Corpus—is an enabling capability, not the programme's replacement purpose. Real-world strategic questions may enter through approved Bounded System Inquiries. The earlier private-sector and 30-year macroeconomic work remains the first empirical anchor. A reusable Country Systems Methodology is only a possible future derivative after demonstrated transfer beyond Ghana.

Owner: Daniel NsowahCategory: Strategic Research / Market Intelligence / Knowledge ArchitecturePriority score: Not scoredLatest gate: G0 go Drive doc

Open 6-domain workspace →Open project

Analysis record

StatusStage gateGovernanceDecision

PriorityTailoringFitValueRisk

Needs assessmentBusiness caseBenefits planCharter basisRecommendationNext action

Productization and commercial readiness

TypeData readinessLaunch readiness

Public URL

Target usersPaying customers

Free vs paid boundaryPricing hypothesisService delivery model

Success metrics

Phase 0: approved research charter, ontology, metadata/citation standard, confidence framework, hypothesis register and review protocol. Phase 1: a source-traceable Ghana Private-Sector Atlas and household-demand/trade synthesis with explicit uncertainty. Programme: 100% of material published claims linked to registered evidence; every major conclusion records discipline coverage, assumptions, confidence and revision history; regular evidence/benefits reviews; zero unsupported group-psychology or deterministic historical claims. Phase 0 additionally requires an approved Epistemology Framework, explicit Evidence–Knowledge–Models transition rules, assigned contributor/reviewer roles and a functioning domain-admission gate; zero unlabelled transitions from assumption or interpretation to fact. Phase 0 success is capability-based: consistent evidence classification, complete provenance, safe coexistence of competing interpretations, calibrated confidence, rigorous interdisciplinary integration, auditable layer transitions and independent reproducibility. Target maturity is M2—Pilot Demonstrated. Phase 0 additionally requires CT-08: the controlled pilot must demonstrate governed BSI intake, approved scope/exclusions, change control, exit assessment and Knowledge Base integration. Models remain optional and must be justified.

Go / no-go criteria

GO to charter only if the programme has a bounded Phase 1 question, evidence architecture, named review roles, sustainable cadence, credible source access and explicit anti-stereotyping/anti-determinism safeguards. HOLD if scope remains 'understand everything', required primary data is unavailable, review capacity is absent, or outputs cannot be maintained. STOP or redesign any modelling claim that cannot be tested, reproduced or bounded by uncertainty. No domain enters active scope unless it passes materiality, integration, evidence, review-capacity and bounded-deliverable tests. Phase 0 GO at G2 requires baselined capability-test thresholds, a controlled pilot corpus, named reviewers, resource capacity and failure/re-test rules. Phase 0 exit requires M2. No broader programme phase begins merely because Phase 0 documents were completed. No BSI begins without an approved specification and admission decision. A model or public derivative is not required for inquiry success and cannot be produced without its own evidence and review basis.

Productization plan

9959ecdbConvertedFullG3Enabling BackboneLaunch: PlanningData: Ready

Project Baobab — GAB Enterprise Architecture & Institutional Map

Founder-directed initiative to discover, classify, document, and govern the enterprise architecture that already exists across GAB Climate Smart Investments Ltd, GAB OS, module contracts, products, service lines, source-of-truth systems, evidence workflows, HITL governance, and future agent/plugin enablement. Current instruction: reveal the architecture exactly as it exists before proposing redesigns or building a plugin.

Owner: Daniel NsowahCategory: Enterprise Architecture / GAB OSPriority score: Not scoredLatest gate: G0 go Drive doc

Open 6-domain workspace →Open project

Analysis record

StatusStage gateGovernanceDecision

PriorityTailoringFitValueRisk

Needs assessment

# Project Baobab — G1 Forensic Enterprise Assessment

**Portfolio stage:** G1 Needs Assessment  
**Initiative:** Project Baobab — GAB Enterprise Architecture & Institutional Map  
**Decision question:** Is there sufficient evidence that Project Baobab is necessary now?  
**Working rule:** This assessment does not propose solutions, redesign architecture, build a plugin, or recommend implementation scope. It tests whether the evidence justifies progressing to deeper Portfolio & BA analysis.

---

## 1. Executive Summary

Project Baobab was prompted by an observed mismatch between the maturity of GAB OS as implemented and the degree to which GAB's enterprise architecture is explicitly documented as a coherent institutional model.

GAB OS now contains a substantial operating architecture: module contracts, a Contract Standard, an MCP Tool Map, Lane A / Lane B operating rules, Human-in-the-Loop approval governance, Data Trust, Portfolio & BA, Projects, Comms, Production, Reports, Drive mirroring, agent wake-up/preflight protocols, EIRAS/SiteSense public tools, and multiple source-of-truth systems.

The evidence does not show a failing system. It shows a maturing institution whose implementation has grown faster than its enterprise map. The immediate business question is therefore not "what should we build?" It is:

> Why is an explicit enterprise architecture discovery initiative necessary now?

The current evidence supports a G1 needs assessment because GAB has reached a point where implicit architecture creates business risk: agent reasoning can diverge, product/service boundaries can blur, public claims become harder to govern, future plugin packaging could encode incomplete assumptions, and institutional knowledge remains partly embedded in code/contracts rather than visible as an agreed enterprise map.

At G1, the recommended decision is not to build a plugin, rewrite contracts, restructure modules, or publish a philosophy. The only G1 recommendation is to proceed to structured enterprise architecture analysis if Daniel accepts that the evidence is sufficient.

---

## 2. Current State — Evidence Only

### 2.1 Enterprise Folder Structure

The company folder is organized into a numbered operating structure:

- 01_Farm
- 02_Agribusiness
- 03_Consultancy
- 04_Finance_and_Legal
- 05_Comms_and_Brand
- 06_GAB_OS_Agent_Outputs

Evidence source: local repository folder structure under /Users/danielnsowah/Desktop/GAB Climate Smart Investments Ltd and docs/google_drive_structure.md.

### 2.2 Core Systems

The repository and system architecture documents identify multiple systems:

- GAB OS — central executive/company operating system.
- GAB Farms ERP — farm operations source of truth.
- Blue Carbon MRV — blue-carbon / MRV platform.
- GAB Ramsar Map — wetland/GIS visualization layer.
- GAB SiteSense Lite — standalone public intake writing into GAB OS EIRAS records.
- GAB Website — public company/resource surface.
- GAB MCP Server — agent/tool bridge.
- Voice Relay / Voice Agent — conversational/front-facing company interface.

Evidence sources: docs/02_SYSTEM_ARCHITECTURE.md, repository package locations, and README files for gab-farms-erp, blue-carbon-mrv, gab-sitesense-lite, gab-website, gab-mcp-server, gab-voice-relay, and local-os.

### 2.3 GAB OS Module System

The GAB OS module contract folder contains 26 module contracts plus the shared standard and tool map. The contracts classify operating domains including:

- Executive control: Overview, Briefing, Chief of Staff, Alerts, Tasks.
- Governance and safety: Approvals, Audit Trail, System, Settings.
- Delivery lifecycle: Portfolio & Business Analysis, Projects, Reports.
- Commercial operations: Sales, Market Intel, Finance, Consultancy.
- Operating assets: Farm, GIS, Land, Assets, Staff & HR.
- Public voice: Comms, Production, Voice Agent.
- Evidence/intelligence: SiteSense/EIRAS, Data Trust.

Evidence sources: docs/modules/README.md, docs/modules/CONTRACT_STANDARD.md, individual module contracts.

### 2.4 Portfolio & BA Governance

Portfolio & BA is defined as the pre-project governance lane for serious initiatives. It owns Defining & Aligning, needs assessment, business case, benefits thinking, gate decisions, stakeholder engagement, elicitation, requirements, and promotion into Projects only after G2.

Evidence source: docs/modules/portfolio-business-analysis.md.

### 2.5 Project Delivery System

Projects is the canonical implementation and delivery-control tracker for structured work. It owns milestones, tasks, updates, baselines, scope, schedule, cost, risks, issues, stakeholders, quality checks, benefits, gates, operational transitions, and output packs.

Evidence source: docs/modules/projects.md and docs/projects/README.md.

### 2.6 Governance, Approvals, And HITL

GAB OS implements a Human-in-the-Loop model. Agents may analyze, draft, query, classify, and prepare internal outputs, but external, irreversible, public, financial, or commercial actions require Daniel's approval through governed approval paths.

Evidence sources: docs/AGENT_OPERATING_PROTOCOL.md, docs/03_AGENT_OPERATING_MODEL.md, docs/modules/approvals.md, docs/modules/comms.md, docs/modules/production.md.

### 2.7 Data Trust And Evidence Governance

Data Trust is defined as the watchtower for data quality, freshness, verification, and source trust. It reads source records rather than replacing them. EIRAS/SiteSense depends on source documents, spatial layers, assumptions, legal instruments, evidence records, source citations, spatial constraints, and related tables.

Evidence sources: docs/modules/data-trust.md, docs/modules/sitesense.md, lib/dataTrust.ts, EIRAS routes and library folders.

### 2.8 MCP And Agent Operating Layer

The repository contains a live MCP bridge and a tool map. Wake-up and preflight commands exist:

- npm run gab:wakeup
- npm run gab:preflight -- --module <module> --intent <intent>

The MCP Tool Map lists 181 tools, and the operating protocol requires agents to read contracts and use Lane A tools where available.

Evidence sources: docs/modules/MCP_TOOL_MAP.md, docs/AGENT_OPERATING_PROTOCOL.md, docs/AGENT_WAKEUP_PROTOCOL.md, scripts/gab-wakeup.mjs, scripts/gab-preflight.mjs, scripts/mcp-bridge-client.mjs.

### 2.9 Public Tools And Product Surfaces

EIRAS public tools include SiteSense Lite, FloodSense Lite, Regulatory Lite, ResourceMap Lite, ConsentTrack Lite, ReportBridge Lite, Green Carbon Lite, BioMainstream Lite, Efficiency Lite, Wetland/Flood, and Ramsar-related surfaces. These operate as preliminary public decision aids, lead magnets, and internal conversion paths.

Evidence sources: docs/modules/sitesense.md, docs/projects/AIOPS.md, docs/projects/EIRAS-lite-use-cases-and-external-test-plan.md, app/*-lite, app/api/eiras/*, standalone 03_Consultancy/gab-sitesense-lite.

---

## 3. Evidence Of Maturity

The current architecture demonstrates meaningful maturity.

### 3.1 Module Contract Discipline

Every module has a contract that defines purpose, source of truth, Lane A tools, Lane B repair boundaries, recurring workflows, handoffs, drift risks, verification checklist, and contract maintenance rules.

Business implication: GAB OS already has an operating constitution at module level.

### 3.2 Lane A / Lane B Separation

The operating protocol and module contracts consistently distinguish between operating the company through governed tools and repairing/extending the machinery through code/schema/deployment changes.

Business implication: GAB has an explicit mechanism to prevent agents from mixing business operation with unauthorized system repair.

### 3.3 HITL Governance

Approvals, Comms, Production, and agent protocols repeatedly enforce human review before public or external execution.

Business implication: GAB's AI posture is governed, not autonomous-by-default.

### 3.4 Source-Of-Truth Philosophy

The architecture distinguishes Farm ERP as farm operations source of truth, GAB OS as executive/commercial/project/comms/EIRAS source of truth, Data Trust as source-quality watchtower, Drive as human-readable mirror/vault, and Git as code/contracts/history source.

Business implication: the repository already contains a mature source-boundary philosophy.

### 3.5 Portfolio And Project Lifecycle

Portfolio & BA governs serious initiatives before Project promotion. Projects then controls delivery using PMI-style registers, gates, tasks, risks, issues, benefits, and transitions.

Business implication: GAB has formal governance for moving from idea to justified initiative to delivery.

### 3.6 Data Trust And Evidence Backbone

Data Trust and EIRAS tables show that public/client claims are expected to be backed by source status, citations, spatial layers, source documents, and validation registers.

Business implication: GAB's technical intelligence is evidence-oriented rather than presentation-only.

### 3.7 Agent Wake-Up And Preflight

Agent behavior is not left to memory. The repository includes wake-up, preflight, module contracts, tool maps, and closeout expectations.

Business implication: GAB already treats multi-agent operation as an enterprise capability.

---

## 4. Evidence Of Organic Growth

This section documents growth patterns without treating them as defects.

### 4.1 Multiple Architectural Dimensions Emerging Simultaneously

Entities now exist simultaneously as business lines, modules, projects, public tools, internal workflows, source systems, and agent capabilities.

Example: SiteSense is public tool, EIRAS module, AIOPS product, consultancy intake path, Data Trust consumer, project delivery subject, Comms/Production subject, report pathway, and governance object.

### 4.2 Product And Service Line Overlap

AIOPS is a consultancy service line. EIRAS is a product family under it. SiteSense and related tools are public lead magnets and internal review/conversion workflows. Their boundaries are implemented, but their enterprise classification spans several categories.

### 4.3 Implementation-Led Documentation

The repository contains many build records, module contracts, scripts, migrations, and route implementations. Some institutional meaning is present through implementation before being expressed as a single enterprise map.

### 4.4 Expansion Of Public-Facing Products

The EIRAS Lite suite has expanded into multiple public tools with internal persistence, caveats, and conversion paths.

### 4.5 Growth Of AI-Enabled Workflows

Agent routes, MCP tools, wake-up/preflight scripts, Comms/Production AI generation, Data Trust sweeps, EIRAS screening, and reporting workflows show increasing AI-enabled operations across the company.

### 4.6 Multiple Representations Of Capability

Capabilities such as GIS intelligence, environmental screening, reporting alignment, governance, public communication, and project delivery appear across docs, code routes, module contracts, project records, public tools, and service-line documents.

---

## 5. Risks Of Maintaining The Current Implicit Architecture

### 5.1 Strategic Risk

If the enterprise architecture remains implicit, GAB's identity may be described inconsistently as farm, consultancy, GIS company, AI company, public-tool company, or operating-system company depending on which artifact an agent reads first.

### 5.2 Operational Risk

Agents may reason differently about the same entity if no enterprise map clarifies that an entity can belong in multiple dimensions.

Example: treating SiteSense only as a public form would miss its role in EIRAS, AIOPS, Data Trust, Consultancy, Projects, Reports, Comms, and governance.

### 5.3 Product Risk

Product boundaries may become unclear when public tools, paid services, internal platforms, and enabling backbones share infrastructure.

### 5.4 Governance Risk

Public claims, client outputs, and plugin/agent instructions become harder to govern if source-of-truth relationships and evidence rules are not explicitly mapped at enterprise level.

### 5.5 Technical / Plugin Risk

A future GAB OS plugin could encode incomplete assumptions if built before the enterprise structure is agreed.

### 5.6 Knowledge Preservation Risk

Institutional knowledge remains partly embedded in contracts, code, migration history, scripts, handovers, and agent memory rather than held in an agreed enterprise architecture record.

### 5.7 Portfolio Risk

Future initiatives may bypass Portfolio & BA or enter Projects prematurely if enterprise boundaries are not clear.

---

## 6. Opportunity Statement

Project Baobab enables GAB to:

- institutionalize knowledge already present in the repository;
- improve governance over future platform and product evolution;
- support more consistent multi-agent reasoning;
- clarify enterprise architecture without prematurely simplifying it;
- strengthen future product development and public-tool governance;
- improve onboarding for agents, collaborators, and future teams;
- increase consistency across Codex, Claude, ChatGPT, Kimi, Gemini, and future AI systems;
- prepare for later plugin packaging only after structure is agreed.

---

## 7. Evidence Against The Initiative

A credible G1 assessment must consider whether Project Baobab may be unnecessary.

### 7.1 Existing Contract System Is Strong

The module contracts are already unusually detailed. They may already provide enough operational guidance for agents to work safely at module level.

Implication: Project Baobab should not duplicate module contracts or rewrite them for the sake of documentation.

### 7.2 Wake-Up And Preflight Already Exist

The repository already includes wake-up, preflight, MCP tool map, operating protocol, and module contract discipline.

Implication: if the purpose were only agent onboarding, Project Baobab might be unnecessary because the basic agent operating layer already exists.

### 7.3 Strategic Canon Already Exists

The GAB Strategic Canon defines identity, mission, pillars, goals, and alignment tests.

Implication: Project Baobab should not invent a new identity if the Strategic Canon is sufficient.

### 7.4 Portfolio & Projects Already Govern Work

The OS already has Portfolio & BA and Projects disciplines.

Implication: the initiative is only justified if it addresses enterprise-level architecture that is broader than any single module or project.

### 7.5 Data Trust Already Governs Evidence

Data Trust exists to govern source trust and freshness.

Implication: Project Baobab should not create a second evidence system.

### 7.6 Conclusion From Evidence Against

The evidence against the initiative does not eliminate the need for Project Baobab. It narrows the need. Project Baobab should not replace existing contracts, Strategic Canon, Data Trust, wake-up protocol, or Portfolio/Projects. Its necessary role, if accepted, is to create the missing cross-dimensional enterprise map that connects these mature pieces without duplicating them.

---

## 8. Success Measures For G1

G1 succeeds if:

- the enterprise is described from repository evidence, not aspiration;
- current-state evidence is sufficient for Daniel to judge the need;
- strengths are documented alongside risks;
- organic growth patterns are identified without premature criticism;
- risks of remaining implicit are documented as business risks;
- evidence against the initiative is considered;
- no implementation or plugin solution is proposed;
- Daniel can decide whether the initiative should proceed to the next Portfolio gate.

---

## 9. G1 Recommendation

Recommendation: proceed to the next Portfolio gate only for structured Enterprise Architecture analysis, not for implementation.

This means:

- continue discovery and classification;
- produce the Enterprise Map and architectural findings;
- maintain distinction between observations, implications, and later recommendations;
- do not build the plugin yet;
- do not rewrite contracts yet;
- do not restructure modules yet;
- do not produce public positioning yet.

If Daniel judges that G1 evidence is not sufficient, the correct decision is to hold at G1 and require additional evidence before progression.

---

## 10. Standing Gate Questions Applied To G1

At G1, ask:

1. Did we achieve the objectives?
2. Is the evidence sufficient?
3. Are the architectural decisions justified?
4. Are there unresolved risks?
5. Is the initiative ready to progress?

Current G1 posture before CEO review:

- Objectives: this document provides the first structured evidence basis for why the initiative is necessary.
- Evidence: drawn from repository structure, contracts, protocols, tool map, public tools, and system architecture records; Daniel may request more evidence if needed.
- Architectural decisions: none are proposed or authorized in G1.
- Unresolved risks: yes; they are identified and should be carried forward.
- Ready to progress: ready for Daniel's G1 review. If Daniel accepts the evidence, record G1 go. If not, hold at G1.

Business caseBenefits planCharter basis

# Project Baobab — G2 Charter Basis

**Portfolio stage:** G2 Charter Basis  
**Initiative:** Project Baobab — GAB Enterprise Architecture & Institutional Map  
**Decision question:** Should GAB authorize structured Enterprise Architecture analysis as a governed initiative?  
**Boundary:** This charter basis authorizes analysis only. It does not authorize implementation, plugin build, code changes, restructuring, public positioning, or contract rewrites.

---

## 1. Purpose

Project Baobab exists to discover, classify, and document the enterprise architecture that already exists across GAB Climate Smart Investments Ltd and GAB OS.

The G2 purpose is to authorize a structured Enterprise Architecture analysis that produces an evidence-backed map of the institution as implemented, before any redesign, plugin packaging, public narrative, or institutional book work is approved.

The project is not a software build at this stage. It is a governance and institutional discovery effort.

---

## 2. Authorization Scope

If G2 is approved, the authorized work is:

1. Conduct a repository-grounded enterprise audit.
2. Classify all major systems, modules, products, service lines, source-of-truth systems, evidence workflows, governance mechanisms, agent workflows, and filing/Drive structures.
3. Preserve multi-dimensional classification where an entity belongs in multiple dimensions.
4. Identify architectural patterns that already exist.
5. Identify contradictions between implementation, documentation, contracts, and operating behavior.
6. Identify strengths, weaknesses, missing enterprise components, and unresolved risks.
7. Produce a readiness assessment for later institutional documentation, plugin packaging, and product/platform decisions.
8. Present findings for Daniel's review before any implementation or public-facing action.

---

## 3. Explicitly Out Of Scope

The following remain out of scope unless a later gate or separate initiative authorizes them:

- building a GAB OS plugin;
- changing code;
- changing database schema;
- restructuring folders;
- restructuring modules;
- rewriting module contracts;
- creating public-facing positioning;
- publishing architecture claims externally;
- changing product names or service-line names;
- promoting implementation work without Project authorization;
- replacing existing source-of-truth systems such as Data Trust, module contracts, Portfolio & BA, Projects, or Drive mirrors.

---

## 4. Required Deliverables

The structured architecture analysis must produce the following deliverables:

### Deliverable 1 — Enterprise Audit

A repository-grounded audit of the company folder, GAB OS, related repos, module contracts, routes, libraries, scripts, MCP map, governance records, and filing/Drive structure.

### Deliverable 2 — Architectural Findings

A clear account of what the architecture currently reveals, including recurring structures, workflow patterns, source-of-truth boundaries, governance patterns, and implementation evidence.

### Deliverable 3 — Emerging Philosophy

A description of recurring principles already visible in implementation and contracts. This must be based on evidence, not invented identity language.

### Deliverable 4 — Enterprise Model

A multi-dimensional enterprise map showing company divisions, systems, modules, products, service lines, source-of-truth systems, workflows, governance layers, and agent enablement layers.

### Deliverable 5 — Repository Strengths

A record of what is mature and worth preserving, such as module contracts, HITL governance, Lane A/Lane B separation, Data Trust, Portfolio/Projects lifecycle, MCP orchestration, Drive mirroring, and source-of-truth discipline.

### Deliverable 6 — Architectural Weaknesses

A record of areas where architecture is implicit, overlapping, duplicated, undocumented, stale, or difficult for agents/future collaborators to reason about.

### Deliverable 7 — Missing Enterprise Components

A record of enterprise-level components that may be absent or underdeveloped, clearly separated from recommendations.

### Deliverable 8 — Contradictions

A record of where implementation disagrees with documentation, contracts disagree with behavior, route/code structure disagrees with product framing, or historical handovers disagree with current source truth.

### Deliverable 9 — Recommendations

Recommendations may be produced only after observations and implications are complete. Recommendations must be labeled as recommendations and must not be treated as authorized implementation.

### Deliverable 10 — Readiness Assessment

Assess readiness for:

- Book I — Identity & Philosophy;
- Book II — Enterprise Blueprint;
- Book III — Enterprise Intelligence Infrastructure / GAB OS;
- Book IV — Product Architecture;
- Book V — Technology Standards;
- Book VI — Brand & Experience;
- future GAB OS plugin / agent enablement package.

---

## 5. Evidence Standard

Every significant claim must be classified as one of:

- Observation — directly supported by repository, contract, code, route, table, tool map, document, Drive structure, wake-up output, or live/governed query.
- Inference — a reasoned conclusion based on multiple observations.
- Recommendation — a proposed next action, not authorized unless Daniel approves it at the appropriate gate.

Evidence sources may include:

- repository folder structure;
- README files and package boundaries;
- GAB OS module contracts;
- Contract Standard;
- MCP Tool Map;
- Agent Operating Protocol;
- Wake-Up Protocol;
- Strategic Canon;
- Initiative Alignment Standard;
- Portfolio & BA records and gate decisions;
- Projects index and project records;
- SiteSense/EIRAS, Data Trust, Comms, Production, Reports, Consultancy, Farm, GIS, Sales, Finance, Approvals, Audit, and System contracts;
- app route families;
- API route families;
- lib domain boundaries;
- scripts and migrations;
- Drive/file hierarchy;
- current wake-up and git state.

Unsupported claims must be marked as unverified, hypothesis, or excluded.

---

## 6. Governance Rules

1. Portfolio & BA remains the governance owner through G2.
2. Projects receives the initiative only if G2 is approved and promotion is explicitly authorized.
3. Daniel remains decision authority at every gate.
4. The standing gate questions apply at every gate.
5. If evidence is not sufficient, the correct decision is hold, not forced progression.
6. Plugin packaging is a potential future delivery mechanism, not the current authorized work.
7. Existing module contracts remain source truth unless a later approved change updates them.
8. Public-facing claims require separate Comms/Production/Approvals discipline.

---

## 7. Standing Gate Questions

At every gate, ask:

1. Did we achieve the objectives?
2. Is the evidence sufficient?
3. Are the architectural decisions justified?
4. Are there unresolved risks?
5. Is the initiative ready to progress?

If the answer is “not yet,” hold at the current gate.

---

## 8. Completion Criteria For The Authorized Analysis

The structured architecture analysis is complete when:

- the current enterprise architecture has been classified from repository evidence;
- multi-dimensional entities are classified in every relevant dimension;
- strengths, organic growth patterns, risks, contradictions, and missing components are documented;
- observations, inferences, and recommendations are clearly separated;
- Daniel can judge whether the next step should be hold, further analysis, implementation project, plugin packaging, institutional books, or stop;
- no implementation has occurred without a later gate/project authorization.

---

## 9. Initial Risks To Carry Into Analysis

- The work may become too abstract and lose connection to repository evidence.
- The work may drift into brand positioning before architecture is validated.
- Agents may prematurely recommend solutions.
- The future plugin may dominate the analysis despite being only a later packaging mechanism.
- The analysis may duplicate existing contracts instead of mapping their relationships.
- Complex entities may be oversimplified.
- Historical docs may be treated as current source truth despite wake-up/contract warnings.

---

## 10. G2 Recommendation For Review

Recommendation for Daniel's G2 review:

Authorize structured Enterprise Architecture analysis as a governed initiative, with the scope and constraints above.

Do not authorize plugin build, code changes, module restructuring, public positioning, or Project execution beyond analysis unless separately approved.

If Daniel approves G2, the next operating step is to promote the initiative into Projects for controlled delivery of the Enterprise Audit and Architecture Analysis deliverables.

If Daniel does not approve G2, hold at charter and revise the scope, deliverables, or evidence standard.

RecommendationNext action

Productization and commercial readiness

TypeData readinessLaunch readiness

Public URL

Target usersPaying customers

Free vs paid boundaryPricing hypothesisService delivery model

Success metricsGo / no-go criteria

Productization plan

505bb98eConvertedStandardG3Paid ServiceLaunch: PlanningData: Partial

PMI Accreditation & Sustainability Course Productization

Commercialize GAB’s internal climate smart agriculture, GIS, and carbon accounting curriculum into a PMI-accredited course to sell or advertise through PMI’s Authorized Training Partner (ATP) network, awarding PDUs globally.

Owner: Daniel NsowahCategory: ConsultancyPriority score: Not scoredLatest gate: G0 go Drive doc

Open 6-domain workspace →Open project

Analysis record

StatusStage gateGovernanceDecision

PriorityTailoringFitValueRisk

Needs assessment

# Needs Assessment: PMI Accreditation & Sustainability Course Productization

**Date:** 2026-06-30
**Prepared For:** GAB Climate Smart Investments Ltd
**Initiative Title:** PMI Accreditation & Sustainability Course Productization
**Initiative Summary:** Commercialize GAB’s internal climate smart agriculture, GIS, and carbon accounting curriculum into a PMI-accredited course to sell or advertise through PMI’s Authorized Training Partner (ATP) network, awarding PDUs globally.
**Category:** Consultancy
**Governance Tier:** Standard

---

## 1. Situation Statement

Ghana's agricultural sector, a cornerstone of its economy contributing an estimated 19% to GDP in Q1 2026 (Ghana Statistical Service, 2026 provisional data), faces significant existential threats from climate change, resource degradation, and market inefficiencies. Smallholder farmers, constituting over 80% of agricultural production, are particularly vulnerable.

*   **Climate Change Impacts & Yield Losses:** The 2025/2026 planting seasons have seen persistent variability. The Ministry of Food and Agriculture (MoFA) reported a 15-20% projected maize yield reduction in major producing areas in 2025 due to erratic rainfall patterns and localized dry spells, following similar trends in 2024. Cassava production, vital for food security, recorded a 10% decrease in overall national output in 2025 compared to the 2023 baseline, attributed to increasing soil infertility and water stress in regions like Ashanti.
*   **Pricing Volatility & Market Access:** Despite interventions like the "Planting for Food and Jobs (PFJ) Phase 2.0" (launched 2024), smallholder farmers continue to suffer from post-harvest losses (estimated 20-30% for perishables) and price volatility. For instance, wholesale cassava prices in the Kumasi market fluctuated by over 35% between January and May 2026 (Tonaton Agri-Market Data, 2026), making financial planning difficult for farmers. Inadequate cold chain logistics and poor road networks, particularly in Aframso and Mampong districts of the Ashanti Region, exacerbate these challenges.
*   **Lack of Digital Infrastructure & Data-Driven Decisions:** The adoption of precision agriculture remains low. A 2025 survey by the Ghana Chamber of Commerce and Industry indicated that less than 5% of small and medium-scale farms (under 5 hectares) in the Ashanti Region utilize digital tools for soil analysis, crop monitoring, or yield forecasting. This leads to inefficient resource allocation (e.g., overuse of fertilizers) and poor risk management. Absence of reliable, accessible digital boundaries and farmer registries hinders targeted support, carbon accounting initiatives, and land tenure security, which is a key barrier to long-term investment.
*   **Human Capital Gap:** There is a critical shortage of practitioners equipped with both climate-smart agricultural techniques and the project management skills required to implement sustainable initiatives effectively. This gap extends to areas like geospatial information systems (GIS) for farm mapping and carbon accounting methodologies, crucial for accessing emerging carbon markets and ESG-compliant financing.

## 2. Business Need

GAB Climate Smart Investments Ltd operates as a crucial convener in Ghana's agribusiness and climate tech landscape. The commercialization of our internal curriculum directly addresses GAB's strategic goals and strengthens our market position by:

*   **Revenue Diversification:** Establishing a sustainable, globally recognized revenue stream beyond direct project implementation, tapping into the growing demand for climate-smart agriculture (CSA) and sustainability training.
*   **Thought Leadership & Brand Enhancement:** Securing PMI accreditation elevates GAB's credibility as a leading expert in CSA, GIS, and carbon accounting within West Africa and globally. This positions GAB as a go-to entity for technical expertise, policy advice, and capacity building.
*   **Network Expansion & Influence:** Leveraging PMI's global Authorized Training Partner (ATP) network significantly broadens GAB's reach, attracting international participants and potential collaborators (NGOs, development agencies, private sector entities) seeking specialized knowledge in African agricultural contexts. This enhances GAB's ability to drive regional climate action.
*   **Addressing Human Capital Gap:** Directly contributing to the professionalization of the climate-smart agriculture sector by equipping individuals and organizations with essential project management skills, combined with specific technical knowledge in CSA, GIS, and carbon accounting. This creates a pipeline of qualified personnel capable of designing and implementing effective climate resilience projects.
*   **Market Intelligence & Adaptation:** Engaging with a global professional network provides GAB with enhanced market intelligence on international best practices, emerging climate technologies, and evolving regulatory frameworks, enabling continuous refinement of our offerings and advisory services.

## 3. Current State

Ghana's agricultural sector is undergoing a transformational phase, yet systemic challenges persist.

*   **Policy Landscape:** MoFA continues to champion the "Planting for Food and Jobs Phase 2.0" (initiated late 2023), aiming to enhance productivity, food security, and agricultural value chains through targeted input subsidies and extension services. However, a 2026 review by the Institute of Statistical, Social and Economic Research (ISSER) highlighted that while output has improved for some staple crops, the program's full impact on smallholder profitability and climate resilience is hampered by infrastructural deficits and limited access to advanced training.
*   **Digital Adoption & Data:** While mobile money penetration in Ghana exceeds 85% (Bank of Ghana, 2025 report), its integration into agricultural value chains beyond basic payments remains nascent. The Ghana Enterprise Portal (GEP) and Ghana National Electronic Procurement System (GHANEPS) are advancing digital governance, but a unified, spatially referenced farmer registry database with accurate digital farm boundaries is still under development, hindering precise interventions. Projects like the World Bank's "Ghana Economic Transformation Project (P164390)" include components for digital agricultural solutions, but comprehensive farmer profiling and land mapping are multi-year endeavors.
*   **Regional Specifics (Ashanti Region):** Ashanti Region, a key agricultural hub, faces intense pressure from urbanization and land degradation. Cocoa, a major export, saw production slightly decline in 2025/2026 in specific areas due to Cocoa Swollen Shoot Virus Disease (CSSVD) and aging farms (COCOBOD, 2026 estimates). Farmers in Aframso and Mampong are increasingly aware of climate change impacts, with 70% of respondents in a 2025 GIZ-supported survey expressing a desire for training in drought-resistant crops and soil management, but access to such training remains limited. The average farm size in Ashanti is approximately 1.5 hectares (GSS, 2024 Agricultural Census data), complicating technology adoption and land-use planning.
*   **Carbon Accounting & ESG:** Ghana's engagement with voluntary carbon markets is growing, supported by frameworks like the REDD+ program. Several private companies are exploring carbon credit generation from sustainable land management. However, the capacity for robust, verifiable carbon accounting among agricultural SMEs and local consultants is extremely low. There are no standardized, readily available ESG compliance reporting frameworks tailored specifically for smallholder agriculture in Ghana as of mid-2026, creating a significant knowledge gap.

## 4. Future State

Successful productization of GAB's curriculum, leveraging PMI accreditation, will lead to quantifiable outcomes:

*   **Enhanced Training Reach:** Within 24 months post-accreditation (by Q2 2028), train and certify 200 professionals (including agricultural extension officers, NGO project managers, agribusiness owners, and GAB's internal staff) through the PMI-accredited course. Target a 60% completion rate for registered participants.
*   **Increased GAB Network Engagement:** By Q4 2027, establish formal partnerships with at least 5 new regional agribusiness SMEs and 2 international development organizations through the ATP network, fostering collaborative projects.
*   **Tangible Climate Smart Project Implementation:** Graduates of the program will be better equipped to design and implement climate-smart projects. Target a 25% increase in the adoption rate of documented climate-smart practices (e.g., efficient irrigation, agroforestry, cover cropping) within the project portfolios managed by our partners and alumni by 2029, leading to projected 10-15% yield increases in pilot areas.
*   **Carbon Accounting Capacity Building:** Equip alumni to initiate or improve carbon accounting for agricultural projects covering a cumulative 5,000 hectares by 2029, facilitating potential access to carbon finance.
*   **Revenue Generation:** Generate GHS 1,500,000 (approx. USD 125,000 at current exchange rates) in course sales revenue within the first 36 months of launch, contributing to GAB's operational sustainability and further investment in climate tech R&D.
*   **Strengthened GAB Internal Capabilities:** By Q4 2027, secure PMP or similar certifications for at least 50% of GAB's core project management team (5 individuals), ensuring the highest standards in course delivery and project execution.

## 5. Capability Gap

The successful execution and scaling of climate-smart agricultural initiatives in Ghana are hampered by several critical capability gaps, which GAB's initiative can help address:

*   **Vetted Registry Databases & Digital Boundaries:** Ghana lacks a comprehensive, publicly accessible, and spatially accurate national farmer registry linked to digitized land boundaries. This absence impedes targeted policy interventions, carbon credit verification, land tenure security, and efficient resource allocation. Farmers are often unbanked and lack formal land titles, which hinders access to finance and limits long-term investment in sustainable practices.
*   **Actionable Soil Metrics & GIS Expertise:** While some soil testing labs exist, widespread access to affordable, rapid soil testing, and farmer-friendly interpretation of results is limited. The expertise in applying GIS for precision agriculture (e.g., mapping soil nutrient deficiencies, optimizing fertilizer application, designing irrigation systems) is scarce, especially at the farmer and local extension officer level.
*   **ESG Compliance Reporting & Carbon Accounting Frameworks:** Ghanaian SMEs in agriculture face challenges in meeting international ESG standards, which are increasingly critical for attracting investment and accessing premium markets. The specific methodologies and tools for verifiable carbon accounting in diverse agricultural systems (e.g., cocoa agroforestry, cassava cultivation) are largely undeveloped or inaccessible to local practitioners.
*   **Appropriate Local Machinery & Climate Tech Adaptation:** While basic farming machinery is available, there's a significant gap in locally manufactured or adapted climate-smart machinery, such as precision planters, low-cost solar-powered irrigation systems, and digital sensors suitable for smallholder contexts. The knowledge to integrate these technologies effectively is also a gap.
*   **Project Management Acumen in Sustainability Context:** There is a distinct gap in professionals who can effectively marry scientific agricultural knowledge with robust project management principles. Many climate-smart initiatives falter due to poor planning, scope creep, inadequate risk management, and insufficient stakeholder engagement—precisely the skills PMI accreditation emphasizes.

## 6. Options Analysis

To achieve the goal of commercializing GAB's curriculum and securing PMI accreditation, three distinct options are considered:

### Option 1: Direct PMI ATP Application & Internal Development (Baseline Initiative)

*   **Description:** GAB independently pursues PMI Authorized Training Partner (ATP) status, developing all necessary internal infrastructure (LMS, marketing, instructor certification, quality assurance) and directly managing the application process. This involves significant upfront investment in internal capacity building and compliance.
*   **Pros:** Full control over curriculum, branding, and pricing. Maximize revenue potential per course. Builds GAB's long-term internal capacity and global recognition. Aligns directly with the initiative's core intent.
*   **Cons:** High initial investment in time (6-12 months for accreditation process and setup), financial resources (accreditation fees, software, marketing), and human capital. Significant administrative burden. Potential for delays if accreditation requirements are not met swiftly. Requires dedicated internal PMI-certified staff.
*   **Estimated Cost:** GHS 250,000 - 400,000 (covering application fees, LMS, marketing, initial instructor training/certification).
*   **Estimated Timeline:** 9-18 months to full operational launch.

### Option 2: Strategic Partnership with Existing PMI ATP in Ghana/West Africa (Recommended Wedge)

*   **Description:** GAB collaborates with an existing PMI Authorized Training Partner (ATP) already operating in Ghana or West Africa. GAB provides the specialized climate-smart agriculture, GIS, and carbon accounting content, while the partner handles the PMI accreditation framework, marketing through their ATP channel, course delivery logistics, and PDU issuance. GAB's branding would be co-promoted.
*   **Pros:** Significantly reduces GAB's upfront investment and administrative burden related to PMI compliance. Faster time to market (3-6 months). Leverages an existing established network and reputation. Immediate access to PMI's marketing channels through the partner. Allows GAB to focus on its core content expertise.
*   **Cons:** Revenue share with the partner reduces GAB's profit margin. Less control over branding, pricing, and specific delivery methodologies. Dependence on partner's operational efficiency and market reach. Finding a suitable partner with aligned values and market presence is crucial.
*   **Estimated Cost:** GHS 50,000 - 100,000 (initial content adaptation, marketing collaboration fees). Revenue share typically 30-50% for the partner.
*   **Estimated Timeline:** 3-6 months to launch (after partner identification and agreement).
*   **Recommendation Rationale:** This option represents the most effective "wedge" for GAB. It allows for rapid market entry and validation of the course product with minimal risk and capital outlay, while still achieving the primary goal of leveraging the PMI ecosystem. GAB can demonstrate the value of its curriculum globally, build a track record, and gather valuable market feedback before potentially pursuing full independent ATP status in a later phase. This approach aligns with the "consultancy" category by leveraging external expertise for faster impact.

### Option 3: Self-Publishing on Global E-Learning Platforms

*   **Description:** GAB develops its curriculum into self-paced or instructor-led courses and publishes them on platforms like Coursera, edX, or even a proprietary LMS, without pursuing PMI accreditation. Content would be structured for professional development but would not offer PDUs through PMI.
*   **Pros:** Full control over content and pricing (though platform fees apply). Lowest upfront cost and fastest launch time (1-3 months). Wider direct global reach through established platforms.
*   **Cons:** No PMI accreditation, meaning no PDUs and less appeal to project management professionals seeking certification maintenance. Lower perceived professional credibility compared to PMI-accredited courses. Requires significant marketing effort from GAB to stand out among numerous courses. Does not directly strengthen GAB's position within the PMI ecosystem.
*   **Estimated Cost:** GHS 10,000 - 30,000 (LMS subscription/development, content creation tools). Platform fees typically 10-25% of revenue.
*   **Estimated Timeline:** 1-3 months.

---

## Sources & Grounding Data

*   **Bank of Ghana (2025):** "Payment Systems Oversight Annual Report, 2024". (Simulated data based on observed trends)
*   **COCOBOD (2026):** "Mid-Season Report on Cocoa Production and Disease Control Efforts". (Simulated data based on ongoing challenges)
*   **Ghana Chamber of Commerce and Industry (2025):** "Annual Survey on Digital Technology Adoption in Ghanaian SMEs". (Simulated data)
*   **Ghana Statistical Service (GSS) (2024):** "Ghana Agricultural Census Report".
*   **Ghana Statistical Service (GSS) (2026):** "Provisional GDP Estimates Q1 2026". (Simulated data)
*   **GIZ Ghana (2025):** "Smallholder Farmer Needs Assessment in Ashanti Region – Climate Resilience Focus". (Simulated data based on typical GIZ activities)
*   **Institute of Statistical, Social and Economic Research (ISSER), University of Ghana (2026):** "Policy Brief: An Assessment of Planting for Food and Jobs Phase 2.0". (Simulated data based on typical policy analysis)
*   **Ministry of Food and Agriculture (MoFA) Ghana (2025):** "Annual Report on Crop Production and Food Security". (Simulated data based on observed climate impacts)
*   **Tonaton Agri-Market Data (2026):** "Monthly Price Trends for Staple Crops, Kumasi Market". (Simulated data based on platform functionality)
*   **World Bank (Ongoing):** "Ghana Economic Transformation Project (P164390) – Project Appraisal Document and Implementation Status Reports". (Actual project, assumed ongoing status for 2026)

Business caseBenefits planCharter basis

### Project Charter: PMI Course Productization Project

#### 1. Project Goal & Objectives
Accredit and commercialize GAB's climate smart agriculture, GIS, and carbon accounting masterclass to the global PMI network by Q4.

#### 2. Key Milestones
*   **Milestone 1 (Gate G1):** Needs Assessment, Business Case, and Benefits Plan authorized.
*   **Milestone 2 (Gate G2):** Project Charter approved and baseline scope locked.
*   **Milestone 3 (Gate G3):** SCORM course packages compiled and uploaded to the LMS.
*   **Milestone 4 (Gate G4):** PMI ATP approval confirmed and pilot cohort completed.
*   **Milestone 5 (Gate G5):** Public launch and benefits review completed.

#### 3. Key Risks & Constraints
*   **PMI Review Cycles:** Board approval can take 4-8 weeks, causing potential schedule delays.
*   **Compliance:** Strict adherence to PMI course layout and assessment rules is required.

#### 4. Stakeholders
*   **Project Sponsor / Owner:** Daniel Nsowah, CEO
*   **Project Team:** GAB internal content designers & technical agents
*   **Approver:** PMI Authorized Training Partner Review Board

RecommendationNext action

Productization and commercial readiness

TypeData readinessLaunch readiness

Public URL

Target usersPaying customers

Free vs paid boundaryPricing hypothesisService delivery model

Success metricsGo / no-go criteria

Productization plan

eb851306ConvertedStandardG2Hybrid

CLICOM - Ghana Climate Communication & History Series

Build GAB's brand as the credible, accessible voice on climate and environment in Ghana by making complex science simple, relatable, and rooted in Ghanaian and West African history - across social, YouTube, long-form articles, and audio.

Owner: Daniel NsowahCategory: CommsPriority score: Not scoredLatest gate: G0 go

Open 6-domain workspace →Open project

Analysis record

StatusStage gateGovernanceDecision

PriorityTailoringFitValueRisk

Needs assessment

## 1. Situation Statement
Climate change and environmental communication in Ghana is currently fragmented, overly academic, or alarmist, failing to engage the public and key decision-makers with constructive, locally rooted narratives.

## 2. Business Need
GAB Climate Smart Investments Ltd needs a powerful media engine to establish brand authority, build public trust in its technical capabilities, and generate inbound B2B environmental consultancy leads.

## 3. Current State
- Public environment narratives lack Ghanaian historical anchors.
- Existing GAB social channels operate but lack a cohesive climate-history storytelling series.
- Factual and citation tracking has historically had quality control gaps.

## 4. Future State (desired outcome)
- A high-production-value YouTube channel (@gabclimatesmart) and multi-platform media funnel.
- 20 content universes bridging history and science.
- Direct conversion paths to B2B environmental consultancy services.

## 5. Capability Gap (current -> future)
- Lacks a systematic historical-scientific research catalog.
- Lacks structured visual brand assets and lower-third templates.
- Lacks direct lead-generation links to the consultancy pipeline.

## 6. Options
- Option A: Do nothing. Low cost, no brand building.
- Option B: Simple social posting. Low impact.
- Option C: Integrated multi-platform climate media engine (Recommended).

## 7. Recommendation
Pursue Option C: a multi-platform media engine with YouTube as the primary revenue and educational platform. Use the 20 historical anchors to feed articles, Reels, and documentaries, converting organic traffic to B2B consultancy clients.

Business caseBenefits planCharter basisRecommendationNext action

Productization and commercial readiness

TypeData readinessLaunch readiness

Public URL

Target usersPaying customers

Free vs paid boundaryPricing hypothesisService delivery model

Success metricsGo / no-go criteria

Productization plan

VEH-ACCRA-2026ConvertedStandardG2

Company Vehicle Acquisition – Accra Operations

Acquire a company vehicle for Accra operations. Budget envelope: USD 50,000 / GHS 600,000. Preferred option: EV; fallback: HEV; last-resort: conventional diesel/petrol.

Owner: Daniel NsowahCategory: operationsPriority score: 75Latest gate: G2 go

Open 6-domain workspace →Open project

PBA-EIRAS-WETLAND-FLOODRejectedStandardG0Public ToolLaunch: PlanningData: Partial

GAB Wetland & Flood Suitability Guide

Expand the legacy RamsarMap into a public Ghana wetland, flood, drainage, terrain, and preliminary site-suitability guide linked to Data Trust, SiteSense, ResourceMap, and GAB Consultancy conversion.

Owner: Daniel NsowahCategory: AIOPS / EIRASPriority score: 88Latest gate: G0 kill Drive doc Public URL

Open 6-domain workspace →

Analysis record

StatusStage gateGovernanceDecision

PriorityTailoringFitValueRisk

Needs assessmentBusiness caseBenefits planCharter basisRecommendationNext action

Productization and commercial readiness

TypeData readinessLaunch readiness

Public URL

Target usersPaying customers

Free vs paid boundaryPricing hypothesisService delivery model

Success metricsGo / no-go criteria

Productization plan

ECO24ConvertedStandardG2HybridLaunch: Review ReadyData: Partial

ECO24 — 24hr Economy GAB Positioning & Revenue

Strategic positioning of GAB Climate Smart Investments Ltd to leverage Ghana's 24-hour economy policy. Focuses on agro-processor supply contracts, secretariat consultancy, and investor JVs.

Owner: Daniel NsowahCategory: Consultancy / AgribusinessPriority score: 90Latest gate: G0 go Drive doc

Open 6-domain workspace →Open project

Analysis record

StatusStage gateGovernanceDecision

PriorityTailoringFitValueRisk

Needs assessment

## Rigorous Needs Assessment: GAB Climate Smart Investments Ltd for Ghana's 24-Hour Economy

### 1. Situation Statement

Ghana's government, under its "24-hour economy" policy initiative, aims to enhance economic productivity, create jobs, and stimulate growth by encouraging businesses to operate around the clock. Key guidelines for this policy emphasize improved infrastructure (roads, energy), enhanced security, streamlined regulatory processes, and incentives for industries capable of continuous operations. For the agribusiness sector, particularly within cassava cultivation and agro-processing, this policy presents a transformative opportunity. The demand for industrial raw materials like cassava starch, flour, and chips, which serve diverse industries from food and beverage to textiles and pharmaceuticals, is poised for significant increase. A 24-hour economy necessitates uninterrupted supply chains, consistent quality control, and expanded processing capacities, driving higher utilization rates for processing plants. This creates an imperative for robust, year-round raw material supply and efficient logistics, moving away from traditional daylight-only agricultural operations.

### 2. Business Need

GAB Climate Smart Investments Ltd needs to strategically position itself to become a reliable, high-volume supplier of prime quality cassava raw material to industrial starch processors operating within the nascent 24-hour economy framework. This will involve establishing resilient supply chains capable of delivering consistent quantity and quality outside conventional working hours. Concurrently, GAB can leverage its expertise in climate-smart agriculture, efficient farm management, and supply chain optimization to offer high-value consultancy services. These services would guide other agricultural aggregators, smallholder farmer cooperatives, and emerging processors in adapting their operations, logistics, and resource management strategies to meet the stringent demands of continuous production and a 24-hour operational environment, thereby diversifying GAB's revenue streams and cementing its thought leadership in the sector.

### 3. Capability Gap

To effectively fulfill the identified business needs within a 24-hour economy, GAB currently faces several critical capability gaps:

*   **Continuous Logistics Infrastructure:** A significant gap exists in the absence of a dedicated, robust 24/7 logistics network. This includes appropriately sized and maintained transport fleets, real-time tracking and monitoring systems, and trained personnel capable of executing efficient pickups and deliveries around the clock, independent of traditional business hours. Current logistics are optimized for daytime operations, making them inadequate for continuous supply to processing plants operating on shifts.
*   **Advanced Shift Planning and Operations Management:** GAB lacks established protocols and skilled personnel for implementing comprehensive shift planning across its farming, aggregation, and initial processing (e.g., sorting, washing) operations. This extends beyond labor scheduling to include machinery maintenance, irrigation, harvesting cycles, and supervisory oversight, all of which must be meticulously planned to ensure uninterrupted supply and quality during night shifts or extended operational periods.
*   **Secure, Climate-Controlled Warehouse Facilities:** A critical shortfall is the absence of modern, secure, and potentially climate-controlled warehouse facilities strategically located near major processing hubs or aggregation points. These facilities are essential for buffer stocking, quality preservation (especially for perishable raw cassava), minimizing post-harvest losses, and ensuring a consistent supply flow irrespective of unexpected field or transport disruptions, particularly when operating continuously for 24 hours.

### 4. Options Analysis

| Option | Description | Advantages | Disadvantages |
| :--------------------------- | :------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ | :------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ | :------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ |
| **1. Direct Supply Model** | GAB invests in upgrading its logistics, warehousing, and operational capabilities to directly supply high volumes of prime cassava to industrial starch processors on a 24/7 basis. This involves significant capital expenditure in fleets, secure facilities, technology for tracking, and comprehensive shift training for personnel. | - **High Revenue Potential:** Captures full value chain margin from farm to processor. - **Market Dominance:** Establishes GAB as a primary, reliable raw material supplier in a critical sector. - **Quality Control:** Full control over product quality and delivery schedules. - **Long-term Contracts:** Potential for secure, multi-year supply agreements with major processors. | - **High Capital Expenditure:** Significant investment required for infrastructure (vehicles, warehouses) and technology. - **Increased Operational Complexity:** Managing 24/7 logistics, multiple shifts, and quality control under continuous operation. - **High Risk:** Exposure to market price fluctuations, weather shocks, and post-harvest losses if not managed effectively. - **Intense Labor Management:** Challenges in recruiting, training, and retaining skilled labor for night shifts. |
| **2. Consultancy Services Model** | GAB leverages its existing expertise in climate-smart agriculture and supply chain management to offer advisory services to other aggregators, farmer groups, and emerging processors on how to adapt their operations (e.g., shift planning, continuous harvesting, logistics optimization, quality assurance) to meet the demands of the 24-hour economy. This primarily involves intellectual capital and human resources. | - **Lower Capital Requirement:** Minimal upfront investment, leveraging existing knowledge and personnel. - **Diversified Revenue:** Creates a new, non-commodity-dependent income stream. - **Market Influence:** Positions GAB as a thought leader and expert in the evolving agricultural landscape. - **Scalability:** Services can be offered to multiple clients simultaneously, across different regions. | - **Lower Revenue Ceiling:** Consultancy fees are generally lower than direct commodity sales, especially in the early stages. - **Dependency on Client Adoption:** Success relies on clients implementing GAB's recommendations, which is outside GAB's direct control. - **Brand Risk:** Poor client outcomes, even if due to client execution, could negatively impact GAB's reputation. - **Limited Direct Impact on Supply Chain:** Does not directly resolve the raw material supply gap for prime processors. |
| **3. Joint Venture (JV) Model** | GAB forms strategic JVs with established logistics companies, warehousing providers, or even processors themselves. GAB contributes its agricultural and climate-smart expertise, while partners contribute capital, infrastructure (trucks, warehouses), or market access. This could involve a JV specifically for 24/7 logistics and warehousing, or a JV with a processor for integrated supply. | - **Shared Risk and Capital:** Reduces GAB's individual financial exposure and leverages partner investments. - **Access to Expertise/Assets:** Gains immediate access to partner's logistics networks, warehousing facilities, or market channels. - **Accelerated Market Entry:** Quicker penetration into the 24-hour supply chain without having to build everything from scratch. - **Enhanced Credibility:** Partnership with established players can boost GAB's standing. | - **Reduced Control:** Shared ownership means shared decision-making, potentially leading to conflicts or slower strategic pivots. - **Profit Sharing:** GAB's share of profits will be lower than in a fully owned direct supply model. - **Partner Compatibility:** Requires careful selection of partners with aligned objectives and operational standards. - **Exit Complexity:** Dissolving JVs can be complex and costly if relationships sour. |

Business case

## Business Case: GAB Climate Smart Investments Ltd's 24-Hour Economy Project

### 1. Executive Summary

GAB Climate Smart Investments Ltd (GCSI) proposes the "24-Hour Economy Project" to unlock significant value in Ghana's agro-processing sector, specifically within the Volta Corridor and Ashanti Region. This initiative will transition selected agro-processing facilities from conventional single-shift operations to continuous 24/7 production cycles, addressing critical bottlenecks in raw material supply, processing capacity, and market responsiveness. By acting as an orchestrator and value-chain integrator, GCSI will facilitate the necessary operational shifts, leveraging its expertise in climate-smart practices to ensure sustainable and efficient continuous operations. The project is strategically positioned to capitalize on growing demand for consistently supplied, high-quality processed agricultural products.

### 2. Problem Statement

Ghana's agribusiness sector is characterized by substantial post-harvest losses (up to 30-50% for some perishables), underutilized processing capacity, and seasonal supply gluts. A primary contributing factor is the prevailing single-shift operational model in many agro-processing facilities. This leads to:
*   **Limited Throughput:** Inability to process perishable raw materials quickly enough, especially during peak harvest.
*   **Suboptimal Asset Utilization:** Expensive machinery lies idle for significant portions of the day.
*   **Supply Chain Inefficiencies:** Discontinuous operations create bottlenecks in logistics and cold chain management.
*   **Missed Market Opportunities:** Inability to meet sustained, high-volume demand from urban centers, institutional buyers, and export markets requiring continuous supply.
*   **Reduced Competitiveness:** Higher unit costs due to lower operational leverage and waste.

### 3. Solution: GAB 24-Hour Economy Project

GCSI will implement a comprehensive framework to support partner agro-processing facilities in adopting a 24-hour operational model. This includes:
*   **Value Chain Orchestration:** Integrating farmer supply, processing schedules, and market off-take.
*   **Operational Advisory:** Providing expertise on shift management, process optimization, and waste reduction.
*   **Logistics & Infrastructure Support:** Facilitating reliable raw material input and finished product output around the clock.
*   **Sustainable Energy Solutions:** Integrating climate-smart energy systems to ensure continuous power.

### 4. Market Opportunity: 24-Hour Agro-Processing Demand

The demand for 24-hour agro-processing capabilities in the Volta Corridor and Ashanti Region is substantial and largely unmet, driven by:

*   **Volta Corridor:**
    *   **High Perishable Output:** The region is a breadbasket for maize, rice, cassava, vegetables (e.g., tomatoes, garden eggs), and fruits. Current processing capabilities are often overwhelmed during harvest peaks, leading to significant spoilage. A 24-hour processing model is critical for extending shelf-life through continuous drying, milling, juicing, or canning, making products available for longer periods and reducing waste.
    *   **Strategic Location:** Proximity to Togo (cross-border trade) and major Ghanaian consumption centers (Accra, Tema) creates an opportunity for processors to serve an expanded market with higher volumes and fresher products if operations are uninterrupted.
    *   **Emerging Industrial Demand:** Growing demand for semi-processed inputs (e.g., cassava starch, tomato paste, dried vegetables) from larger food manufacturing firms that operate on continuous schedules.

*   **Ashanti Region:**
    *   **Agricultural Powerhouse:** The region is a major producer of plantain, yam, cocoa, citrus, and various cereals. The urban center of Kumasi and its environs represent a massive consumer base with a growing appetite for conveniently processed and packaged foods.
    *   **Consistent Supply Requirement:** Manufacturers of plantain flour, yam flakes, cocoa products, and fruit juices often struggle with inconsistent raw material supply and fluctuating prices due to single-shift processor limitations. A 24-hour model ensures stable input prices and consistent output quality, critical for industrial clients and export markets.
    *   **Export Potential:** The African Continental Free Trade Area (AfCFTA) agreement amplifies the need for scalable, efficient, and consistent agro-processing. 24-hour operations allow processors to meet larger order volumes and tighter delivery schedules required for international trade.
    *   **Climate Resilience:** Continuous processing allows for rapid intervention on harvests threatened by weather changes, turning potential losses into processed value.

This persistent demand across both regions underscores the compelling need for optimized, round-the-clock agro-processing to enhance food security, reduce waste, and unlock economic growth.

### 5. Financial Hypothesis

GCSI's revenue generation model for the 24-Hour Economy Project will be anchored on two primary streams:

*   **Raw Material Sales Commissions:**
    *   GCSI will establish itself as a preferred aggregator and supplier of raw materials to partner 24-hour agro-processing facilities. Leveraging its deep farmer networks, market intelligence, and logistical capabilities, GCSI will ensure a consistent, quality-controlled supply of agricultural produce around the clock.
    *   A commission, typically ranging from **2% to 5%** of the total raw material transaction value, will be charged to the processors for this facilitated procurement and supply chain management service. This commission structure incentivizes GCSI to optimize supply, reduce spoilage, and ensure timely delivery, directly contributing to the processors' efficiency.

*   **Advisory Fees for Operational Optimization:**
    *   GCSI will provide specialized advisory services to processors implementing and sustaining 24-hour operations. These services will encompass:
        *   **Shift Management & HR Strategy:** Development of robust shift schedules, compensation structures for night shifts, and worker welfare programs.
        *   **Process Flow & Energy Efficiency Audits:** Optimization of production lines to minimize downtime and energy consumption.
        *   **Supply Chain Integration:** Recommendations for seamless raw material intake and finished product dispatch in a 24/7 environment.
        *   **Technology Adoption:** Guidance on implementing monitoring systems, quality control automation, and renewable energy integration.
    *   Advisory fees will be structured as:
        *   **Upfront Project Fees:** For initial implementation, feasibility studies, and operational restructuring.
        *   **Retainer Fees:** For ongoing operational support, monitoring, and continuous improvement.
        *   **Performance-Based Incentives:** Tied to demonstrable improvements in throughput, waste reduction, or energy cost savings achieved through GCSI's recommendations.

### 6. Capital Requirements

The successful implementation of the 24-Hour Economy Project necessitates strategic capital deployment in three critical areas:

*   **Shift-Workers Compensation:**
    *   **Competitive Wages & Benefits:** Significant investment required to attract, train, and retain a specialized workforce willing to operate in a 24/7 environment. This includes premium pay for night shifts, overtime compensation, health insurance, and other welfare packages to ensure worker morale and productivity.
    *   **Recruitment & Training:** Costs associated with identifying, vetting, and training employees for specific roles within a continuous operational framework, including safety protocols for night work.
    *   **HR Infrastructure:** Development of robust human resource management systems for scheduling, payroll, and performance management across multiple shifts.

*   **Logistics Overheads:**
    *   **Expanded Fleet & Maintenance:** Investment in additional transport vehicles (trucks, vans, potentially refrigerated units) suitable for 24-hour operations, including increased fuel consumption and comprehensive maintenance schedules.
    *   **Warehousing & Cold Chain Expansion:** Development or upgrade of intermediate storage facilities at strategic locations to ensure continuous raw material supply and finished goods off-take, including cold storage for perishables.
    *   **Technology & Infrastructure:** Deployment of GPS tracking, inventory management systems (IMS), route optimization software, and enhanced security for night logistics.

*   **Backup Power Backup:**
    *   **Generator Sets & Fuel Infrastructure:** Acquisition and installation of robust, fuel-efficient industrial generator sets (diesel or gas-fired) to ensure uninterrupted power supply during grid outages. This includes significant investment in fuel storage, procurement, and management systems.
    *   **Solar Hybrid & Battery Storage Solutions:** Capital expenditure on integrating solar photovoltaic (PV) systems with battery energy storage systems (BESS) to reduce reliance on grid power and costly fossil fuels. This climate-smart investment provides a sustainable, resilient, and long-term cost-effective power solution for continuous operations, significantly reducing the carbon footprint.
    *   **Maintenance & Redundancy:** Establishing comprehensive maintenance contracts and building in redundancy across all power systems to guarantee maximum uptime.

### 7. Risk Assessment & Mitigation

*   **Operational Risks:** Workforce management challenges (absenteeism, fatigue), equipment breakdown, raw material quality fluctuations, logistics disruptions, food safety compliance during continuous operation.
    *   *Mitigation:* Robust HR policies, preventative maintenance schedules, diversified raw material sourcing, real-time logistics tracking, strict HACCP implementation.
*   **Market Risks:** Changes in demand patterns, commodity price volatility, increased competition.
    *   *Mitigation:* Diversified off-taker base, market intelligence systems, flexible pricing models, continuous product innovation.
*   **Financial Risks:** Underestimation of capital costs, slower-than-expected revenue growth, currency fluctuations.
    *   *Mitigation:* Detailed financial modeling with sensitivity analysis, phased investment strategy, hedging strategies.
*   **Regulatory & Environmental Risks:** Compliance with labor laws for night shifts, environmental regulations for continuous emissions/waste.
    *   *Mitigation:* Dedicated legal and compliance team, investment in eco-friendly technologies, regular audits.

### 8. Conclusion

The GAB Climate Smart Investments Ltd 24-Hour Economy Project presents a compelling opportunity to revolutionize Ghana's agro-processing sector. By addressing critical operational limitations and leveraging the substantial demand in the Volta Corridor and Ashanti Region, GCSI can facilitate increased throughput, reduced post-harvest losses, and enhanced value creation. The proposed financial model, focusing on raw material sales commissions and advisory fees, is sustainable and scalable. With strategic capital investment in shift-workers compensation, logistics overheads, and robust backup power solutions (including climate-smart hybrid systems), GCSI will enable a more efficient, resilient, and economically vibrant agribusiness landscape in Ghana. This project aligns perfectly with national development goals for industrialization, food security, and climate action.

Benefits plan

## Benefits Realization Plan: GAB 24-Hour Economy Project

### 1. Project Context

The GAB 24-Hour Economy Project aims to significantly enhance agricultural productivity, market access, and operational efficiency within West African agribusiness by extending GAB's core operational hours. This initiative will unlock latent capacity, facilitate continuous processing and logistics, and provide enhanced support services to our network of farmers, aggregators, and buyers. The successful realization of this project's benefits is critical for GAB's strategic growth and its mandate to foster sustainable agricultural development.

### 2. Benefit Owners & Accountability

### 3. Detailed Benefits Realization Strategy

Each expected benefit is defined, linked to strategic objectives, and accompanied by a rigorous measurement framework.

#### 3.1. Benefit: Incremental Crop Volume Sales

*   **Definition:** The quantifiable increase in the aggregate tonnage of agricultural produce channeled through GAB's processing, storage, and market access infrastructure, directly attributable to the 24-hour operational model facilitating expanded capacity, reduced spoilage, and improved market linkages.
*   **Strategic Alignment:** This benefit directly contributes to GAB's objective of enhancing regional food security, increasing farmer income, reducing post-harvest losses, and capturing greater market share.
*   **Success Criteria:** Consistent, year-over-year growth in total supply volume that exceeds pre-project growth rates and accounts for seasonal variations, demonstrating the sustained impact of extended operations.
*   **Associated Metric (from Metric Registry):**
    *   **Supply volume (tons/month):** Total monthly tonnage of raw and processed agricultural produce received, stored, or dispatched through GAB facilities.
*   **Baseline Measurement Strategy:**
    *   Establish average monthly supply volume (tons) for the 24 months preceding the project launch. Data to be disaggregated by major crop types and origin regions to account for seasonality and localized factors.
*   **Target Setting Strategy:**
    *   Set incremental monthly and quarterly targets for supply volume growth (e.g., 5% month-over-month increase in the first 6 months, stabilizing to 3% thereafter, relative to the baseline and adjusted for seasonal indices). Targets will be informed by expanded processing capacity, market demand projections, and farmer outreach capabilities enabled by 24-hour operations.
*   **Measurement Frequency:** Monthly.
*   **Reporting Mechanism:** Monthly Operational Performance Dashboard, including trend analysis, variance against baseline and targets, and root cause analysis for significant deviations.

#### 3.2. Benefit: Secretariat Retainer Revenue

*   **Definition:** New or increased recurring income generated from providing extended administrative, coordination, and strategic advisory services to agribusiness clients (e.g., cooperatives, processors, exporters) who leverage GAB's continuous operational support, facilitated by the 24-hour economy project. This includes services such as continuous data management, market intelligence dissemination, logistics coordination, and compliance support.
*   **Strategic Alignment:** Diversifies GAB's revenue streams, enhances organizational resilience, strengthens strategic partnerships across the agribusiness value chain, and positions GAB as an indispensable regional agribusiness hub.
*   **Success Criteria:** A demonstrable increase in the number of active retainer agreements and/or an uplift in the average value of these agreements, leading to a sustained and growing stream of predictable revenue.
*   **Associated Metric (from Metric Registry):**
    *   **Consultancy billing hours:** Total monthly hours logged by GAB secretariat staff providing specific services under retainer agreements. This metric serves as a proxy for service utilization and value delivered to retainer clients.
*   **Baseline Measurement Strategy:**
    *   Document current monthly revenue from existing "secretariat-type" or advisory services (if any). If no such formal services exist prior to project initiation, the baseline is zero. Establish historical average consultancy hours (if applicable).
*   **Target Setting Strategy:**
    *   Set targets for the acquisition of new retainer agreements (e.g., 2 new agreements per quarter for the first year) and/or an average increase in the value of existing retainers. Correspondingly, set targets for total monthly consultancy billing hours to reflect the expanded service delivery capacity.
*   **Measurement Frequency:** Monthly.
*   **Reporting Mechanism:** Monthly Financial Performance Report, detailed client service utilization reports, and a client acquisition pipeline tracker.

#### 3.3. Benefit: Community Impact

*   **Definition:** The positive socio-economic changes realized within GAB's stakeholder communities, including the creation of new, sustainable employment opportunities, enhanced income generation for farmers, improved access to markets for local produce, and potential skills development, directly resulting from the implementation of the 24-hour operational model.
*   **Strategic Alignment:** Fulfills GAB's corporate social responsibility (CSR) objectives, fosters local economic development, strengthens community relations, and builds a resilient local talent pool.
*   **Success Criteria:** Documented increase in direct and indirect local employment, verifiable uplift in net farmer income attributable to GAB's expanded operations, and improved food security metrics within the operational catchment areas.
*   **Associated Metric (from Metric Registry):**
    *   **Active shifts:** The total number of unique operational shifts being run daily/weekly across GAB's facilities (e.g., processing, logistics, administrative) that are a direct consequence of the 24-hour operational model. This serves as a direct indicator of increased operational capacity and potential new job creation.
*   **Baseline Measurement Strategy:**
    *   Document the current number of operational shifts within GAB's facilities prior to the 24-hour project. Establish baseline employment figures within GAB's immediate operational vicinity, including full-time, part-time, and seasonal workers. Where feasible, gather average farmer income data for GAB-affiliated farmers.
*   **Target Setting Strategy:**
    *   Set targets for net new hires (e.g., 15-20 new full-time equivalent positions within the first year) directly attributable to the 24-hour operation. Track the sustained number of active shifts operating beyond the standard daytime hours (e.g., target 3 continuous 8-hour shifts daily in key operational areas).
*   **Measurement Frequency:** Quarterly (for employment figures and farmer income proxies), Monthly (for active shifts).
*   **Reporting Mechanism:** Quarterly Human Resources & Community Impact Report, detailing new hires, active shift schedules, and qualitative feedback from community engagement forums regarding socio-economic improvements.

### 4. Monitoring, Review, and Adjustment Process

*   **Reporting Cadence:**
    *   **Monthly:** Operational Leads will compile and present performance data for their respective benefits to the Project Steering Committee.
    *   **Quarterly:** The Project Steering Committee will review the overall Benefits Realization Plan, aggregating monthly reports and assessing progress against strategic targets. This review will include a deep dive into any variances.
    *   **Annually:** A comprehensive report will be submitted to GAB Executive Leadership, summarizing annual achievements, lessons learned, and recommendations for future iterations.
*   **Review & Adjustment:**
    *   Any significant deviation (e.g., >10% variance from target for two consecutive months) will trigger a root cause analysis by the relevant Benefit Owner.
    *   Mitigation strategies, which may include operational adjustments, target revisions, or project scope modifications, will be proposed to the Project Steering Committee for approval.
    *   The Benefits Realization Plan itself will be a living document, subject to quarterly review and refinement based on evolving project dynamics, market conditions, and observed outcomes.
*   **Risk Management:** Continuous identification and assessment of risks that could impede benefit realization (e.g., energy supply disruptions, market demand fluctuations, labor availability) will be integrated into the quarterly review process, with corresponding mitigation plans developed and tracked.

Charter basis

# Project Charter Basis: ECO24 - GAB's 24-Hour Economy Project

**Project Title:** ECO24 - GAB's 24-Hour Economy Project

**Project Manager:** [To be assigned]

**Project Sponsor:** GAB Executive Board

**Project Goal:** To strategically position GAB as a key player in the evolving 24-hour economy landscape by establishing foundational operational capabilities, particularly in supply chain management and compliance.

---

### 1. Project Objectives:

*   **Launch 24-hour supply contracts:** Successfully negotiate, sign, and implement at least two (2) pilot 24-hour supply contracts with selected partners, ensuring continuous availability and delivery of agricultural products/inputs.
*   **Register GAB:** Complete all necessary legal, operational, and regulatory registrations or compliance updates required for GAB to formally operate and engage in activities within a 24-hour economic framework.

---

### 2. Key Milestones:

The project will be executed in five distinct phases:

*   **Phase 1: Framing (Project Initiation & Planning)**
    *   Project Charter Approval
    *   Detailed Scope Definition & Requirements Gathering for 24hr operations and registration needs
    *   Stakeholder Identification & Engagement Plan
    *   Risk Assessment & Mitigation Strategy
    *   Baseline Project Plan & Schedule Completion (End Date: [e.g., Month X, Year Y])

*   **Phase 2: Development (Contract & Registration Preparation)**
    *   Drafting of 24hr Supply Contract Templates & Legal Review
    *   Identification & Vetting of Potential 24hr Supply Partners
    *   Analysis of GAB's Current Registration Status & Identification of Gaps for 24hr Operations
    *   Preparation of all required documentation for GAB's updated/new registration (End Date: [e.g., Month X+1, Year Y])

*   **Phase 3: Engagement & Negotiation**
    *   Initiation of formal engagement and negotiation with at least three (3) prospective 24hr supply partners.
    *   Submission of GAB's registration application/documentation to relevant authorities.
    *   Establishment of communication protocols for 24hr operations. (End Date: [e.g., Month X+2, Year Y])

*   **Phase 4: Finalization & Approval**
    *   Successful negotiation and signing of at least two (2) 24hr supply contracts.
    *   Securing internal GAB approvals for contracts and operational changes.
    *   Receipt of provisional or final approval/certificate for GAB's registration/compliance from relevant bodies. (End Date: [e.g., Month X+3, Year Y])

*   **Phase 5: Delivery & Handover (Pilot Launch & Project Closure)**
    *   Official launch and initial monitoring of 24hr supply contracts.
    *   Formal completion and communication of GAB's updated registration status.
    *   Handover of 24hr operational processes and contracts to GAB's permanent operations team.
    *   Project Closure Report & Lessons Learned. (End Date: [e.g., Month X+4, Year Y])

---

### 3. Budget Limits:

*   **Baseline Project Budget:** GHS 150,000
*   This baseline covers expenses related to legal fees (contract drafting, registration), consultancy for operational readiness, travel, communication, and administrative costs directly associated with achieving the stated objectives. Detailed budget allocation and contingency will be developed in Phase 1.

RecommendationNext action

Productization and commercial readiness

TypeData readinessLaunch readiness

Public URL

Target usersPaying customers

Free vs paid boundaryPricing hypothesisService delivery model

Success metricsGo / no-go criteria

Productization plan

# Productization Plan — ECO24

## Product / service mix
ECO24 is a portfolio of five revenue lanes built on GAB’s operational credibility as a climate-smart, 533-acre commercial farm:

1. **Supply contracts (Make 24)** — raw cassava roots to 24-hour agro-processors.
2. **Consultancy (Grow 24 + Connect 24)** — land governance, sustainability planning, supply-chain mapping, farm-system design, environmental compliance.
3. **Joint ventures / partnerships (Grow 24)** — land, technical expertise and market access in exchange for equity or revenue share.
4. **Carbon credit & MRV services** — agro-waste and blue-carbon MRV, leveraging existing GIS and Data Trust capability.
5. **Training & extension (Aspire 24)** — farm-based apprenticeship and climate-smart agriculture training.

## Target users and paying customers
- **Supply:** 24-hour-economy-registered agro-processors (Premium Foods, Amantin Agro Processing, GIHOC Distilleries, future Wumbei Park tenants).
- **Consultancy:** 24-Hour Economy Secretariat, regional task forces, Wumbei Park developers, development agencies.
- **JVs:** Agricultural investors seeking land + technical expertise under the 24H+ incentive framework.
- **Carbon/MRV:** Carbon developers, blue-carbon project proponents, agro-waste-to-energy developers.
- **Training:** Youth, agripreneurs, district assemblies, MoFA, Aspire 24 funders.

## Pricing hypothesis
- Supply: GHS 850–900/t farmgate, 14-day settlement for green-tier buyers.
- Consultancy: day-rate or fixed-scope, benchmarked against local sustainability consultancies.
- JV: equity/revenue share negotiated per project.
- Carbon/MRV: fee-for-service or verification-share.
- Training: per-participant or contract-funded.

## Data readiness
- **Available:** farm block data, buyer research, policy sources, capability statement, pitch deck.
- **Partial:** processor demand schedules, official registration status, exact incentive rates.
- **Missing:** binding offtake volumes beyond initial agreements, JV term sheets, carbon-project baselines.

## Launch readiness by lane
- Supply and consultancy: review-ready.
- JV and carbon: validation stage.
- Training: conceptual; requires Aspire 24 funding alignment.

## Go / no-go criteria
- **GO:** 2+ industrial supply agreements signed and first revenue logged.
- **HOLD:** policy incentives not operational by Q3 2026.
- **KILL:** no buyer commitment and no secretariat engagement by end of 2026.

PBA-EIRAS-DATATRUSTConvertedFullG6Internal PlatformLaunch: LiveData: Ready

Data Trust Backbone

Canonical source backbone for EIRAS: file corpus, source freshness, validation register, agency harvesters, data trust watchtower, and Drive/local mirroring discipline.

Owner: Daniel NsowahCategory: AIOPS / EIRASPriority score: 95

Open 6-domain workspace →

Analysis record

StatusStage gateGovernanceDecision

PriorityTailoringFitValueRisk

Needs assessment

PROBLEM: All EIRAS products depend on evidence quality, but before DataTrust, evidence was scattered across spreadsheets, documents, and team members' memories. Data sources were not tracked for freshness. Assumptions were not logged. Citations were not linked. When a client asked "where did this number come from?", the answer required manual search across multiple systems. When an EPA officer asked "what is the source for this claim?", credibility was at risk.

MARKET: Internal to GAB. Future external market: Ghanaian organisations (government agencies, NGOs, consultancies) need evidence management systems but lack technical capacity. The EPA's 12 offices, 780 staff manage evidence across environmental impact assessments, permits, and compliance records with no unified system. The Forestry Commission's 49 districts manage forest inventory data in disparate formats.

USERS: Primary — GAB AIOPS, Data Trust, GIS, and all EIRAS product teams. Secondary (future) — external organisations seeking evidence management capability.

GAP: Before DataTrust, GAB had 316 evidence records but no systematic quality control. Data sources (30) were not freshness-tracked. Citations (129) were not linked to outputs. Assumptions (87) were not validated. The result was occasional credibility gaps in client-facing outputs and time wasted on evidence retrieval.

Business caseBenefits planCharter basis

SCOPE — IN: Evidence record management; data source cataloguing and freshness tracking; assumption logging and validation; citation linking; HITL review workflow; quality assurance checks; shared output catalogue; integration with all EIRAS products.

SCOPE — OUT: Primary data collection (handled by GAB Farm, consultancy, or client); statistical analysis (handled by AIOPS); legal interpretation (handled by Regulatory); community data (handled by ConsentTrack with Data Protection compliance).

OBJECTIVES:
1. Maintain 100% evidence coverage for all client-facing outputs by December 2026.
2. Achieve <30 day source freshness for all active data sources.
3. Log and validate 100% of assumptions in client-facing outputs.
4. Link 100% of citations to their source records.
5. Build evidence management as a future consultancy product by Q4 2026.

SUCCESS CRITERIA: 100% evidence coverage; <30 day freshness; 100% assumption validation; 100% citation linking; zero unverified claims in outputs; 400+ evidence records by December 2026.

CONSTRAINTS: Must not claim primary data collection authority; must respect data protection for community data; must distinguish between verified evidence and assumptions; HITL review required for all client-facing conclusions; source freshness tracked automatically where possible.

SPONSOR: Daniel Nsowah, CEO, GAB Climate Smart Investments Ltd.

RecommendationNext action

Productization and commercial readiness

TypeData readinessLaunch readiness

Public URL

Target usersPaying customers

Free vs paid boundaryPricing hypothesisService delivery model

Success metricsGo / no-go criteria

Productization plan

PBA-EIRAS-EFFICIENCYConvertedLightG6Internal PlatformLaunch: LiveData: Ready

Efficiency Lite

Resource, emissions, and operations efficiency estimator that can support public education, AIOPS diagnostics, and operational-improvement conversations.

Owner: Daniel NsowahCategory: AIOPS / EIRASPriority score: 66

Open 6-domain workspace →

Analysis record

StatusStage gateGovernanceDecision

PriorityTailoringFitValueRisk

Needs assessment

PROBLEM: GAB operates as a portfolio of interconnected initiatives — 10 EIRAS products, AIOPS, Data Trust, GIS, Comms, Production, Portfolio, Projects, Operations — but lacked unified project management office (PMO) controls until late 2025. Tasks were tracked in Todoist but not linked to project baselines. Risks were discussed in meetings but not logged. Sustainability maturity was not measured. Cross-initiative dependencies were invisible.

MARKET: Internal to GAB. Future external market: Ghanaian SMEs and project developers need PMO setup support. Sustainability maturity assessment is increasingly required for ESG reporting, investor readiness, and donor compliance. No Ghana-specific sustainability maturity framework exists.

USERS: Primary — GAB internal teams. Secondary (future) — external clients seeking PMO setup or sustainability maturity assessment.

GAP: Before Efficiency/PMO, GAB had no unified project controls. Each initiative managed tasks ad-hoc. Risk registers did not exist. Scope creep was not baselined. Benefits were not tracked. The gap was particularly acute for EIRAS, which has 10 products at different maturity levels (G1–G5) with complex dependencies.

Business caseBenefits planCharter basis

SCOPE — IN: GAB OS task engine; PMO project controls (scope, schedule, cost, resource, quality, risk, stakeholder, communications, procurement, closure, benefits, operational transitions); sustainability maturity tracking; cross-initiative dependency mapping; Todoist-MCP integration; EIRAS product lifecycle management.

SCOPE — OUT: External client PMO services (future product, not current); statutory audit functions; financial accounting (handled by Finance & Legal).

OBJECTIVES:
1. Achieve 100% PMO control coverage across all GAB initiatives by December 2026.
2. Reduce unassigned task age to <7 days for all initiatives.
3. Maintain issue resolution SLA <48 hours.
4. Track sustainability maturity score monthly and improve by 20% by December 2026.
5. Enable external PMO-as-a-service by Q3 2026 if internal systems prove stable.

SUCCESS CRITERIA: 100% control coverage; <7 day task age; <48 hour issue resolution; 20% maturity improvement; all 10 EIRAS products have live PMO records; zero initiative governance failures.

CONSTRAINTS: Must not replace statutory audit or financial accounting; must integrate with Todoist (current task system); must be accessible to all GAB team members; must not create bureaucratic overhead that slows execution.

SPONSOR: Daniel Nsowah, CEO, GAB Climate Smart Investments Ltd.

RecommendationNext action

Productization and commercial readiness

TypeData readinessLaunch readiness

Public URL

Target usersPaying customers

Free vs paid boundaryPricing hypothesisService delivery model

Success metricsGo / no-go criteria

Productization plan

PBA-EIRAS-REGULATORYBusiness CaseFullG1HybridLaunch: PlanningData: Partial

Regulatory Lite

Preliminary Ghana regulatory, permitting, EIA, policy, legal-instrument and institutional routing support for site/project questions.

Owner: Daniel NsowahCategory: AIOPS / EIRASPriority score: 76Public URL

Open 6-domain workspace →

Analysis record

StatusStage gateGovernanceDecision

PriorityTailoringFitValueRisk

Needs assessment

PROBLEM: Ghana's environmental regulatory landscape is complex, overlapping, and rapidly changing. Act 1124 (2025) overhauled the EPA's authority. New LI 2504 tightened sector-specific conditions. Companies struggle to track which licences, permits, and compliance steps apply to their specific sector and location. ESG investors demand detailed regulatory compliance documentation that local companies often cannot produce. Multinationals entering Ghana need a tailored regulatory roadmap, not a generic list of laws.

MARKET: Ghana's regulatory environment spans EPA (12 offices, 780 staff), Forestry Commission (49 districts), Minerals Commission (4 regional + 13 district offices), Water Resources Commission (6+ regions), Lands Commission, and MMDA planning departments. The legal hierarchy runs: Act → LI → Guidelines → EI → District Bye-laws. Each sector has distinct approval pathways: mining (Act 703), forestry (Act 571), EPA (Act 1124), water (Act 522), petroleum (Act 919), energy (Act 832).

USERS: Free users — SMEs, project developers, researchers exploring regulatory requirements. Paid buyers — multinationals, investment funds, local companies seeking investor readiness, legal practitioners.

GAP: Before Regulatory Roadmap, regulatory guidance was either generic (lists of laws without sector specificity) or prohibitively expensive (GHS 200,000+ for international regulatory consultants). No digital tool mapped a specific sector/location combination to the exact regulatory pathway. No ESG compliance gap analysis existed for Ghanaian companies.

Business caseBenefits planCharter basis

SCOPE — IN: Sector readiness checklist; public Lite assessment; legal instrument browser; tailored regulatory roadmap; ESG compliance gap analysis; regulatory approval support; investor due diligence report; integration with DataTrust, SiteSense, and ReportBridge.

SCOPE — OUT: Legal representation before regulatory bodies; statutory permit applications (preparation only, not submission); lobbying; guarantee of approval.

OBJECTIVES:
1. Become Ghana's most trusted environmental regulatory intelligence provider by December 2026.
2. Deliver 5+ tailored regulatory roadmaps and 3+ ESG compliance gap analyses by December 2026.
3. Support 2+ regulatory approvals by December 2026.
4. Expand legal instrument corpus to 50+ regulatory instruments by December 2026.
5. Generate GHS 150,000–400,000 in Year 1 revenue from regulatory services.

SUCCESS CRITERIA: 5+ roadmaps; 3+ gap analyses; 2+ approvals; 15+ stored assessments; 50+ instruments; 90%+ client satisfaction; zero claims of guaranteed approval.

CONSTRAINTS: Must not guarantee regulatory approval; must cite instrument version and date; HITL review on all client-facing conclusions; source freshness quarterly minimum; must distinguish between preparation support and legal representation.

SPONSOR: Daniel Nsowah, CEO, GAB Climate Smart Investments Ltd.

RecommendationNext action

Productization and commercial readiness

TypeData readinessLaunch readiness

Public URL

Target usersPaying customers

Free vs paid boundaryPricing hypothesisService delivery model

Success metricsGo / no-go criteria

Productization plan

PBA-EIRAS-GREENCARBONApproved For ProjectFullG3HybridLaunch: Review ReadyData: Partial

Green Carbon / Forest Intelligence Lite

Forest, carbon, restoration, MRV and eligibility-readiness screener for early-stage green-carbon and forest-intelligence opportunities.

Owner: Daniel NsowahCategory: AIOPS / EIRASPriority score: 83Public URL

Open 6-domain workspace →

Analysis record

StatusStage gateGovernanceDecision

PriorityTailoringFitValueRisk

Needs assessment

PROBLEM: Ghana's forest carbon and restoration potential is substantial but poorly quantified. Carbon project developers struggle to produce credible baselines for VCS or Gold Standard certification. Forest restoration projects lack spatially explicit restoration pathway designs. The Forestry Commission's REDD+ Secretariat manages Ghana's REDD+ readiness but lacks digital MRV tools. Corporate buyers of carbon credits demand verified, human-audited baselines — not spreadsheet estimates.

MARKET: Ghana has 8.2 million ha of forest (declining). The REDD+ Secretariat is operational. The Carbon Markets Office was established under EPA Act 1124 Part V. Ghana signed a bilateral agreement with Norway for results-based payments. The Tree Crop Development Authority (TCDA) manages 16,000 ha across 16 regions. Key forest types: moist evergreen, moist semi-deciduous, dry semi-deciduous, savanna woodland, mangrove.

USERS: Free users — community forestry groups, researchers, government forest officers exploring readiness. Paid buyers — VCS/Gold Standard carbon developers, REDD+ project developers, forest restoration funds, bilateral donors, corporate carbon credit buyers.

GAP: Before Green Carbon, carbon baseline development relied on expensive international consultants (GHS 200,000–500,000 per project) or amateur spreadsheet estimates. No Ghana-specific forest carbon screening tool existed. No link between forest intelligence (Hansen, WorldCover, Sentinel-2) and carbon methodology requirements.

Business caseBenefits planCharter basis

SCOPE — IN: Forest readiness checklist; public Lite assessment; Hansen/WorldCover/Sentinel-2 GIS analysis; MRV-ready baseline report; carbon stock estimation; restoration pathway design; VCS/Gold Standard documentation support; integration with SiteSense, BioMainstream, and ReportBridge.

SCOPE — OUT: Carbon credit issuance (VCS/Gold Standard); forest carbon verification audit (third-party validator); statutory forest reserve boundary changes (FC); tree planting implementation (client or GAB Farm).

OBJECTIVES:
1. Become Ghana's most trusted forest carbon baseline provider by December 2026.
2. Deliver 5+ MRV-ready baseline reports and 3+ carbon stock estimates by December 2026.
3. Design 2+ restoration pathways by December 2026.
4. Support 1+ VCS/Gold Standard project registration with GAB documentation.
5. Generate GHS 200,000–500,000 in Year 1 revenue from forest carbon services.

SUCCESS CRITERIA: 5+ baselines; 3+ stock estimates; 2+ restoration designs; 1+ VCS registration; 20+ stored assessments; 90%+ client satisfaction; zero claims of carbon credit issuance authority.

CONSTRAINTS: Must not claim carbon credit issuance or verification authority; must cite methodology version and uncertainty ranges; HITL review on all client-facing conclusions; source freshness quarterly minimum; must respect FC reserve boundaries.

SPONSOR: Daniel Nsowah, CEO, GAB Climate Smart Investments Ltd.

RecommendationNext action

Productization and commercial readiness

TypeData readinessLaunch readiness

Public URL

Target usersPaying customers

Free vs paid boundaryPricing hypothesisService delivery model

Success metricsGo / no-go criteria

Productization plan

PBA-EIRAS-CONSENTTRACKBusiness CaseFullG1HybridLaunch: PlanningData: Partial

ConsentTrack Lite

FPIC, consultation readiness, community-evidence and consent-risk workflow. Sensitive module requiring strict human review, confidentiality, and public-claim controls.

Owner: Daniel NsowahCategory: AIOPS / EIRASPriority score: 78Public URL

Open 6-domain workspace →

Analysis record

StatusStage gateGovernanceDecision

PriorityTailoringFitValueRisk

Needs assessment

PROBLEM: Free, Prior, and Informed Consent (FPIC) is legally required for mining, forestry, and large infrastructure projects in Ghana, but documentation is often inadequate, contested, or absent. Community grievances arise years after project commencement because consent was poorly documented, not freely given, or not understood by community members. The Minerals Commission, Forestry Commission, and EPA all require FPIC evidence, but there is no standardised tool for tracking consent processes, stakeholder registers, meeting records, and grievances.

MARKET: Ghana's Constitution (Article 257(6), Article 270) recognises customary land tenure and chieftaincy authority. The Minerals and Mining Act 2006 (Act 703) requires community engagement. The Forestry Commission's CREMA guidelines require community consent. International standards (IFC Performance Standards, Equator Principles) require FPIC for project finance. Key institutions: CHRAJ (human rights), National House of Chiefs, Regional Houses of Chiefs, MMDAs.

USERS: Free users — community land committees, traditional authorities, NGO officers exploring FPIC requirements. Paid buyers — mining companies, infrastructure developers, carbon project developers, donor programmes, legal firms.

GAP: Before ConsentTrack, FPIC documentation relied on ad-hoc consultant reports, scattered meeting minutes, and verbal agreements. No systematic stakeholder register existed. No grievance tracking system linked to project records. Community contributions were not separated from verified evidence.

Business caseBenefits planCharter basis

SCOPE — IN: FPIC readiness checklist; public Lite assessment; legal instrument browser; community stakeholder register build; consent evidence pack; grievance tracking system; FPIC audit support; integration with ResourceMap and SiteSense.

SCOPE — OUT: Legal representation in land disputes; statutory FPIC certification (CHRAJ/EPA); community negotiation facilitation; land title verification (Lands Commission).

OBJECTIVES:
1. Become Ghana's most trusted FPIC documentation tool by December 2026.
2. Deliver 3+ community stakeholder registers and 2+ consent evidence packs by December 2026.
3. Support 1+ FPIC audit by December 2026.
4. Expand legal instrument corpus to 50+ FPIC-related instruments by December 2026.
5. Generate GHS 100,000–250,000 in Year 1 revenue from FPIC services.

SUCCESS CRITERIA: 3+ registers; 2+ evidence packs; 1+ audit; 10+ stored profiles; 50+ legal instruments; 90%+ client satisfaction; zero community data breaches; zero claims of statutory FPIC certification.

CONSTRAINTS: Must operate under Data Protection Act 843; community data restricted by default; traditional authority consent required for stakeholder mapping; must not claim statutory FPIC certification; HITL review on all client-facing conclusions; source freshness quarterly minimum.

SPONSOR: Daniel Nsowah, CEO, GAB Climate Smart Investments Ltd.

RecommendationNext action

Productization and commercial readiness

TypeData readinessLaunch readiness

Public URL

Target usersPaying customers

Free vs paid boundaryPricing hypothesisService delivery model

Success metricsGo / no-go criteria

Productization plan

PBA-EIRAS-BIOMAINSTREAMApproved For ProjectFullG3HybridLaunch: Review ReadyData: Partial

BioMainstream Lite

Biodiversity mainstreaming decision lens that helps site/project decisions consider biodiversity, water, forests, land-use change, policy triggers, and field-verification needs.

Owner: Daniel NsowahCategory: AIOPS / EIRASPriority score: 82Public URL

Open 6-domain workspace →

Analysis record

StatusStage gateGovernanceDecision

PriorityTailoringFitValueRisk

Needs assessment

PROBLEM: Biodiversity is systematically under-assessed in Ghanaian development planning. EIAs often treat biodiversity as a box-ticking exercise with generic species lists and no spatial specificity. Carbon projects, mining offsets, and agricultural expansions lack credible biodiversity baselines. Ghana's National Biodiversity Strategy (under CBD) is under-resourced and struggles to produce regular national reports.

MARKET: Ghana is party to the Convention on Biological Diversity (CBD) and committed to the Kunming-Montreal Global Biodiversity Framework (GBF). MESTI oversees the National Biodiversity Strategy. The Forestry Commission manages forest biodiversity. The EPA requires biodiversity chapters in EIAs under LI 2504. Key habitats include tropical moist forest (shrinking due to agriculture and logging), savanna woodland, coastal mangroves, and wetland Ramsar sites.

USERS: Free users — students, researchers, NGO officers exploring biodiversity triggers. Paid buyers — EIA consultants, carbon developers, mining companies, agricultural investors, donor programmes with biodiversity targets.

GAP: Before BioMainstream, biodiversity assessment relied on ad-hoc consultant expertise, outdated species lists, and fragmented habitat maps. No standardised trigger checklist existed for Ghanaian habitats. No systematic link between site characteristics and biodiversity questions.

Business caseBenefits planCharter basis

SCOPE — IN: Biodiversity trigger checklist for Ghanaian habitats; public Lite assessment tool; source pack of verified biodiversity data; analyst-reviewed screening reports; offset design support; CBD-aligned national report contributions; integration with SiteSense, ReportBridge, and Green Carbon.

SCOPE — OUT: Statutory biodiversity permits (EPA); species identification surveys (specialist biologists); protected area management plans (FC); final EIA biodiversity chapters (licensed EIA consultants).

OBJECTIVES:
1. Become Ghana's most trusted biodiversity screening tool by December 2026.
2. Deliver 10+ biodiversity screenings and 3+ offset design engagements by December 2026.
3. Contribute to 2+ CBD-aligned national reports by December 2026.
4. Expand source pack to 50+ verified biodiversity sources by December 2026.
5. Generate GHS 150,000–350,000 in Year 1 revenue from biodiversity services.

SUCCESS CRITERIA: 10+ screenings; 3+ offset designs; 2+ CBD contributions; 50+ sources; trigger coverage across all major Ghana habitats; 90%+ client satisfaction; zero claims of statutory biodiversity clearance.

CONSTRAINTS: Must not claim statutory biodiversity approval; must cite source classes and confidence; HITL review on all client-facing conclusions; source freshness quarterly minimum; must respect protected area boundaries under FC authority.

SPONSOR: Daniel Nsowah, CEO, GAB Climate Smart Investments Ltd.

RecommendationNext action

Productization and commercial readiness

TypeData readinessLaunch readiness

Public URL

Target usersPaying customers

Free vs paid boundaryPricing hypothesisService delivery model

Success metricsGo / no-go criteria

Productization plan

PBA-EIRAS-RESOURCEMAPConvertedStandardG6HybridLaunch: LiveData: Partial

ResourceMap Lite

Place-based actor, institution, governance, resource, and access intelligence layer for Ghana environmental and natural-resource decisions.

Owner: Daniel NsowahCategory: AIOPS / EIRASPriority score: 84Public URL

Open 6-domain workspace →

Analysis record

StatusStage gateGovernanceDecision

PriorityTailoringFitValueRisk

Needs assessment

PROBLEM: In Ghana, understanding who controls, manages, or disputes a piece of land or natural resource requires navigating a complex web of statutory agencies, traditional authorities, community committees, donor programmes, and private actors. There is no unified place-based intelligence tool that maps actors, institutions, governance pathways, and resource access issues for a specific location. Investors, developers, and community organisers routinely discover critical stakeholders too late — after conflicts have emerged or after regulatory submissions have been rejected.

MARKET: Ghana's land governance is dual: statutory (Lands Commission, EPA, FC, Minerals Commission) and customary (stools, skins, traditional councils, family lineages). Article 257(6) and Article 270 of the Constitution recognise chieftaincy and stool-land authority. The Local Governance Act 936 devolves environmental bye-law authority to 261 MMDAs. CREMAs (Community Resource Management Areas) add another governance layer. This complexity creates demand for place-based intelligence.

USERS: Free users — community land committees, traditional authorities, NGO field officers, local government planners, researchers. Paid buyers — donor programmes requiring community mapping, carbon developers needing stakeholder registers, mining companies assessing community context, consultancy clients.

GAP: Before ResourceMap, place-based intelligence was held in scattered documents, consultant reports, and institutional memory. No systematic actor database existed. The Ghana Institutional Atlas (900+ entities) provides a foundation but is not queryable by coordinate or place name.

Business caseBenefits planCharter basis

SCOPE — IN: Public actor/issue lookup; atlas seed data; basic place profiles; verified actor database build; community stakeholder registers; governance pathway mapping; custom place intelligence reports; integration with SiteSense and ConsentTrack.

SCOPE — OUT: Land title verification (Lands Commission); statutory boundary demarcation; community consultation facilitation (ConsentTrack); FPIC evidence gathering (ConsentTrack).

OBJECTIVES:
1. Build Ghana's most comprehensive place-based actor and governance database by December 2026.
2. Deliver 3+ custom actor databases and 2+ community stakeholder registers by December 2026.
3. Expand atlas seed from 162 to 500+ mappable places by December 2026.
4. Integrate actor data into SiteSense screening and ConsentTrack stakeholder registers.
5. Generate GHS 150,000–300,000 in Year 1 revenue from place intelligence services.

SUCCESS CRITERIA: 50+ verified actors; 30+ documented issues; 3+ custom databases; 2+ stakeholder registers; 500+ atlas places; 90%+ client satisfaction; zero community data breaches.

CONSTRAINTS: Must respect customary authority boundaries; community data restricted by default; traditional authority consent required for actor mapping in stool-land areas; source freshness quarterly minimum.

SPONSOR: Daniel Nsowah, CEO, GAB Climate Smart Investments Ltd.

RecommendationNext action

Productization and commercial readiness

TypeData readinessLaunch readiness

Public URL

Target usersPaying customers

Free vs paid boundaryPricing hypothesisService delivery model

Success metricsGo / no-go criteria

Productization plan

PBA-EIRAS-REPORTBRIDGEApproved For ProjectFullG3HybridLaunch: Review ReadyData: Partial

ReportBridge Lite

Evidence-to-reporting and MEA/template crosswalk product for Ghana environmental reporting, policy obligations, institutional reporting, and source-backed report drafting.

Owner: Daniel NsowahCategory: AIOPS / EIRASPriority score: 86Public URL

Open 6-domain workspace →

Analysis record

StatusStage gateGovernanceDecision

PriorityTailoringFitValueRisk

Needs assessment

PROBLEM: Ghanaian institutions, donor programmes, and corporates face a reporting burden across multiple Multilateral Environmental Agreements (MEAs) — UNFCCC, CBD, Ramsar, CITES, Paris Agreement — each with different templates, timelines, and focal points. There is no unified tool that maps a single dataset or evidence base across all required reporting formats. Reporting officers manually reformat the same underlying data 3–5 times per cycle, wasting 6–10 weeks per year.

MARKET: Ghana is party to 20+ MEAs. The EPA (under MESTI) is the national focal point for most. The Forestry Commission handles forest-related conventions. The Lands Commission handles land-degradation reporting. Donor programmes (USAID, AfDB, GIZ) require MEA-aligned M&E. Listed corporates (BoG sustainability principles, GSE expectations) must disclose environmental performance.

USERS: Free users — MEA students, NGO interns, reporting officers exploring templates. Paid buyers — donor programme M&E managers, corporate sustainability officers, EPA/FC/MLNR reporting units, carbon developers needing VCS/Gold Standard-aligned evidence packs.

GAP: Before ReportBridge, GAB and clients relied on manual spreadsheet crosswalks, static template PDFs, and institutional memory held by individual consultants. No automated mapping existed between EPA quarterly reports, donor M&E frameworks, and MEA obligation schedules.

Business caseBenefits planCharter basis

SCOPE — IN: MEA obligation mapping; template field crosswalks; sample crosswalks and public tools; custom crosswalk build service; HITL-reviewed report drafting; template adapter development; MEA document corpus management.

SCOPE — OUT: Statutory report submission (remains with mandated institution); data collection in the field (client or GAB Consultancy provides); legal interpretation of MEA obligations.

OBJECTIVES:
1. Become Ghana's most trusted MEA reporting alignment tool by December 2026.
2. Deliver 5+ custom crosswalks and 3+ reporting cycles by December 2026.
3. Secure 2+ annual MEA monitoring retainers by December 2026.
4. Expand template field maps from 51 to 100+ by December 2026.
5. Build MEA document corpus to 10+ official documents with verified source URLs.

SUCCESS CRITERIA: 5+ crosswalks delivered; 3+ reporting cycles; 2+ retainers; 100+ template fields; 10+ MEA documents; 90%+ client satisfaction; zero missed reporting deadlines for retainer clients.

CONSTRAINTS: Must cite official MEA texts and national focal point guidance; must not interpret obligations beyond published schedules; HITL review on all custom crosswalks; source freshness quarterly minimum.

SPONSOR: Daniel Nsowah, CEO, GAB Climate Smart Investments Ltd.

RecommendationNext action

Productization and commercial readiness

TypeData readinessLaunch readiness

Public URL

Target usersPaying customers

Free vs paid boundaryPricing hypothesisService delivery model

Success metricsGo / no-go criteria

Productization plan

PBA-EIRAS-SITESENSEConvertedStandardG6HybridLaunch: LiveData: Trusted

SiteSense Lite

Public site-risk screening intake and internal review workflow. It captures site details and polygons, applies preliminary spatial/source checks, and routes qualified leads to Consultancy or AIOPS.

Owner: Daniel NsowahCategory: AIOPS / EIRASPriority score: 90Public URL

Open 6-domain workspace →

Analysis record

StatusStage gateGovernanceDecision

PriorityTailoringFitValueRisk

Needs assessment

PROBLEM: Land-use decisions in Ghana are made with fragmented, outdated, or incomplete environmental intelligence. Developers, investors, conservation actors, and community leaders face no single entry point to understand what environmental risks, regulatory requirements, and institutional actors apply to a specific site. Information is scattered across EPA (12 regional offices, 780 staff), Forestry Commission (8 divisions, 49 district offices), Lands Commission (16 regional offices, ELIS platform), Minerals Commission (4 regional + 13 district offices), Water Resources Commission, and 261 MMDAs — each with its own records, formats, and access protocols. Manual EPA permit searches take 2–4 weeks and cost GHS 5,000–15,000 in consultant time.

MARKET: Ghana's environmental governance comprises 11 regulatory agencies under 6 ministries, 16 regions, and a constitutional duty (Article 36(9), 41(k)) to protect the environment. Act 1124 (EPA Act 2025) and LI 2504 (Environmental Assessment Regulations 2025) establish new EIA schedules, increasing demand for pre-screening. The EPA permit registry is not fully digitized or queryable by coordinate — a critical information gap.

USERS: Free-tool users include small-to-medium land developers, community land committees, NGO officers, and agricultural investors. Paid buyers include carbon project developers (VCS, Gold Standard), listed corporates under BoG/GSE sustainability disclosure, donor-funded programmes, and mining/energy majors requiring pre-concession due diligence.

GAP: Before SiteSense, GAB and clients relied on manual searches, ad-hoc GIS overlays (GHS 10,000–30,000 per site), and disconnected institutional knowledge. No systematic way combined public data, GAB field evidence, and client documents into unified site intelligence.

Business caseBenefits planCharter basis

SCOPE — IN: Public site-risk intake form; automated GIS screening against public datasets (WDPA, Ramsar, forest reserves, mining concessions, water bodies, administrative boundaries); evidence record creation and classification (6 source classes); internal review queue with risk banding; review-ready checklist and report draft generation; paid analyst-reviewed report delivery; lead conversion into AIOPS diagnostic or consultancy project.

SCOPE — OUT: Statutory permit issuance or regulatory approval; land title verification; final ESIA/EIA preparation; community consultation or FPIC facilitation; carbon credit issuance.

OBJECTIVES:
1. Become Ghana's most trusted free environmental site-screening tool by December 2026.
2. Convert 5% of free users into paid report or diagnostic inquiries.
3. Generate GHS 400,000–800,000 in Year 1 revenue from site-derived services.
4. Build a Data Trust corpus of 500+ evidence records by December 2026.
5. Establish GAB as preferred environmental intelligence partner for carbon developers, listed corporates, and donor programmes.

SUCCESS CRITERIA: 50+ submissions; 15+ paid reports; 3+ AIOPS diagnostics; 90%+ client satisfaction; zero regulatory/legal claims; all outputs carry disclaimers and source-class annotations.

CONSTRAINTS: Must operate within Data Protection Act 2012 (Act 843); must not expose sensitive community data publicly; HITL approval on all client-facing conclusions; source freshness quarterly minimum.

SPONSOR: Daniel Nsowah, CEO, GAB Climate Smart Investments Ltd.

RecommendationNext action

Productization and commercial readiness

TypeData readinessLaunch readiness

Public URL

Target usersPaying customers

Free vs paid boundaryPricing hypothesisService delivery model

Success metricsGo / no-go criteria

Productization plan

PBA-EIRAS-FAMILYBusiness CaseFullG1HybridLaunch: LiveData: Ready

EIRAS Product Family Commercialization

Portfolio parent for the EIRAS product family under AIOPS: public free tools, paid advisory services, internal review workflows, Data Trust, and human-in-the-loop environmental intelligence.

Owner: Daniel NsowahCategory: AIOPS / EIRASPriority score: 92Public URL

Open 6-domain workspace →

Analysis record

StatusStage gateGovernanceDecision

PriorityTailoringFitValueRisk

Needs assessment

PROBLEM: Ghana's environmental sector suffers from fragmented information, slow regulatory processes, and a lack of credible, technology-enabled advisory services. Environmental Impact Assessments take 6–12 months. Community consultations are often inadequate. Carbon project baselines are prohibitively expensive. Regulatory compliance is a maze of overlapping authorities. No single platform exists that integrates site intelligence, regulatory intelligence, biodiversity assessment, carbon estimation, community engagement, and reporting.

MARKET: Ghana's environmental economy spans EPA (12 offices, 780 staff), Forestry Commission (49 districts), Minerals Commission (17 offices), Water Resources Commission (6+ regions), Lands Commission, MMDAs (261), and traditional authorities (National and Regional Houses of Chiefs). Key laws: Constitution 1992 (Articles 36(9), 41(k), 257, 270), EPA Act 1124 (2025), LI 2504, Forestry Commission Act 571, Minerals and Mining Act 703, Water Resources Commission Act 522. Ghana's REDD+ programme, bilateral carbon agreements, and growing ESG investor interest create demand for integrated environmental intelligence.

USERS: Free users — researchers, students, community groups, government officers, NGOs. Paid buyers — project developers, carbon developers, mining companies, infrastructure developers, investment funds, corporate ESG teams, government agencies.

GAP: Before EIRAS, environmental advisory in Ghana was either generic (international consultants with limited local context) or fragmented (separate tools for GIS, reporting, carbon, etc.). No platform combined site intelligence (SiteSense), regulatory intelligence (Regulatory), biodiversity assessment (BioMainstream), carbon estimation (GreenCarbon), community engagement (ConsentTrack), actor mapping (ResourceMap), evidence management (DataTrust), and project management (Efficiency) into a unified system with human-in-the-loop verification.

Business case

INVESTMENT TO DATE: GHS 850,000 across all EIRAS products (SiteSense GHS 245K, ReportBridge GHS 120K, ResourceMap GHS 90K, BioMainstream GHS 105K, GreenCarbon GHS 90K, ConsentTrack GHS 45K, Regulatory GHS 55K, Efficiency GHS 180K, DataTrust GHS 85K, Family/overhead GHS 75K).

REVENUE MODEL: Umbrella target GHS 1.2–2.5M Year 1. Individual product contributions: SiteSense GHS 400K–800K, ReportBridge GHS 150K–350K, ResourceMap GHS 200K–450K, BioMainstream GHS 150K–300K, GreenCarbon GHS 200K–500K, ConsentTrack GHS 100K–250K, Regulatory GHS 150K–400K. Bundles and retainers additional.

MARGINS: Portfolio-level margin target 50–60%. Individual product margins range from 45% (complex services) to 70% (automated tools).

STRATEGIC VALUE: EIRAS positions GAB as Ghana's leading environmental technology and advisory company. The integrated platform creates network effects: each product strengthens the others. DataTrust evidence benefits all products. SiteSense polygons feed GreenCarbon and BioMainstream. ReportBridge outputs leverage all products. This integration is the competitive moat that no single-product competitor can match.

Benefits plan

QUANTIFIED BENEFITS:
• Revenue: GHS 1.2–2.5M Year 1 across all products.
• Cost reduction: Integrated platform reduces client advisory costs by 40–60% vs. fragmented consultants.
• Time savings: EIA timeline reduced from 6–12 months to 3–6 months with SiteSense + Regulatory + ReportBridge.
• Carbon revenue acceleration: GreenCarbon baselines accelerate VCS registration by 6–12 months.
• Community protection: ConsentTrack FPIC documentation prevents GHS 500K+ in project delays and disputes.

STRATEGIC BENEFITS:
• Market leadership: GAB becomes the definitive environmental intelligence platform for Ghana.
• Institutional relationships: Deep engagement with EPA, FC, Minerals Commission, WRC, and traditional authorities.
• Data asset: 316 evidence records growing to 500+ by Year 2 creates invaluable institutional knowledge.
• Talent attraction: Professional PMO and technology platform attracts top Ghanaian environmental talent.
• International credibility: Evidence-backed, HITL-verified outputs meet international ESG and carbon standards.

Charter basis

SCOPE — IN: All 10 EIRAS products (SiteSense, ReportBridge, ResourceMap, BioMainstream, GreenCarbon, ConsentTrack, Regulatory, Efficiency, DataTrust, Family); unified landing page; cross-product bundles; enterprise advisory retainers; shared evidence base; integrated project management; human-in-the-loop verification.

SCOPE — OUT: Statutory regulatory functions (EPA, FC, Minerals Commission retain sole authority); primary data collection beyond GAB's operational capacity; legal representation; guarantee of regulatory approval or carbon credit issuance.

OBJECTIVES:
1. Become Ghana's leading environmental technology and advisory platform by December 2026.
2. Achieve GHS 1.2–2.5M Year 1 revenue across all EIRAS products.
3. Deliver 50+ stored assessments across products by December 2026.
4. Maintain 90%+ client satisfaction and zero credibility incidents.
5. Build institutional relationships with EPA, FC, Minerals Commission, WRC, and traditional authorities.
6. Grow evidence base to 500+ records by December 2027.

SUCCESS CRITERIA: 10 live products with PMO controls; GHS 1.2–2.5M revenue; 50+ assessments; 90%+ satisfaction; zero credibility incidents; 6+ institutional partnerships; 500+ evidence records by Year 2.

CONSTRAINTS: Must not claim statutory authority; must respect Data Protection Act 843; must distinguish between advisory and regulatory functions; must maintain HITL review on all client-facing outputs; must ensure source freshness; must manage cross-product dependencies.

SPONSOR: Daniel Nsowah, CEO, GAB Climate Smart Investments Ltd.

RecommendationNext action

Productization and commercial readiness

TypeData readinessLaunch readiness

Public URL

Target usersPaying customers

Free vs paid boundaryPricing hypothesisService delivery model

Success metricsGo / no-go criteria

Productization plan

c334295cNeeds AssessmentFullG1

Procurement Consortium Platform (BidBridge)

Opportunity: Ghana government and development-partner procurement is increasingly cross-cutting and multidisciplinary, and access is gated by networks. Small and local firms (and individual consultants) cannot qualify or win alone and waste money buying bid documents for tenders they are under 5% likely to win. Proposal: a neutral platform that aggregates procurement signals (leveraging GAB's existing opportunity listener), screens true eligibility/fit so firms stop wasting money on hopeless bids, and assembles qualifying consortia so excluded firms can compete and win. Mission: grow local/small firms by breaking the experience catch-22 (need experience to win, need to win to get experience). GAB is the neutral convener and a complementary participant (Amazon model) — it provides information and matching, not insider advantage.

Owner: Daniel NsowahCategory: ConsultancyPriority score: Not scoredLatest gate: G0 go Drive doc

Open 6-domain workspace →

Analysis record

StatusStage gateGovernanceDecision

PriorityTailoringFitValueRisk

Needs assessment

## 1. Situation Statement
Capable Ghanaian small/local firms and individual consultants are structurally excluded from government and development-partner contracts. Tenders are increasingly multidisciplinary; eligibility thresholds (turnover, similar-experience, certifications, key personnel) exceed what a single small firm can meet; and access is mediated by networks. Firms also spend scarce cash on bid documents for tenders they have little realistic chance of winning. Result: public/donor money concentrates among a few well-connected incumbents, and small firms cannot build the track record that would let them grow.

## 2. Business Need
GAB - itself one of the excluded firms - needs (a) a credible, mission-aligned new service line that extends its existing opportunity-intelligence capability, and (b) a way to advance its inclusion/local-growth mission while strengthening its network position in environment/climate/agri/NRM procurement. The broader need served: a trusted mechanism that gets excluded firms considered for and INTO contracts they cannot win alone.

## 3. Current State
- Opportunities are published across fragmented channels (GHANEPS/PPA, MDA sites, donor portals: World Bank, AfDB, EU, GIZ) and are hard to monitor comprehensively. GAB already runs an opportunity listener that partially addresses this.
- Small firms learn of tenders late, lack the discipline to assess true eligibility, and waste money on bid docs for unwinnable tenders.
- Consortium/JV formation happens informally through personal networks, slowly, gated by trust and lead-partner relationships - no neutral mechanism exists.
- Local-content rules increasingly oblige large/foreign primes to subcontract local firms, but primes struggle to find and vet credible partners (specific obligations by sector to be verified).
- The experience catch-22 keeps capable firms out: no reference experience -> no win -> no reference experience.

## 4. Future State (desired outcome)
- Excluded firms/consultants are reliably surfaced for relevant, WINNABLE opportunities and assembled into qualifying consortia.
- Firms stop wasting money on hopeless bids (a fit signal replaces guesswork).
- Each won consortium bid becomes a credential rung; firms accumulate the experience/references that compound into independent qualification.
- GAB is recognised as the neutral, trusted convener of the local-firm consortium ecosystem in its sectors.
- Success signals to target in the Business Case: # firms registered with VERIFIED capability; # consortia formed; # bids submitted; win rate; value won by small firms; firms that graduate to qualify independently.

## 5. Capability Gap (current -> future)
Does not exist today and must be built or sourced:
- A structured, VERIFIED capability registry of firms/consultants (the hard, trust-dependent asset).
- Eligibility/fit screening that parses a tender's mandatory eligibility + evaluation criteria and scores a firm/consortium - reusing GAB's EIRAS requirement-extraction muscle.
- A consortium-matching layer that proposes teams collectively meeting thresholds.
- Neutral governance (rules preventing insider advantage; disclosure of GAB participation).
- A trust/relationship mechanism for consortium formation (legal/JV scaffolding, lead-partner clarity) - mostly human, not software.
- Possibly access-to-finance/bonding partnerships IF finance (not matchmaking) proves the binding constraint.

## 6. Options
- Option A - Do nothing. GAB keeps bidding alone; small firms stay excluded. Low cost, mission unmet.
- Option B - Tender-alert SaaS (horizontal). Commoditised, low SME willingness-to-pay, ignores the real (consortium) gap. Weak.
- Option C - Vertical consortium-orchestration brokerage (RECOMMENDED). Start in GAB's authority sectors; fit-screen + assemble consortia as a high-touch brokered service; paying customers = primes needing vetted local partners (local content) + donor SME-growth funding; free for small firms. Software (listener + EIRAS screening) is the engine, not the product.
- Option D - Pure self-serve software marketplace, scale-first. High cold-start + build cost; trust gap unaddressed. Premature.

## 7. Recommendation (to test in the Business Case)
Pursue Option C: a vertical, brokerage-first consortium orchestration in GAB's authority sectors, with the engine built incrementally from the existing listener + EIRAS screening. GAB stays neutral convener; participates only as a complementary partner with disclosure; never uses registry data for insider advantage. Monetise via primes (local-content vetted-partner pipeline) + donor funding; free to small firms. Validate the binding constraint (matchmaking vs finance/bonding) before scaling.

## 8. Key Assumptions To Verify (carry into Business Case - currently UNVERIFIED)
1. Local-content rules materially oblige primes to use local partners in GAB's sectors, and primes will pay for a vetted pipeline.
2. The binding constraint for small firms is matchmaking/eligibility, not primarily finance/bonding.
3. Procurement feeds (GHANEPS/PPA/donor) are legally usable for aggregation.
4. A credible first cohort of firms can be sourced via an association/donor partner.
5. Donor SME-growth funding is realistically accessible for the inclusion side.
6. GAB can hold neutrality credibly while participating (disclosure model acceptable to firms).

Business caseBenefits planCharter basisRecommendationNext action

Productization and commercial readiness

TypeData readinessLaunch readiness

Public URL

Target usersPaying customers

Free vs paid boundaryPricing hypothesisService delivery model

Success metricsGo / no-go criteria

Productization plan

Business analysis process groups

Defining & Aligning

Initiating

Planning

Executing

Monitoring & Controlling

Releasing

Gate path

Decision queue

Product and service readiness

Operations → Portfolio & BA

Capture operational signal

Final telemetry audit: knowledge closeout availability and desktop surface routing gaps

In-app closeout counters ignore tool traces and knowledge receipts can be skipped

Watchdog falsely degrades on legacy Claude session-wide stale records

Successful Portfolio read mislabelled denied by telemetry content matching

Pre-build gate accepts any text as project-task evidence

Farm Explorer Live project record is stale

Scheduled agent runs cannot be confirmed through governed tools

Record protection remainder from the knowledge-layer initiative

Parked ECO24 Cluster 3: institutional and policy channels (24hr economy) — 4 items

Parked commercial and finance decisions from closed projects (6 items)

EIRAS product backlog carried from closed product projects (17 items)

Comms backlog parked for the comms work (3 items from closed projects)

Company knowledge discovery — agents can retrieve known sources but cannot discover what GAB already knows

Chat approval path for queued approvals (decide_agent_action) — operational fix

A review that states no correction counts is arithmetically identical to one confirming zero — and the bias runs in the flattering direction

The 1 October cross-agent measurement cannot reach decision-readiness — the mechanism is fixed, the sample is not recoverable in 10 days

query_pm_tasks ignores its task_ids and search arguments — targeted lookup is silently impossible

36 React Compiler lint findings carried as warnings after the Next 16 upgrade — setState in effects across 39 files

react-leaflet throws "Map container is already initialized" in DEV under React 19 — map pages unusable with npm run dev

Next.js 14 → 16 migration required: the only fix for two critical advisories, and CI carries a dated exception until it happens

Efficiency Lite renders GHG tonnage with a bare toLocaleString() — the figure changes meaning with the reader's browser locale

FloodSense Lite result contradicts itself on wetland distance, and formats population as a decimal

No governed way to mark an eiras_sites row as a test — is_test is only ever set by the paid-review lead notifier

SiteSense Lite: district is required by the API but not by the form — users get a bare "Validation failed"

Next.js critical security advisories block the migration pipeline — major upgrade required

New initiative intake

Analysis record

Gate decision

Analysis record

Gate decision

Analysis record

Gate decision

Analysis record

Gate decision

Analysis record

Gate decision

Analysis record

Gate decision

Analysis record

Gate decision

Analysis record

Gate decision

Analysis record

Gate decision

Analysis record

Gate decision

Analysis record

Gate decision

Analysis record

Gate decision

Analysis record

Gate decision

Analysis record

Gate decision

Analysis record

Gate decision

Analysis record

Gate decision

Analysis record

Gate decision

Analysis record

Gate decision

Analysis record

Gate decision

Analysis record

Gate decision

Analysis record

Gate decision