An interactive review and implementation workspace for the Open Intelligence Architecture — a nine-layer reference architecture for enterprise intelligent systems.
Live · oia.agentics.org · Reader's Digest · Decision Log · ADR-0001
The OIA Model — Open Intelligence Architecture — is the Agentics Foundation's reference architecture for intelligent systems at enterprise scale. It defines nine layers (physical compute at the bottom, the human and browser interface at the top), two state-holding layers that bracket the operational core, and six horizontal cross-layer spans that cannot be localised to any one layer.
This repository contains the OIA Model Reader: a single-page application that turns the two published digests (the Reader's Digest and the Decision Log Digest) into a practical, interactive system — allowing reviewers, architects, vendors, regulators, and researchers to use the model rather than just read it.
Every section of the source documents is preserved verbatim. Every section also carries a tied-in operator tool that produces a real Markdown artefact (brief, spec, memo, plan) you can paste straight into your operations.
- Masthead and foreword in their original institutional + personal registers
- §1 Introduction through §8 Deferrals, the Decision Log Digest, and the Closing — every paragraph preserved
- Six figure-quality SVG visualisations positioned inline with the prose they illuminate
| Tool | What it produces |
|---|---|
| Workspace (hub) | Live KPIs + single-file Markdown Review Package + JSON bundle for team sharing |
| Stack Diagram | Interactive 10-layer × 6-span canvas with Assessment overlay |
| Architecture Assessment | Rate each layer COHERENT / GAP / NOT YET / N/A — live span coherence scores |
| Provider Footprint | 33-vendor × 10-layer matrix; flags "vertical extension" (Decision 10) |
| Implementation Roadmap | Per-layer checklists, Markdown plan export |
| Explorer | Search 55 concerns, 91 reference technologies, 42 open questions |
| Decision Review Register | Per-decision vote (REOPEN / HOLD / CONFIRM) + Markdown export |
| Feedback | Form backed by SQLite, five Foundation-featured questions + eleven review prompts |
Each prose section has a tied-in micro-tool that turns the content into a business deliverable:
| Section | Operator tool | Artefact |
|---|---|---|
| On Reading | Reading Plan Generator | Team-channel reading plan by role × depth |
| Foreword | Era Transition Brief | Exec-level "we are at this moment" memo |
| §1 Introduction | System Scope Generator | Project-charter scope statement |
| §2.1 Lineage | Governance Overlay | Framework-stack mapping to OIA |
| §2.2 Context | Urgency Brief | Board-level urgency memo |
| §2.3 Properties | Layer Priority Plan | Ranked investment plan |
| §3 Nine Layers | Layer Scope Memo | RACI-style owned-layers memo |
| §5 Spans | Span Gap Memo | CISO / audit gap memo |
| §6 Relationships | Stack Spec | Governance stack spec for ARB / RFP |
| §7 Adoption | Adoption Playbook | Four-step program kickoff |
| §8 Deferrals | Community Input | Submission to the Foundation |
| Decision Log | Reviewer Brief | Stance-tailored review protocol |
All tools persist state to localStorage; every exported artefact is plain Markdown.
┌──────────────────────────────────────────────────────────────────────┐
│ Client (oia.agentics.org) │
│ ┌──────────────────────────────────────────────────────────────┐ │
│ │ Single-page app — React 18 + TypeScript + TailwindCSS │ │
│ │ Telemetry-dashboard design system (palette-locked) │ │
│ └──────────────────────────────────────────────────────────────┘ │
└───────────────────────┬──────────────────────────────────────────────┘
│
HTTPS (SNI oia.agentics.org · CORS allow-list)
│
┌───────────────────────▼──────────────────────────────────────────────┐
│ Static hosting — GitHub Pages (oia.agentics.org) │
│ │
│ Cross-origin fetch ↓ /api/* │
│ │
│ Cloud Run service oia-model-reader (us-central1, Gen 2) │
│ Node 20 + Express 5 + better-sqlite3 │
│ │
│ SQLite volume mount: │
│ gs://agentics-oia-model-data → /data/feedback.db │
└──────────────────────────────────────────────────────────────────────┘
Design references
- Telemetry UI aesthetic homage — see
animated-telemetry-dashboard-section (1).tsx - Palette:
#000000/#FFFFFF/#F05122(Agentics orange) — strict - All visual motion gated behind
IntersectionObserverand disabled underprefers-reduced-motion
- Node.js ≥ 20
- npm ≥ 10
- For local feedback DB: no setup; the API writes to
data/feedback.dbon first run - For containerised builds: Docker ≥ 20.10
git clone https://github.com/agenticsorg/OIA-Model
cd OIA-Model
npm install
npm run devThe SPA and feedback API run together on http://localhost:5173 via a Vite middleware plugin (server/vite-api-plugin.ts).
npm run build # type-check + Vite build → dist/
npm run server:build # compile server TS → server-dist/
npm run server:start # run dist/ + API on PORT=8080docker build --platform=linux/amd64 -t oia-model-reader .
docker run --rm -p 8080:8080 \
-v $(pwd)/data:/data \
-e DB_PATH=/data/feedback.db \
oia-model-readerProduction runs on Google Cloud Run (Gen 2) in project agenticsorg, region us-central1.
| Component | Resource |
|---|---|
| Container | us-central1-docker.pkg.dev/agenticsorg/gcf-artifacts/oia-model-reader |
| Service | Cloud Run oia-model-reader, --execution-environment=gen2, allow-unauthenticated |
| Storage | gs://agentics-oia-model-data mounted at /data for SQLite persistence |
| Static | GitHub Pages (gh-pages branch) serving the SPA on oia.agentics.org |
| DNS | Cloudflare CNAME oia.agentics.org → agenticsorg.github.io (DNS-only) |
| API CORS | Allow-list: oia.agentics.org, agenticsorg.github.io, localhost |
# 1. Build and push new image
TAG="v$(date +%Y%m%d-%H%M)"
IMG="us-central1-docker.pkg.dev/agenticsorg/gcf-artifacts/oia-model-reader:$TAG"
docker build --platform=linux/amd64 -t "$IMG".
docker push "$IMG"# 2. Deploy
gcloud run deploy oia-model-reader --image="$IMG" \
--region=us-central1 --project=agenticsorgexport VITE_API_BASE="https://oia-model-reader-2slhzrawja-uc.a.run.app"
npm run build
echo oia.agentics.org > dist/CNAME
cp dist/index.html dist/404.html
# Push dist/ contents to the gh-pages branchAgentics Foundation members submit structured feedback via the Feedback section. Submissions are stored in SQLite (data/feedback.db locally, /data/feedback.db on Cloud Run, backed by GCS).
- Identity — optional name / email / organisation / role;
share_nametoggle redacts from public listing - Foundation-featured prompts — hyperscaler monopolies, L8 certification, legacy vendor mapping, Foundation funding, regulatory alignment
- Review prompts drawn from the digests — shape assessment, decisions to reopen + reasons, open-question contribution, missing concerns / tech / spans, vertical-extension experience, improvements, general notes
# Human-readable
npm run feedback:list
# Every field, every row
npm run feedback:full
# Arbitrary SELECT (SELECT only; mutations refused)
npm run feedback:query -- "SELECT id, role, substr(shape_response, 1, 120) FROM feedback"# Sync production DB down for local inspection
gcloud storage cp gs://agentics-oia-model-data/feedback.db ./data/feedback.dbOIA-Model/
├── src/
│ ├── sections/ 14 top-level sections + Workspace + Feedback
│ ├── components/
│ │ ├── viz/ Six SVG figures inline with the digest prose
│ │ └── micro/ 13 operator tools (Markdown-artefact generators)
│ ├── content/ Verbatim digest text · layers · spans · refs · decisions
│ ├── lib/ Workspace aggregation + export
│ └── styles/ Telemetry design tokens
├── server/ Production Express app + Vite dev middleware + SQLite
├── scripts/ Feedback CLI (list / full / query)
├── docs/ Source digests (v0.1)
└── plans/ ADR-0001 (the architecture decision record)
The OIA Model is open for community review. Two paths:
- Through the UI — use the Feedback section on oia.agentics.org. All five Foundation-featured questions and eleven review prompts are structured for machine query.
- Through Git — open a pull request against
main. Please follow the palette, verbatim-source, and section-structure conventions inplans/ADR-0001-oia-model-reader-ui.md.
For substantive architectural disagreement, cite a specific decision number in the form reopen decision NN, because … — as the Decision Log Digest itself requests.
- The feedback API enforces a CORS allow-list (see
server/handlers.ts). - Public listings of feedback redact name, email, and organisation for any submission where the reviewer left share_name unchecked. The
listFeedbackPublic()view enforces this at the SQL level. - No secrets are stored in the repository. Cloudflare and Google credentials live in Google Secret Manager under the
agenticsorgproject. - Cloud Run is public (
--allow-unauthenticated) — please do not submit confidential material. Only review-relevant content should be posted.
Released under the MIT License. See LICENSE.
Copyright © 2026 Agentics Foundation
- Reuven Cohen — Founder, Agentics Foundation, author of the OIA Model Foreword.
- Source documents — the OIA Model v0.1 Reader's Digest and the Decision Log Digest are the normative texts this UI renders.
- Frameworks the model builds on — ISO/IEC 7498 (OSI), NIST Cybersecurity Framework, MITRE ATT&CK / ATLAS, NIST AI Risk Management Framework, OWASP Top 10 for LLM Applications, ISO/IEC 42001, Model Context Protocol.