You have OKRs. Dashboards. Risk registers. Compliance checklists.
A beautiful deck from Q1 that everyone already forgot.
But nobody tracks what must be TRUE for any of it to work.
"We'll win mid-market on simplicity."
Based on what belief?
"EU customers will pay a sovereignty premium."
When did we last validate that? Who owns it?
Six months later the bet fails — and nobody can name the assumption that broke.
That's not bad execution. That's flying blind on beliefs you never wrote down.
Stratum is the layer corporations never built.
A living system of record for strategic assumptions, dependencies, and decision-ground-truth — the connective tissue between quarterly planning and what your team actually does on Tuesday.
- Assumption graph — what depends on what
- Drift radar — beliefs going stale before they break you
- Signal feed — evidence that contradicts what you still "know"
- Alignment matrix — where teams secretly disagree
- Board brief — assumption health, not vanity metrics
Clone it. Plug your API key. 60 seconds. Free. Local.
- ✅ Strategy deck from January → nobody can list the 5 beliefs it depends on
- ✅ Product ships X, Sales promises Y → same company, different reality
- ✅ Post-mortem: "we assumed churn was about price" → never validated, never logged
- ✅ Board asks "what's our biggest strategic risk?" → you list KPIs, not assumptions
- ✅ You have GRC, OKRs, ERP → no system of record for what must be true to win
Nodded twice? Your house has mess. Stratum is the flashlight.
| Tool | Version | Check |
|---|---|---|
| Python | 3.10+ | python --version |
| Node.js | 18+ (for npm scripts) | node -v |
git clone https://github.com/Sunradiance/stratum.git
cd stratum
cp .env.example .env
# Edit .env — add your 3 API keys (see below)npm run setup # installs Python deps
npm run dev # starts Stratum| # | Key | What it powers | Get a key |
|---|---|---|---|
| 1 | LLM_API_KEY |
AI analysis, premortems, blind spots, brief enhancement, assumption extraction | Groq (recommended) |
| 2 | TAVILY_API_KEY |
Live web signal scanning against your assumptions | Tavily |
| 3 | SERPER_API_KEY |
Supplementary Google search for contradicting evidence | Serper |
Minimum to start: LLM_API_KEY alone enables all AI features. Add Tavily + Serper for web signal scanning.
LLM_API_KEY=gsk_...
LLM_BASE_URL=https://api.groq.com/openai/v1
LLM_MODEL_NAME=qwen/qwen3-32b
TAVILY_API_KEY=tvly-...
SERPER_API_KEY=...Any OpenAI-compatible LLM works — change LLM_BASE_URL and LLM_MODEL_NAME for OpenAI, Together, DeepSeek, etc.
- Command Center — org assumption health score, drift radar, cascade risk
- Research Thesis — synthesis of enterprise research on the strategy-execution gap
- Strategy Pillars — link bets to the assumptions they stand on
- Assumption Graph — dependency + contradiction visualization
- Signal Feed — contradicting evidence (manual + AI web scan)
- Decision Trace — major calls linked to assumptions
- Premortem Lab — AI-generated failure scenarios
- Alignment Matrix — cross-team belief disagreements
- Board Brief — exportable markdown + AI enhancement
All data is local-first (IndexedDB in your browser). API keys stay on your machine in .env — never sent to a Stratum server.
stratum/
├── .env.example # Copy to .env
├── index.html # Frontend SPA
├── css/ js/ # UI
├── backend/
│ ├── run.py # Start here
│ ├── requirements.txt
│ └── app/
│ ├── api/ # REST endpoints
│ └── services/ # LLM + search
└── package.json # npm run dev
| Method | Path | Description |
|---|---|---|
| GET | /api/health |
Check which APIs are configured |
| POST | /api/ai/premortem |
Generate premortem scenario |
| POST | /api/ai/blind-spot |
Generate blind-spot question |
| POST | /api/ai/analyze |
Analyze org assumption health |
| POST | /api/ai/enhance-brief |
AI-enhance board brief |
| POST | /api/ai/extract-assumptions |
Extract assumptions from strategy text |
| POST | /api/signals/scan |
Web scan for contradicting signals |
| App | Port | Layer |
|---|---|---|
| Stratum | 8791 | What must be true to win |
| Keepline | 8792 | What you're committed to pay |
| Whyline | 8793 | What was decided and why |
MIT — use it, fork it, stop flying blind on beliefs.
Stratum — illuminate what you depend on but never tracked.