End-to-end platform for investment committee workflow automation:
- deal intake
- due diligence execution
- IC approval and voting
- post-investment monitoring
- compliance-grade audit trail
Send this repo to IT; they can run docker compose up --build and give you a URL.
Recommended handoff steps:
- IT copies
.env.firm.exampleto.envand fills internal infrastructure values. - IT runs
docker compose up --build. - IT shares frontend URL (default
http://localhost:5173) with your deal team.
cp .env.example .env
docker compose up --build
python scripts/seed_demo.pyOpen:
- Frontend:
http://localhost:5173 - API docs:
http://localhost:8000/docs
Walkthrough script: docs/Demo script.md
- Document intelligence (OCR + parsing)
- Extraction engine (heuristic / Ollama / Hugging Face)
- Workflow orchestration (state transitions + task generation)
- Collaboration UI (React command center)
- Compliance data layer (
audit_logs,monitoring_alerts)
| Mode | Cost | Typical latency | Setup effort | Typical extraction accuracy |
|---|---|---|---|---|
| Heuristic | Free | Fastest | Minimal | Baseline, deterministic |
| Ollama | Free (local compute) | Medium | Medium (install model runtime) | Better on nuanced text |
| Hugging Face free tier | Free tier limits | Medium-high | Low-medium (token + model) | Good when model prompt follows schema |
Default mode in .env.example is AI_PROVIDER=heuristic.
- Install Ollama on internal host.
- Pull a local model (example):
ollama pull llama3.1:8b - Set:
AI_PROVIDER=ollama
AI_MODEL=llama3.1:8b
OLLAMA_BASE_URL=http://<internal-host>:11434- Restart API/Celery services.
Extraction mode examples and tests: docs/extraction-modes.md
Built-in demo users endpoint: GET /api/v1/demo/users
Default demo users:
- Partner:
partner.demo@firm.local - Analyst:
analyst.demo@firm.local - Compliance:
compliance.demo@firm.local
Example request:
curl -s http://localhost:8000/api/v1/deals \
-H "X-User: analyst.demo@firm.local" \
-H "X-User-Role: Analyst"- DD checklist without code changes: data/dd_checklist.yaml
- IC voting rules via env:
IC_QUORUM_REQUIREDIC_QUORUM_PERCENTAGEIC_REQUIRED_ROLESIC_APPROVAL_THRESHOLD
- Branding:
FIRM_NAMEFIRM_LOGO_URL- Optional upload endpoint:
POST /api/v1/branding/logo
- Dependency health:
GET /health(DB, Redis, Celery broker/worker) - Portfolio risk snapshot:
GET /api/v1/monitoring/portfolio - Alert records:
GET /api/v1/monitoring/alerts - Trigger alert scan:
POST /api/v1/monitoring/alerts/run
Logging guide: docs/logging.md Deployment and hardening: docs/deployment.md Stress test script and latest report:
- scripts/stress_test.py
- docs/reports/stress-test-report-2026-02-26.md
- docs/reports/artifacts/stress-results-2026-02-26.json
- Visual presentation built from stress artifacts only:
- Audit model: docs/compliance-audit-model.md
- RegTech/supervisory view: docs/regtech-supervisory-view.md
- Audit export script:
python scripts/export_audit_log.py --output data/processed/audit_log_export.csvThis platform is designed for private deployment inside your own environment (VPC/on-prem):
- Postgres + Redis + Chroma can run fully private.
- Document storage can stay internal.
- Heuristic and Ollama modes avoid external AI dependencies.
- Developer template: /.env.example
- Firm deployment template: /.env.firm.example
src/backend API, services, workflows, tasksfrontend/React dashboardscripts/seed/export/init utilitiesdata/uploaded docs, checklist config, demo assetsdocs/operating model, deployment, governancetests/unit and integration coverage



