Dome Process Analyzer
Map, govern, and automate your business processes with AI.
Paste a plain-language process description and get back a structured analysis: process steps, systems inventory, governance flags, and automation opportunities — with a Mermaid flowchart.
Built by Dome — the AI layer for governed operations.
Part of the DOME portfolio · Phase: Discover
Live
What you get
Submit any business process in plain language and receive:
- Process map — each step sequenced, typed (manual / automated / decision / approval / external), with actor, duration, wait time, and bottleneck flags
- Systems inventory — every tool and platform mentioned or implied, with shadow IT detection
- Governance flags — segregation of duties gaps, missing audit trails, approval threshold issues, data handling risks, and regulatory exposure, each with severity (critical / major / minor)
- Automation opportunities — specific recommendations by type (RPA, API integration, workflow engine, AI extraction, LLM agent, rule engine) with implementation complexity and prerequisites
- Metrics — automation coverage %, bottleneck count, shadow IT flag, critical flag count
- Mermaid flowchart — auto-generated process diagram
- Clarifying questions — when the description is ambiguous, the tool surfaces the gaps and lets you refine the analysis
Analyses are anonymous by default. Sign in with a magic link to save results to your account.
Repo Structure
dome-process-analyzer/
├── backend/ # FastAPI (Python 3.12) — deployed on Railway
└── frontend/ # Next.js 14 App Router — deployed on Vercel
Local Development
Backend
cd backend
python3 -m venv venv
source venv/bin/activate
pip install -r requirements.txt
cp .env.example .env # fill in ANTHROPIC_API_KEY + Supabase credentials
python3 -m uvicorn app.main:app --reload --port 8000
Frontend
cd frontend
npm install
cp .env.example .env.local # set NEXT_PUBLIC_API_BASE=http://localhost:8000
./node_modules/.bin/next dev # runs on port 3000
Note: The frontend calls the backend directly (not via the Next.js proxy) because analysis takes ~40 seconds, which exceeds the proxy timeout. NEXT_PUBLIC_API_BASE must point at the backend URL.
Environment Variables
Backend (backend/.env)
| Variable | Required | Description |
|---|
LLM_PROVIDER | Yes | claude | azure_openai | ollama — only claude is functional in v1 |
ANTHROPIC_API_KEY | Yes | Anthropic API key |
AZURE_OPENAI_ENDPOINT | No | Azure OpenAI endpoint (stubbed — not usable in v1) |
AZURE_OPENAI_KEY | No | Azure OpenAI key (stubbed — not usable in v1) |
AZURE_OPENAI_DEPLOYMENT | No | Azure OpenAI deployment name (stubbed) |
OLLAMA_URL | No | Ollama base URL, default http://localhost:11434 (stubbed — not usable in v1) |
SUPABASE_URL | Yes | Supabase project URL |
SUPABASE_ANON_KEY | No | Supabase anon key (magic link emails) |
SUPABASE_SERVICE_ROLE_KEY | Yes | Supabase service role key for backend DB access |
ENVIRONMENT | Yes | development | production |
ALLOWED_ORIGINS | Yes | Comma-separated CORS origins |
SITE_URL | Yes | Frontend URL — used in magic link email redirects |
AUTH_CALLBACK_URL | No | Magic link callback URL; defaults to https://domelayer.com/auth/callback |
CACHE_TTL_SECONDS | No | In-memory analysis cache TTL, default 3600 |
Frontend (frontend/.env.local)
| Variable | Description |
|---|
NEXT_PUBLIC_API_BASE | Direct backend URL (bypasses Next.js proxy) |
NEXT_PUBLIC_AUTH_BACKEND | Backend auth endpoint (usually same as above) |
API Reference
All endpoints are prefixed /api/v1.
Analysis
| Method | Path | Auth | Description |
|---|
POST | /analysis | None | Submit a process description; returns analysis + session_token |
POST | /analysis/{id}/refine | X-Session-Token header | Answer clarifying questions; returns updated analysis |
GET | /analysis/{id} | Bearer token | Retrieve a saved analysis |
POST | /analysis/{id}/save | Bearer token | Save analysis (requires consent: true in body) |
GET | /analysis | Bearer token | List saved analyses (paginated, default 20) |
DELETE | /analysis/{id} | Bearer token | Delete a saved analysis |
Auth
| Method | Path | Auth | Description |
|---|
POST | /auth/magic-link | None | Send magic link email |
POST | /auth/verify | None | Verify magic link token; returns access_token |
DELETE | /auth/session | Bearer token | Logout |
Health
| Method | Path | Description |
|---|
GET / HEAD | /health | Health check including Supabase status |
Rate limits: 10 req/60 s on POST /analysis, 20 req/60 s on POST /analysis/{id}/refine, 5 req/60 s on POST /auth/magic-link (per IP).
Deployment
| Layer | Platform | Config |
|---|
| Frontend | Vercel | Root dir: frontend, env vars set in dashboard |
| Backend | Railway | Root dir: backend, Dockerfile, env vars set in dashboard |
See backend/README.md for per-layer details.
Supabase keep-alive
The Supabase free tier pauses after 7 days of inactivity. Add an UptimeRobot monitor on /api/v1/health (free tier, 5-minute interval) to keep the project active independently of any local setup.
Stack
- Backend: FastAPI · Claude (Anthropic) · Supabase · structlog · Python 3.12
- Frontend: Next.js 14 · TypeScript · Tailwind CSS v3 · Mermaid · DOMPurify
- Auth: Supabase magic links via Resend SMTP
- Design: DOME Design System v2.0 — Space Grotesk headings, JetBrains Mono labels, electric blue
#0080FF
- Data: Zero retention by default — process descriptions are never persisted; analyses stored only on explicit user consent
Data & Privacy
- Process descriptions are never written to the database — they exist only in a server-side in-memory cache with a 1-hour TTL, then are discarded.
- Saving an analysis requires explicit consent (
consent: true). Only structured metadata and the analysis JSON are stored, never the original description.
- A SHA-256 hash of the input is logged in the governance event for audit purposes — not the content itself.
