# DRISHTI **AI-Powered Intelligent Disaster Response & Resource Coordination — Prototype** > "From fragmented disaster information to coordinated action." This is a hackathon-grade **prototype**, not a production disaster-management system. It demonstrates: emergency report → AI extraction → transparent priority scoring → resource recommendation → map visualization → simulated safe route → live dashboard. --- ## 1. Run the backend ```bash cd backend python -m venv venv source venv/bin/activate # Windows: venv\Scripts\activate pip install -r requirements.txt cp .env.example .env # Optional: put your ANTHROPIC_API_KEY in .env for real AI extraction. # If left blank, DRISHTI automatically uses a deterministic fallback # analyzer, so the demo works with zero setup and never crashes. uvicorn main:app --reload --port 8000 ``` The API starts at `http://localhost:8000`, creates `drishti.db` (SQLite) automatically, and seeds demo resources/shelters/incidents on first run. Check it's alive: open `http://localhost:8000` — should show `{"status": "DRISHTI API operational", ...}`. ## 2. Run the frontend ```bash cd frontend npm install cp .env.example .env npm run dev ``` Open `http://localhost:5173`. --- ## 3. Demo script (~3 minutes) 1. **Dashboard** — show seeded incidents, critical/high counts, available resources, map. 2. **Report Emergency** — click "Use example (Assam flood demo scenario)", then **Analyze with AI**. 3. Review the structured extraction and the **"CRITICAL because…"** explanation. 4. Click **Confirm & Create Incident** — you land on the Incident Details page showing the recommended resource (e.g. "Boat 01 → Incident #N") and why it was picked. 5. Click **Show Safe Route** — see the incident → shelter path on the map. 6. Go back to **Dashboard** — stats and resource availability have updated live. --- ## Responsible AI note DRISHTI is an **AI-assisted decision-support** tool. The AI only extracts structured facts from free text. Priority scoring and resource matching are deterministic, rule-based, and shown transparently in the UI — never a black box. Final response decisions remain with authorized human responders. All resource, shelter, and route data in this prototype is **simulated**, not live government data. ## What's deliberately out of scope for this prototype No auth, no real routing engine, no live data feeds, no notifications, no mobile app, no multi-agent AI orchestration, no container orchestration. See the architecture doc discussed with the team for the full list. # -DRISHTI-AI-Powered-Intelligent-Disaster-Response-Resource-Coordination-Prototype