🛡️ Rakshak AI

Zero-Touch, AI-Driven Autonomous Crisis Command Platform

🔴 Live Working Prototype

(Click to view the deployed application)

⚠️ Note (Free Tier Hosting): Because this prototype is hosted on free-tier services, the backend servers may "spin down" after a period of inactivity. If the dashboard or AI seems unresponsive or fails on the very first try, please wait and refresh the page. It just takes a moment for the AI microservices to wake up!

---

The Problem: Hospitality venues face unpredictable, high-stakes emergencies that demand instantaneous reactions. Critical information is often siloed, fracturing communication between distressed guests, staff, and first responders resulting in fatal delays.

The Solution: Rakshak AI bypasses human panic by acting as an autonomous digital dispatcher. It replaces manual 911 calls and fragmented radio chatter with an AI pipeline that automatically analyzes threats, dynamically maps the nearest specific emergency stations, and executes automated SMS dispatches to first responders.

✨ Standout Features

• 🧠 Autonomous Agent Pipeline: A Python-powered LangGraph agent that ingests incident data and autonomously evaluates threat severity without human bottleneck.
• 👁️ Transparent "Chain-of-Thought": A dedicated UI panel exposes the AI’s internal logic, showing human operators exactly why it escalated a threat and how it plans to mitigate it.
• 🗺️ Hyper-Local Dynamic Mapping: Integrates with the Overpass API to scan the immediate geographical radius of the venue to pinpoint the exact nearest Police, Fire, or Medical units.
• 💬 Zero-Touch SMS Dispatch: Automatically formats and blasts critical SOS SMS messages to the nearest identified emergency responders via Twilio.
• 🏢 Digital Twin & Spatial Awareness: Includes a 3D Facility Twin to give on-site security spatial layout awareness of the threat location within the building.
• ❤️ Rakshak Neural Core: A central, visually dynamic pulsing orb that acts as the real-time heartbeat of the system, instantly shifting states (Cyan to Red) to provide visceral threat awareness.
• 🎮 Built-In Crisis Simulator: Allows stakeholders to artificially inject emergencies into the system to run fire drills and test AI response latency safely.

🛠️ Cutting-Edge Tech Stack

Rakshak AI is built on a scalable Microservices Architecture:

• AI & Logic Engine: Google Gemini LLM, LangChain, LangGraph, Python 3, FastAPI, Uvicorn.
• Frontend Command Center: React.js (Vite), Framer Motion, React Flow (@xyflow/react), Modern CSS.
• API Gateway & Orchestration: Node.js, Express.js.
• External Integrations: Twilio SMS API, Overpass API (OpenStreetMap).

🏗️ System Architecture

flowchart TB
    classDef frontend fill:#1e293b,stroke:#3b82f6,stroke-width:2px,color:#fff,rx:5px,ry:5px;
    classDef nodejs fill:#1e293b,stroke:#10b981,stroke-width:2px,color:#fff,rx:5px,ry:5px;
    classDef python fill:#1e293b,stroke:#f59e0b,stroke-width:2px,color:#fff,rx:5px,ry:5px;
    classDef external fill:#1e293b,stroke:#ef4444,stroke-width:2px,color:#fff,rx:5px,ry:5px;

    subgraph Client ["🖥️ Client Tier (Frontend)"]
        UI[React.js Command Dashboard]:::frontend
        Twin[Facility Digital Twin]:::frontend
        Graph[Live Agent Graph UI]:::frontend
    end

    subgraph Gateway ["⚙️ Orchestration Tier (Node.js)"]
        Node[Express.js API Gateway]:::nodejs
    end

    subgraph AI ["🧠 AI Processing Tier (Python Microservice)"]
        Fast[FastAPI & Uvicorn]:::python
        Lang[LangGraph Agent]:::python
        LLM[Google Gemini Core LLM]:::python
        
        Fast <--> Lang
        Lang <--> LLM
    end

    subgraph External ["🌍 External APIs & Infrastructure"]
        Twilio[Twilio SMS API]:::external
        Maps[Overpass API]:::external
    end

    %% Connections
    UI <-->|HTTP/WS| Node
    Twin -.-> UI
    Graph -.-> UI
    
    Node <-->|REST Relay| Fast
    Lang -->|Radius Search| Maps
    Node -->|Automated Dispatch| Twilio

Loading

---

🚀 Local Development

Install dependencies for all workspaces:

npm run setup

Run all three services together concurrently:

npm run dev

Default local ports:

• Frontend: http://localhost:5173
• Node relay: http://localhost:3000
• Python agent: http://localhost:8000

☁️ Deployment

The included render.yaml handles deploying the microservices:

1. rakshak-backend as the public web app and Node relay
2. rakshak-agent as the Python LangGraph agent

Important deployment behavior:

• The Node service builds the frontend and serves frontend/dist directly.
• The Node service talks to the Python agent through PYTHON_AGENT_URL.
• On Render, PYTHON_AGENT_URL is populated from the Python service hostport, which the Node backend normalizes to an internal http:// URL.

🔐 Required Environment Variables

Python Agent (backend/backend-python/.env):

• GEMINI_API_KEY
• TOMTOM_API_KEY (Optional fallback)

Node Relay (backend/backend-node/.env):

• TWILIO_ACCOUNT_SID
• TWILIO_AUTH_TOKEN
• TWILIO_PHONE_NUMBER

Frontend (frontend/.env - Optional):

• VITE_BACKEND_URL (If the frontend is served by the Node relay in production, no override is required).