Janmitra - Voice Companion for Rural India

Toll-free voice companion that makes every rural citizen financially smart & infra-connected

Janmitra is a dialect-aware voice AI that connects rural Indian citizens to government services, bridging the information gap that affects 330 million people.

The Problem

Rural India faces a critical information asymmetry crisis:

• 68% of rural adults cannot name one government credit scheme (RBI 2024)
• <9% use internet in their local language (IAMAI 2023)
• Average 3.2 physical visits & ₹260 spent per failed loan application
• 27 days average delay in complaint redressal

Middlemen exploit this gap, language barriers isolate citizens, and outdated IVR systems fail to serve local dialects.

The Solution

Janmitra is a missed-call away, dialect-aware voice bot that:

• Answers only in local tongues (Hindi, Awadhi, Bundeli, Telugu, and more)
• Pulls verified information from government APIs and databases
• Warm-transfers callers to the exact officer when human escalation is needed
• Works offline-capable for areas with poor connectivity

Technical Architecture

User Speech → LiveKit Frontend (React) → LiveKit Server → Sarvam AI Agent
                                      ↓                           ↓
                               WebRTC Connection          Dialect Detection
                                      ↓                           ↓
                               Search Tools → Government APIs   RAG System
                                      ↓                           ↓
                               Verified Information       Contextual Responses

Core Components

• Frontend: Next.js React app with LiveKit Components for WebRTC voice interface
• Backend: LiveKit Agents Node.js server with Sarvam AI model
• AI Pipeline: End-to-end audio processing with automatic dialect detection
• Tools: Web search for government information, RAG for verified knowledge
• Future: Exotel telephony integration for toll-free access

Key Features

• 🔊 Automatic dialect detection and response in local languages
• 📞 Voice-first interface optimized for low-literacy users
• 🔍 Verified government data from official sources
• 👥 Officer directory with direct contact information
• 📱 Progressive enhancement from web to telephony
• 🌐 Offline-capable responses for poor connectivity

Getting Started

Prerequisites

• Node.js 18+ and pnpm
• LiveKit Cloud account (sign up)
• Sarvam AI API Key

Installation

1. Clone the repository

   git clone <repository-url>
   cd janmitra

2. Install dependencies

   # Frontend
   cd frontend && pnpm install
   
   # Backend
   cd ../backend && pnpm install

3. Set up environment variables

   Frontend (.env.local)

   LIVEKIT_API_KEY=your_livekit_api_key
   LIVEKIT_API_SECRET=your_livekit_api_secret
   LIVEKIT_URL=https://your-livekit-server-url

   Backend (.env.local)

   LIVEKIT_URL=https://your-livekit-server-url
   LIVEKIT_API_KEY=your_livekit_api_key
   LIVEKIT_API_SECRET=your_livekit_api_secret
   SARVAM_AI_API_KEY=your_sarvam_ai__api_key

4. Download AI models (backend)

   cd backend
   pnpm run download-files

5. Start the services

   # Terminal 1: Backend
   cd backend && pnpm run dev
   
   # Terminal 2: Frontend
   cd frontend && pnpm dev

6. Open your browser Visit http://localhost:3000

Development

Project Structure

janmitra/
├── frontend/                # Next.js React app
│   ├── app/                 # Next.js app router
│   ├── components/          # React components
│   └── lib/                 # Utilities
├── backend/                 # LiveKit Agents server
│   ├── src/                 # TypeScript source
│   └── dist/                # Compiled output
└── docs/                    # Documentation

Available Scripts

Frontend:

• pnpm dev - Start development server
• pnpm build - Build for production
• pnpm lint - Run ESLint
• pnpm format - Format code with Prettier

Backend:

• pnpm run dev - Start development server
• pnpm run build - Build TypeScript
• pnpm run typecheck - Type checking
• pnpm run lint - Run ESLint
• pnpm test - Run tests

Code Style

• TypeScript for type safety
• ESLint + Prettier for code formatting
• Prettier plugin for import sorting
• Follow existing patterns in the codebase

Demo Scenarios

Janmitra should handle these conversations naturally:

PM Kisan Scheme Query:

User: "मेरे खेत में कितने एकड़ हैं, क्या मुझे पीएम किसान योजना का पैसा मिलेगा?"
Agent: "हाँ, अगर आपके पास 2 हेक्टेयर तक जमीन है तो आप PM Kisan योजना के लिए योग्य हैं..."

Loan Eligibility:

User: "मैं ST/SC category का हूँ, कौन सी government loan schemes हैं?"
Agent: "आपके लिए विशेष schemes हैं जैसे PM SVANidhi, Stand-Up India..."

Officer Transfer:

User: "मेरी pension नहीं मिली, किससे contact करूँ?"
Agent: "आपके district का pension officer का number है +91-XXXXXXXXXX..."

Deployment

LiveKit Cloud

1. Deploy backend to LiveKit Cloud

   lk app deploy

2. Update frontend environment with production URLs

3. Configure domain and SSL certificates

Docker Deployment

# Build backend image
cd backend && docker build -t janmitra-backend .

# Run with environment variables
docker run -p 3001:3001 \
  -e LIVEKIT_URL=$LIVEKIT_URL \
  -e SARVAM_AI_KEY=$SARVAM_AI_API_KEY \
  janmitra-backend

Future Roadmap

• Phase 1: Exotel telephony integration for toll-free access
• Phase 2: Direct government API integration
• Phase 3: Officer directory with real-time availability
• Phase 4: Multi-modal support (SMS, WhatsApp)
• Phase 5: Offline PWA capabilities

Contributing

We welcome contributions! Please:

1. Fork the repository
2. Create a feature branch
3. Make your changes
4. Add tests if applicable
5. Submit a pull request

Development Guidelines

• Follow the existing code style
• Add tests for new features
• Update documentation
• Ensure all tests pass

License

• Code: MIT License
• Content: CC-BY-4.0 License

Acknowledgments

Built with LiveKit Agents and powered by Sarvam AI. Inspired by the need to bridge India's rural-urban digital divide.

---

Contact: For questions or partnerships, reach out to the development team.