JanSetu AI is a multilingual AI platform designed to transform citizen voices, text, and development requests into structured intelligence for policymakers.
It helps identify community problems, understand their context, classify development needs, assess urgency, and generate AI-assisted recommendations for better infrastructure and public-service planning.
AI recommends. Policymakers make the final decision.
- 🌐 Multilingual Citizen Input — Marathi, Hindi, and English
- 🎤 Voice-Based Requests — browser speech recognition for citizen voices
- 🤖 AI-Powered Analysis — converts natural-language requests into structured data
- 📊 Development Intelligence — identifies category, urgency, affected groups, and problem details
- 🎯 Priority Scoring — combines citizen demand, urgency, infrastructure gap, population impact, and vulnerability
- 🏛️ Policy Recommendations — AI-assisted recommendations for development planning
- 📍 Location-Aware Requests — captures district and state information
- 📱 Fully Responsive UI — designed for desktop, tablet, and mobile
- 🔐 Privacy-Aware Design — follows data minimisation and responsible AI principles
Citizen
│
├── Text
├── Voice
└── Regional Language
│
▼
Citizen Portal
│
▼
AI Request Analysis
│
├── Problem Identification
├── Category Classification
├── Urgency Detection
└── Affected Groups
│
▼
Development Intelligence
│
├── Citizen Demand
├── Infrastructure Gap
├── Population Impact
└── Vulnerability
│
▼
Priority Scoring
│
▼
AI Policy Recommendation
│
▼
Policymaker Decision
┌───────────────────────────────┐
│ Citizen Portal │
│ React + TypeScript │
│ │
│ Marathi • Hindi • English │
│ Text + Voice Input │
└───────────────┬───────────────┘
│
│ REST API
▼
┌───────────────────────────────┐
│ Backend API │
│ Node.js + TypeScript │
│ │
│ Request Processing │
│ AI Analysis │
│ Priority Intelligence │
│ Policy Recommendation │
└───────────────┬───────────────┘
│
┌───────┴────────┐
▼ ▼
┌──────────────┐ ┌──────────────┐
│ PostgreSQL │ │ Groq AI │
│ + Prisma │ │ LLM │
└──────────────┘ └──────────────┘
- React
- TypeScript
- Vite
- Tailwind CSS
- React Icons
- Browser Web Speech API
- Node.js
- TypeScript
- REST API
- Prisma ORM
- PostgreSQL
- Groq API
- AI-powered request analysis
- AI-assisted policy recommendations
- Multilingual citizen understanding
- Git
- GitHub
- ESLint
- TypeScript
JanSetu-AI/
│
├── frontend/
│ ├── src/
│ │ ├── hooks/
│ │ │ └── useSpeechRecognition.ts
│ │ ├── services/
│ │ │ └── api.ts
│ │ ├── types/
│ │ │ └── speech.d.ts
│ │ ├── App.tsx
│ │ ├── main.tsx
│ │ └── index.css
│ ├── package.json
│ └── .env.example
│
├── backend/
│ ├── src/
│ │ ├── controllers/
│ │ ├── routes/
│ │ ├── services/
│ │ ├── config/
│ │ ├── middleware/
│ │ ├── types/
│ │ └── utils/
│ ├── prisma/
│ ├── package.json
│ └── .env.example
│
├── docs/
├── .gitignore
├── README.md
└── LICENSE
git clone https://github.com/SRCarlo/JanSetu-AI.git
cd JanSetu-AIcd backend
npm installCreate a .env file:
DATABASE_URL=your_postgresql_connection_string
GROQ_API_KEY=your_groq_api_key
PORT=5000Run Prisma:
npx prisma generate
npx prisma migrate devStart the backend:
npm run devBackend API:
http://localhost:5000
Open another terminal:
cd frontend
npm installCreate .env:
VITE_API_URL=http://localhost:5000/apiStart the frontend:
npm run devOpen the URL shown by Vite, usually:
http://localhost:5173
JanSetu AI uses the browser's Speech Recognition API.
Supported languages:
| Language | Recognition Code |
|---|---|
| Marathi | mr-IN |
| Hindi | hi-IN |
| English | en-IN |
For the best experience, use a browser with Web Speech API support such as Google Chrome.
Microphone permission is required for voice input.
JanSetu AI calculates development priority using multiple signals:
Citizen Demand
+
Urgency
+
Infrastructure Gap
+
Population Impact
+
Vulnerability
↓
Priority Score
Priority levels:
80–100 → CRITICAL
60–79 → HIGH
40–59 → MEDIUM
0–39 → LOW
The scoring system is designed to help policymakers identify requests that may require greater attention.
A citizen request can be transformed into structured information such as:
{
"category": "WATER",
"urgency": "HIGH",
"problem": "Lack of reliable drinking water",
"affectedGroups": [
"Residents",
"Children",
"Elderly"
],
"summary": "The community lacks reliable access to drinking water.",
"confidence": 0.91
}The structured output can then be used by the intelligence layer for priority assessment and recommendations.
JanSetu AI is designed around India's linguistic and geographic diversity.
The platform can support expansion toward:
- Regional Indian languages
- Voice-first citizen interaction
- WhatsApp-based reporting
- District-level dashboards
- Infrastructure intelligence
- Government scheme mapping
- Public grievance intelligence
- Development priority heatmaps
JanSetu AI follows a human-led approach.
- Understand citizen requests
- Structure unstructured information
- Identify development categories
- Estimate urgency
- Generate recommendations
- Assist prioritisation
- Verify information
- Evaluate feasibility
- Allocate resources
- Make final decisions
JanSetu AI assists decision-making; it does not replace human governance.
JanSetu AI aims to bridge the gap between:
Citizen Voice
↓
AI Understanding
↓
Development Intelligence
↓
Policy Priorities
↓
Better Public Infrastructure
The long-term vision is to build a scalable Digital Public Good that enables governments and communities to turn citizen feedback into actionable development intelligence.
- 22+ Indian language support
- WhatsApp citizen integration
- Government/policymaker dashboard
- District-level analytics
- Interactive infrastructure maps
- Priority heatmaps
- Government scheme recommendation
- Offline/low-connectivity support
- Advanced demographic and vulnerability analysis
- Cloud deployment and scalable infrastructure
SRCarlo
GitHub:
https://github.com/SRCarlo
This project is intended for educational, research, and hackathon purposes.
Add an appropriate open-source license before production/public deployment.
Giving every citizen a voice in development.
Listen to citizens. Understand their needs. Prioritize development.