A full-stack real-time meeting analyzer that uses an LLM to extract structured insights from meeting transcripts and broadcasts them live to the frontend in under 1 second.
| Layer | Technology |
|---|---|
| Frontend | Next.js, TypeScript, Tailwind CSS |
| Backend | Node.js, Express |
| Database | PostgreSQL |
| Real-time | WebSockets (ws library) |
| AI | Groq LLM API (llama-3.3-70b-versatile) |
- Submit transcript chunks via REST API (
POST /api/transcript) - Extracts 4 structured insight types: summary, action items, decisions, topic
- Broadcasts AI insights to all connected clients via WebSocket in under 1 second
- Decoupled AI service layer enabling independent scaling
- Persistent storage of meetings, transcripts, and insights in PostgreSQL
LiveMeet/
├── backend/
│ └── src/
│ ├── database/ # PostgreSQL connection (db.ts)
│ ├── routes/ # REST endpoints (meetings.ts)
│ └── services/ # Groq AI service (ai.ts)
└── frontend/
└── app/ # Next.js pages and components
- Node.js 18+
- PostgreSQL
- Groq API key (free at console.groq.com)
# Clone the repo
git clone https://github.com/MM871/LiveMeet.git
cd LiveMeet
# Backend
cd backend
npm install
cp .env.example .env # fill in your credentials
npm run dev # runs on port 4000
# Frontend (new terminal)
cd frontend
npm install
npm run dev # runs on localhost:3000- User submits a transcript chunk via the frontend
- Backend receives it via
POST /api/transcriptand stores it in PostgreSQL - The AI service sends it to Groq LLM which returns structured insights
- Insights are broadcast to all connected WebSocket clients in real time
- Frontend dashboard updates instantly with summary, action items, and decisions