Kontext Meeting Agent

Real-time meeting transcription, summarization, and integration platform powered by AI.

Kontext is a self-hosted meeting assistant that captures audio from your microphone, transcribes it in real-time using OpenAI's Whisper, generates intelligent summaries with Groq's LLMs, and automatically sends them to email, Notion on request, or stores them locally.

Enable streaming: Set ENABLE_STREAMING=true in .env

• Real-time Audio Capture: WebRTC-based audio streaming from browser to backend
• ** Transcription**: Streaming transcription using faster-Whisper (CPU-friendly)
• Smart Summaries: Automatic meeting summaries with key points, decisions, and action items
• ** Email Integration**: Send meeting summaries directly to team members via email
• Notion Integration: Push summaries to Notion databases for team documentation
• ** RAG Search**: Query meeting content using semantic search with Pinecone
• Post-Meeting Chat: Ask questions about meetings after they end
• Local Storage: SQLite database keeps all data on your server
• ** Docker Ready**: One-command deployment with Docker Compose

Features

Component	Technology	Purpose
Backend	FastAPI + Uvicorn	Real-time WebSocket streaming
Speech-to-Text	faster-Whisper (base/medium)	CPU-efficient transcription
Summarization	Groq API (Llama 70B)	High-quality meeting summaries
Embeddings	Sentence-Transformers	Semantic search for meeting content
Vector DB	Pinecone	Store and query meeting embeddings
Frontend	Vanilla JavaScript + HTML	Lightweight, no frameworks
Database	SQLite	Local persistent storage
Email	SMTP (Gmail)	Send summaries to team
Notion API	Official Python Client	Push to shared Notion databases

---

Quick Start

Prerequisites

• Docker & Docker Compose
• Groq API key (free at https://console.groq.com/keys)
• (Optional) Notion API token
• (Optional) Email credentials for sending transcripts

Setup

1. Clone the repository

git clone <repo-url>
cd Kontext-Agent

2. Create .env file

cp .env.example .env

3. Configure environment variables (edit .env)

# Groq API (required for streaming mode)
GROQ_API_KEY=your_groq_api_key_here

# Streaming configuration
ENABLE_STREAMING=true           # Set to false for local batch mode
FRAME_SIZE_MS=20               # Audio frame size in milliseconds
SUMMARY_INTERVAL_MIN=1         # Update summary every 1 minute

# Notion integration (optional)
NOTION_API_TOKEN=your_notion_token
NOTION_DATABASE_ID=your_database_id

# Email integration (optional)
SMTP_SERVER=smtp.gmail.com
SMTP_PORT=587
SENDER_EMAIL=your_email@gmail.com
SENDER_PASSWORD=your_app_password

4. Start the application

docker-compose up -d

5. Access the UI

http://localhost:8000

---

Usage

Starting a Meeting

1. Open http://localhost:8000
2. Click Start Recording
3. Allow microphone access when prompted
4. Speak naturally - transcription happens in real-time
5. See live captions with:
   • 🔴 Gray italic text = interim (being processed)
   • ⚫ Black bold text = final (confirmed)

During Meeting

• Transcripts appear in real-time (streaming mode)
• Summaries auto-update every 1 minute
• Live indicator shows when meeting is active

After Meeting

• Click Stop Recording
• Transcript and summary automatically saved
• Email sent (if configured)
• Notion database updated (if configured)

---

Configuration

Streaming vs Batch Mode

Streaming Mode (Default - Fast)

ENABLE_STREAMING=true
GROQ_API_KEY=your_groq_api_key_here
FRAME_SIZE_MS=20
SUMMARY_INTERVAL_MIN=1

• Pros: 100x faster, real-time, low memory
• Cons: Requires API key, rate limited on free tier
• Best for: Live meetings, demos, presentations

Batch Mode (Local - Privacy)

ENABLE_STREAMING=false
GROQ_API_KEY=    # Not needed
SUMMARY_INTERVAL_MIN=10

• Pros: No API needed, unlimited, local processing
• Cons: 30-60s latency, high memory (2-3GB), slow startup
• Best for: Privacy-first, offline, recorded audio files

Audio Configuration

# Frame size (20ms = 320 samples at 16kHz)
FRAME_SIZE_MS=20          # Recommended: 20-40ms
FRAME_SIZE_MS=10          # Faster but more overhead
FRAME_SIZE_MS=40          # Slower but less overhead

# Summary interval
SUMMARY_INTERVAL_MIN=1    # Streaming mode
SUMMARY_INTERVAL_MIN=10   # Batch mode

API Integration

Groq Streaming API

GROQ_API_KEY=your_key_here
ENABLE_STREAMING=true

• Get free key: https://console.groq.com/keys
• Free tier: 30 requests/minute (sufficient for ~6 hours of meetings)
• No setup needed beyond API key

Notion Integration

NOTION_API_TOKEN=your_token
NOTION_DATABASE_ID=your_database_id

Email Integration

SMTP_SERVER=smtp.gmail.com
SMTP_PORT=587
SENDER_EMAIL=your_email@gmail.com
SENDER_PASSWORD=your_app_password

---

Architecture

Components

┌─────────────────────────────────────────────────────┐
│             Web UI (index.html)                      │
│  • Audio capture (16ms frames in streaming mode)    │
│  • Real-time transcript display                     │
│  • Meeting controls                                 │
└────────────────┬────────────────────────────────────┘
                 │ WebSocket (AUDIO_FRAME)
                 ▼
┌─────────────────────────────────────────────────────┐
│        WebSocket Handler (websocket.py)             │
│  • Streaming transcription worker                   │
│  • Batch transcription worker (fallback)            │
│  • Broadcast partial/final results                  │
└────────────────┬────────────────────────────────────┘
                 │
        ┌────────┴────────┐
        ▼                 ▼
   ┌─────────┐      ┌──────────┐
   │   Groq  │      │  Local   │
   │  API    │      │  Whisper │
   │(Stream) │      │ (Batch)  │
   └────┬────┘      └────┬─────┘
        └────────┬───────┘
                 ▼
     ┌───────────────────────┐
     │  Audio Processing     │
     │  • Normalization      │
     │  • Noise handling     │
     │  • Format conversion  │
     └───────────────────────┘
                 │
        ┌────────┴────────┐
        ▼                 ▼
   ┌─────────┐      ┌──────────────┐
   │ Database │     │ Summarizer    │
   │ Storage  │     │ (Every 1 min) │
   └──────────┘     └─────┬────────┘
                          ▼
                   ┌──────────────┐
                   │ Integrations │
                   │ • Email      │
                   │ • Notion     │
                   │ • Services   │
                   └──────────────┘

Data Flow (Streaming Mode)

Speech → Browser (16ms frames) → WebSocket
  ↓
Groq API (100-300ms) → Partial results
  ↓
UI (real-time captions) → Database
  ↓
Summarizer (1-min interval) → Email/Notion

---

## File Structure

Kontext-Agent/ ├── app/ │ ├── init.py │ ├── main.py # FastAPI app │ ├── config.py # Configuration │ ├── database.py # Database operations │ ├── api/ │ │ ├── routes.py # HTTP endpoints │ │ └── websocket.py # WebSocket handlers │ ├── core/ │ │ ├── audio_processor.py # Audio normalization │ │ ├── transcriber.py # Local Whisper │ │ ├── rag_engine.py # RAG for summaries │ │ └── summarizer.py # Summary generation │ ├── services/ │ │ ├── streaming_transcriber.py # Groq streaming │ │ ├── meeting_service.py │ │ ├── email_service.py │ │ ├── notion_service.py │ │ └── meeting_service.py │ ├── models/ │ │ └── schemas.py # Pydantic models │ └── ui/ │ ├── index.html # Web UI │ └── audio_recorder.py ├── data/ │ └── meeting_transcripts/ # Stored transcripts ├── logs/ # Application logs ├── docker-compose.yml ├── Dockerfile ├── requirements.txt ├── start_server.py └── README.md

## Contributing 

Contributions welcome! Please:

1. Fork the repository
2. Create feature branch: `git checkout -b feature/amazing-feature`
3. Commit changes: `git commit -m 'Add amazing feature'`
4. Push to branch: `git push origin feature/amazing-feature`
5. Open Pull Request



- **Email**: funboy.ea@gmail.com

## Why I Built This 

I attend multiple meetings daily at Payvite and other commitments. Traditional meeting assistants are expensive. I wanted:

- **Privacy**: All data stays on my server
- **Cost**: No monthly subscriptions
- **Control**: Integrate with our tools (Notion, email)
- **Simplicity**: One-click to start recording
- **Extensibility**: Add new integrations easily

 Maybe it'll help you too!

---