Zoomer — AI Meeting Assistant

An accessible AI-powered meeting assistant that provides real-time transcription, topic tracking, AI-generated summaries, transcript translation, and post-meeting Q&A.

---

Features

🎙️ Real-time Meeting Bot

• Integrates with Recall.ai to join Zoom meetings (Google Meet and Microsoft Teams coming soon...)
• Real-time transcription via webhooks
• Topic tracking with periodic updates sent to meeting chat

📝 Post-Meeting Recap

• Video playback with synced transcript navigation (click transcript to seek)
• AI-generated summaries using Gemini
• Transcript translation to 10+ languages (Spanish, French, German, Portuguese, Chinese, Japanese, Korean, Hindi, Arabic)
• Q&A interface — ask questions about the meeting content
• PDF export of meeting summaries

📚 Meeting History

• Browse and load previous meeting transcripts
• View past recordings and regenerate summaries

---

Screenshots

Summary	Q&A

---

Tech Stack

Layer	Technology
Backend	FastAPI, Python 3.9+
Frontend	Next.js 15 (App Router), React 19, TypeScript
Styling	Tailwind CSS 4
AI/LLM	OpenAI-compatible API (GPT-4o-mini), Google Gemini
Translation	deep-translator
Meeting Bot	Recall.ai

---

Prerequisites

• Python 3.9+
• Node.js 18+ (20+ recommended)
• ngrok installed + logged in (ngrok config add-authtoken ...)
• A Recall.ai API key
• An LLM API key (OpenAI-compatible) for topic tracking and Q&A
• A Google Gemini API key for summaries

---

Project Structure

zoomer/
├── backend/
│   ├── main.py              # FastAPI app with all endpoints
│   ├── llm_client.py        # OpenAI/Gemini LLM integration
│   ├── qa_engine.py         # Q&A engine for post-meeting questions
│   ├── topic_tracker.py     # Real-time topic detection
│   ├── tangent_detector.py  # Meeting tangent detection
│   ├── translate.py         # Translation utilities
│   ├── store.py             # In-memory meeting state
│   ├── transcripts/         # Saved transcript files (JSONL)
│   ├── requirements.txt
│   └── .env                  # Backend config (do not commit)
├── frontend/
│   ├── app/
│   │   ├── page.tsx         # Home page (start meeting / view history)
│   │   ├── meeting/         # Live meeting view
│   │   └── post-meeting/    # Recap page (video, transcript, summary, Q&A)
│   ├── components/ui/       # Reusable UI components
│   └── package.json
├── docs/images/             # Screenshots for README
├── .gitignore
└── README.md

---

Setup

1. Backend Setup

cd backend
py -m venv .venv
.\.venv\Scripts\Activate.ps1
pip install -r requirements.txt

Create backend/.env:

# Recall.ai
RECALL_API_KEY=YOUR_RECALL_KEY
RECALL_BASE_URL=https://us-west-2.recall.ai

# Public URL for webhooks (set after starting ngrok)
PUBLIC_BASE_URL=

# Webhook token auth
WEBHOOK_TOKEN=dev-secret-token

# Topic tracking
TOPIC_TRACKER_ENABLED=true
TOPIC_CHECK_INTERVAL_S=60

# LLM (OpenAI-compatible) — for topic tracking and Q&A
LLM_API_KEY=YOUR_LLM_KEY
LLM_MODEL=gpt-4o-mini
LLM_BASE_URL=https://api.openai.com/v1

# Gemini — for meeting summaries
GEMINI_API_KEY=YOUR_GEMINI_KEY

2. Frontend Setup

cd frontend
npm install

3. Start ngrok

ngrok http 8000

Copy the forwarding URL (e.g., https://xxxxxx.ngrok-free.dev) and update PUBLIC_BASE_URL in backend/.env.

---

Running the App

Terminal A — Backend

cd backend
.\.venv\Scripts\Activate.ps1
uvicorn main:app --reload --port 8000 --env-file .env

Verify: http://localhost:8000/healthz → {"ok": true}

Terminal B — Frontend

cd frontend
npm run dev

Open: http://localhost:3000

---

API Endpoints

Method	Endpoint	Description
GET	/healthz	Health check
GET	/transcripts	List saved transcripts
POST	/start-meeting-bot	Start a bot to join a meeting
POST	/meeting/{bot_id}/agenda	Set/update meeting agenda
GET	/meeting/{bot_id}/topic	Get current meeting topic
GET	/meeting/{bot_id}/status	Get bot status + recording URL
GET	/meeting/{bot_id}/transcript	Get full transcript
GET	/meeting/{bot_id}/summary	Generate AI summary
POST	/meeting/{bot_id}/leave	Tell bot to leave meeting
POST	/qa	Ask a question about a meeting
POST	/translate-file	Translate transcript to another language
POST	/recall/webhook/realtime/	Webhook for real-time transcripts
POST	/recall/webhook/bot-status/	Webhook for bot status changes

---

Usage

Start a New Meeting

1. Open http://localhost:3000
2. Enter your meeting link (Zoom, Google Meet, etc.)
3. Optionally add a meeting topic/context
4. Click Connect Assistant
5. The bot will join and start transcribing

View Meeting Recap

After the meeting ends (or select a previous meeting from the home page):

• Transcript tab: View/translate the full transcript, click to seek video
• Summary tab: AI-generated meeting summary with PDF export
• Questions tab: Ask questions about the meeting content

---

Common Issues

Bot joins but no transcript appears

• Ensure ngrok is running and PUBLIC_BASE_URL is set correctly
• Restart the backend after editing .env
• Check that webhook URLs include ?token=... matching WEBHOOK_TOKEN

Recording not available

• Recording processing can take a few minutes after the meeting ends
• The post-meeting page will poll for up to 2 minutes

Translation not working

• Ensure deep-translator is installed (pip install deep-translator)
• Check that the backend has internet access

---

.gitignore

backend/.env
backend/.venv/
backend/__pycache__/
backend/transcripts/
frontend/node_modules/
frontend/.next/