🎬 RevVision — "See it. Know it. Review it."

A real-time video AI agent that identifies movies and TV shows from your screen share or camera, then instantly surfaces reviews, ratings, and AI-powered sentiment summaries from RevApp.

Built for the WeMakeDevs Vision Possible: Agent Protocol Hackathon using Stream's Vision Agents SDK.

✨ Features

• 🖥️ Screen Share Recognition — Share your Netflix/Disney+ window, RevVision identifies what you're watching
• 📸 Camera Recognition — Point your webcam at movie posters, DVD covers, or streaming thumbnails
• 🤖 AI Review Summaries — Hear AI-generated review summaries from aggregated critic and user reviews
• ⭐ Instant Ratings — IMDB ratings fetched in real-time
• 🗣️ Voice Output — The agent speaks review summaries aloud via ElevenLabs TTS
• ⚡ Ultra-Low Latency — Under 30ms video latency via Stream's edge network

🏗 Architecture

User Video Feed → Stream Edge Network → RevVision Agent (Python)
                                              ↓
                                        Gemini Realtime (Vision AI)
                                              ↓
                                        RevApp API (Reviews)
                                              ↓
                                        Voice + UI Response

🛠 Tech Stack

Component	Technology
Agent Backend	Python 3.12+, Vision Agents SDK
Vision AI	Google Gemini Realtime (2 fps frame analysis)
Text-to-Speech	ElevenLabs
Movie Data	RevApp (TMDB + review aggregation + AI summaries)
Video Network	Stream Edge Network
Frontend	React + Vite + Stream Video React SDK

🚀 Quick Start

Prerequisites

• Python 3.12+
• uv (Python package manager)
• Node.js 18+
• API keys for: Stream, Google AI Studio, ElevenLabs

1. Clone the repo

git clone https://github.com/YOUR_USERNAME/RevVision.git
cd RevVision

2. Set up the Agent (Python backend)

cd agent
cp .env.example .env    # Fill in your API keys
uv sync                 # Install dependencies

3. Set up the Frontend

cd frontend
cp .env.example .env    # Add your Stream API key
npm install

4. Run

Terminal 1 — Start the agent:

cd agent
uv run python main.py --call-type default --call-id revvision

Terminal 2 — Start the frontend:

cd frontend
npm run dev

Open http://localhost:5173, enter the session ID revvision, and start watching! 🎬

📂 Project Structure

RevVision/
├── agent/
│   ├── main.py              # Vision Agent entry point
│   ├── revapp_client.py      # RevApp API client
│   ├── pyproject.toml        # Python dependencies
│   └── .env                  # API keys (not committed)
├── frontend/
│   ├── src/
│   │   ├── App.jsx           # Main app with Stream Video SDK
│   │   ├── components/
│   │   │   └── MovieCard.jsx # Movie info + review card
│   │   └── index.css         # Dark theme styles
│   ├── index.html
│   └── .env                  # Frontend config (not committed)
└── README.md

🎯 How It Works

1. You share your screen (or point a camera) via the web app
2. Video frames are sent to the RevVision agent at 2 fps via Stream's edge network
3. Gemini Realtime analyzes frames and identifies the movie/show
4. RevApp API is queried for reviews from IMDB, Twitter, Reddit, and Letterboxd
5. AI Summary is generated and spoken aloud via ElevenLabs
6. Movie card with ratings, reviews, and sentiment appears in the sidebar

🏆 Built For

WeMakeDevs Vision Possible: Agent Protocol Hackathon (Feb 22 – March 1, 2026)

Prizes: $2,000 (1st) · $1,500 (2nd) · $500 (Best Blog)

📝 License

MIT