Your AI-powered meeting assistant that runs locally

CatsUp is a stealthy, cross-platform meeting assistant designed to run locally on your machine. It detects when you join a meeting, records the session, transcribes in real-time, and uses AI to generate summaries and answer questions about the discussion.

• Stealth Mode — Unobtrusive system tray/menu bar application
• Cross-Platform — Runs on Windows, macOS, and Linux
• Meeting Detection — Automatically detects Teams, Zoom, Slack, Google Meet
• Smart Recording — Captures video (window), system audio, and microphone
• Real-Time Transcription — Live speech-to-text with speaker detection
• Screen Vision — Understands what's on screen using OCR and vision AI models
• Trigger Word Detection — AI assistant activates when your name is spoken
• AI-Powered Responses — Contextual answers using OpenAI, Anthropic, Google, or local models
• Live Mini Window — Floating overlay with transcript, AI responses, quick actions
• Private Options — Use cloud APIs or 100% local processing with Ollama/llama.cpp

brew tap geiserx/catsup
brew install --cask catsup

Or install directly:

brew install --cask geiserx/catsup/catsup

Download the latest release from the Releases page.

1. Configure: Click menu bar icon → Settings → Add API keys
2. Set Trigger Word: Add your name (e.g., "Sergio") in Trigger Words tab
3. Start: Click "Start Detection" in menu bar dropdown
4. Join Meeting: CatsUp auto-detects and starts recording

CatsUp is built with a decoupled architecture to ensure cross-platform compatibility:

• Windows: .NET 6 SDK
• macOS: Xcode 15+
• Linux: Python 3.10+, FFmpeg, Xlib
• UI: Node.js 20+

# Clone the repository
git clone https://github.com/GeiserX/CatsUp.git
cd CatsUp

# Build the UI
cd ui
npm install
npm run build

Windows:

cd platform/windows
dotnet run

macOS:

cd platform/macos
swift run

Linux:

cd platform/linux
python main.py

The project includes GitHub Actions workflows for automated cross-platform builds.

CatsUp supports multiple AI providers for transcription, vision, and chat:

Contributions are welcome! Please check out the development branch and read our contributing guidelines.

This project is licensed under the GNU General Public License v3.0 — see the LICENSE file for details.