How can we help?

Download the installer from gotchi.ai, open the DMG (Mac) or EXE (Windows), and drag to Applications. Gotchi will guide you through your first model download on launch.

Any model available through Ollama - the full open-source library including Llama 3.1, DeepSeek V3, Qwen 2.5, Mixtral, Gemma, and more. One-click install from the model library.

8GB minimum for small models (7B parameters). 16GB recommended for medium models (13-32B). 32GB+ for large models (70B+). Apple Silicon unified memory is particularly efficient.

Yes. All inference runs locally. No data is sent to our servers or any cloud service, and there is no telemetry. Connector tokens are stored in your OS keychain.

Absolutely. Once models are downloaded, everything works without internet. Connectors that access cloud services (Gmail, Slack) obviously need connectivity, but core AI features are fully offline.

Go to Settings → Connectors and click the service you want to connect. Follow the OAuth flow to authorize access. Credentials are stored in your system keychain.

The open-source core is free, with local inference and chat. One All-in-one plan (€49.99/year) unlocks everything else - unlimited agents, every connector, the MCP marketplace, and fine-tuning. Free updates included. Billed yearly, cancel anytime.

Use the feedback form in Settings → Help, or file an issue on our GitHub repository. Include your OS version, Gotchi version, and steps to reproduce.