Floating AI chat for macOS
Tomo
Capture selected text or a screenshot from any app and send it to an OpenAI-compatible cloud, local, or self-hosted model endpoint.
Requires macOS 14 or later and a configured model provider.
How Tomo works
- Select text or take a screenshot in the app you are already using.
- Tomo opens with that context attached, ready for your question.
- Dismiss the window and return to the active app.
Examples
These examples show selected text, screenshots, and documents used as chat context.
Model endpoints
Tomo supports OpenAI-compatible cloud, local, and self-hosted endpoints. It stores provider, model, and reusable persona configurations.
OpenAI · Anthropic · Gemini · OpenRouter · Ollama · Custom
Privacy
Tomo sends captured context to the selected model endpoint. It does not run commands or modify files.
- Credentials
- Provider keys are stored in the macOS Keychain.
- Context
- Only the text and images you capture enter a conversation.
- Providers
- Anything you send is handled according to your chosen endpoint's policies.
- Local models
- Connect Ollama or another local endpoint to keep model traffic on your Mac.
Questions
How does Tomo capture content from my screen?
Use customizable hotkeys to grab selected text or screenshots from any app. The content appears instantly in your chat. No manual copy-paste needed.
Is Tomo an "AI agent"?
No. Tomo does not run commands or modify your files. It sends the context you provide to the selected model endpoint and displays the response.
What model providers does it support?
Any OpenAI-compatible chat completion API: OpenAI, Anthropic, Gemini, DeepSeek, local and self-hosted models, and more. Just add your URL and API key.
What else do I need to install?
Nothing else for cloud providers: add your API key in Tomo and start chatting. To run models locally, install and start Ollama.