Guides

Setting up AI

Connect your own AI model for slips, images and in-app chat.

Todebaran's AI features run on your own model and API key — you bring the provider, you stay in control of the key, and there's no middleman. This powers AI Prompt and image blocks in slips and the in-app AI chat.

AI chat is a Pro feature

The floating in-app AI chat is part of Pro. AI Prompt and AI image blocks in slips work with your own key. See pricing.

Add a model#

Open Settings → AI Settings.
Click Add Model and choose a provider:
- OpenAI, Anthropic or Google, or
- Custom — for any OpenAI-compatible server, including local ones like Ollama, LM Studio and vLLM.
Enter the Model ID (each provider suggests a sensible default), and paste your API key.
Mark one model as the primary — that's the one used by default.

Screenshot coming soon

Settings → AI Settings, with a model's provider, model ID and API key.

Your key stays private

API keys are stored securely on your Mac and are never put into your synced settings or sent to Todebaran. Only you and your chosen provider ever see them.

Image generation#

To generate images inside slips, turn on image generation for a model and set its image model ID, then pick a primary image model. Image blocks set to the AI source will use it.

Local and custom servers#

Running a model locally? Choose the Custom provider and set the Base URL (under a model's advanced options) to your server's address. This keeps AI fully on your machine — no data leaves your Mac.

In-app chat#

Turn on Enable AI chat to show the floating chat button and chat with your model right inside Todebaran. Chat requires a Pro subscription.

Work with your tasks from another AI client

Todebaran can also act as an MCP server, so an outside AI client like Claude Desktop can read and act on your tasks. See MCP integration.

If AI isn't responding, see AI isn't responding.