Choose the brain that helps me think

There are two different brains. Do not mix them up.

YOUR AI AGENT — ChatGPT, Claude, Cursor, Grok, OpenClaw — talks to Bommel through the Agent Gateway. That agent stays yours. Public read needs no Settings toggle and no new name.

The in-app librarian — the chat on this site after you sign in — can use a simple built-in brain, a model on your computer, or a cloud key in Settings. That is only for the box on this website.

What it is (in-app)

Behind the scenes, the in-app chat uses an AI model to write titles, pick tags, and understand what you're asking for. You get to choose which model it uses. There are three routes, and you can switch anytime.

Why you'll love it

• Start clever for free — no setup, no keys.
• Go fully private with a model on your own machine.
• Or plug in a top cloud model if you already have one.

Your three in-app choices

1. My built-in brain (offline). Works out of the box, needs nothing, and never sends your saves anywhere. It's simple but honest, and always available.
2. A model on your own computer. Run Ollama or LM Studio and point me at it. Private, free, and quite clever — see Use Ollama or LM Studio.
3. A big cloud model. Bring your own key from OpenAI, Anthropic (Claude), Gemini, Grok, or Perplexity — see Use OpenAI, Claude, Gemini, Grok, or Perplexity. Or see Pro is the library, not a brain we pay for if you expected us to host one.

How to use it

1. Open Settings and find the AI section.
2. Pick which brain the in-app chat should use.
3. Save. That's it — I'll start using it right away.

Which should you pick? If YOUR AI AGENT should read public Bommels, skip Settings and open the Agent Gateway. If you just want the site chat to work, start with the built-in brain. If you care most about privacy for that chat, use a local model. If you already pay for OpenAI or Gemini, plug that in.

Good to know

• Choosing an in-app model is Free; some cloud options cost money on the provider's side.
• Whatever you choose, your saves stay yours — a local model keeps in-app completions on your machine.
• Pairing YOUR AI AGENT to a private library is a different door: Connect YOUR AI AGENT.

Where to next

• Connect YOUR AI AGENT
• Use Ollama or LM Studio (private and free)
• Use OpenAI, Claude, Gemini, Grok, or Perplexity
• Pro is the library, not a brain we pay for