Vodou's AI features are optional, off by default, and run entirely against your own local Ollama installation — nothing is sent to a cloud AI service.

The easy way

☰ → Settings → Local AI → Set up Local AI… installs Ollama for you on Windows (it downloads and opens the official installer — you agree to its terms there, nothing runs silently) and pulls the default model, llama3.2. Turning on Local AI (Ollama) without Ollama installed offers this same wizard automatically. On macOS/Linux, the wizard points you to ollama.com/download instead; the automated installer step is currently Windows-only.

The manual way

Install Ollama yourself, run ollama pull llama3.2 (or any chat-capable model), then turn the feature on: ☰ → Settings → Local AI → Local AI (Ollama).

If Ollama isn't running or has no models pulled, the AI panel says so directly — "Couldn't reach Ollama. Is it running? (start it with ollama serve)" — rather than failing silently. Everything else in Vodou works normally either way.