Skip to main content
Version: 2026.09

AI Chat & Providers

This is the reference for configuring the AI chat provider. For how to use AI chat in the viewer, see AI Chat.

"AI Chat Provider" is its own entry in Settings (a full screen on mobile, a dialog on desktop) and is also reachable from the AI chat panel. Pick a provider from the row of provider cards; the panel below swaps to match.

AI provider settings on desktopDesktop
AI provider settings on mobileMobile

Falcon AI (hosted)​

The built-in provider, powered by Google Gemini through Falcon's secure backend — no API key or setup needed. The panel shows your current plan, remaining trial quota, and a Free/Pro/Premium comparison.

Local AI (Ollama) Pro​

Desktop only. Run a local model via Ollama:

  • A Ping Daemon check confirms Ollama is reachable.
  • Set the daemon endpoint.
  • Register models and pick a default, or remove them.

OpenAI & OpenRouter (bring your own key) Pro​

Use your own OpenAI-compatible provider:

  • Named profiles you can add, switch between, and delete.
  • Provider chips (OpenAI / OpenRouter / Groq / Together / Custom) pre-fill the default model.
  • An API key field with show/hide and Verify.
  • Model tag, optional base URL, and Standard / Advanced tool-capability options.

Chat limits​

Limits are counted per document and enforced by the backend. See the table in AI Chat.