AI write assist in the desktop app

The desktop app supports the same AI write assist modes as the web platform — with multi-provider support including a local Ollama option for fully offline AI.

Configuring AI in the desktop app

Go to Settings → AI in the desktop app. The AI settings have three tabs:

Anthropic
Enter your Anthropic API key. The key is stored in the OS keychain — Windows Credential Manager on Windows, Keychain on macOS.

Ollama (local models, no internet required)
If you run Ollama locally, enter the server URL (default is localhost port 11434). Click Check to see available models, then select one from the dropdown. This option lets you use AI completely offline.

OpenAI-compatible
Any provider with an OpenAI-compatible API. Enter the base URL, model name, and your API key.

Selecting a provider

After configuring a provider, click Use [Provider] to make it active. The active provider shows a dot indicator on its tab.

Codex context in the desktop app

Entity names recognized in your local Codex are highlighted in amber in the editor. When you use AI assist, the local entity data is included in the context, so the AI is aware of your world even when offline.