Local AI

QuillArk’s AI runs on your computer. Your manuscript is never sent anywhere, and there is no code in QuillArk that contacts an online AI service.

The built-in engine

Works immediately, with nothing to download. It summarises chapters and scenes, organises brain dumps into topics, and pulls out likely characters, places, objects and tasks. It works from what you have written rather than inventing new text, which makes it reliable for summaries and notes.

Using a local language model

If you run Ollama or a llama.cpp server on this computer, choose it in Settings → Local AI and give the address, such as http://127.0.0.1:11434. QuillArk checks that the address is on your own machine and refuses anything else.

Nothing is replaced without asking

AI output is always shown as a suggestion. You see it, compare it, and choose whether to insert or replace. Before any of your writing is replaced, a revision is taken.