Documentation your users actually read
Your users are humans and their AI agents — GitBook publishes for both. Write in the editor, drive it from your agent, or sync from Git.
How do you want to work?
The same docs, three ways in. Most teams use more than one.
Write and publish visually. Blocks, change requests, and review without touching a terminal.
Drive GitBook from Claude, Cursor, or any MCP client. Draft, restructure, and open change requests.
Sync from Git, script with the CLI, or build on the API and ship your own components.
From first draft to published site
Follow it end to end, or pick up where you left off.
Start fresh or bring what you have.
Review before anything goes live.
Structure, brand, and control access.
See what readers need next.
AI out of the box
Two work for you while you write. Three work for your readers after you publish.
Keep going
Last updated
Was this helpful?
