Skip to main content

Ask AI on your documentation

Your documentation is where people go to figure out how to do something with your product, and where they go when they are stuck: mid-setup, mid-integration, mid-error. Either way, what they find is pages, and a question rarely maps to one page: the answer is spread across a guide here, a reference there, a changelog entry from last month. So they scan, open tabs, search again, and too many give up and file a ticket, or quietly stop using the product. People do not want to search; they want to ask. An assistant on your documentation site lets them do exactly that, right where they already are: ask in their own words, get an answer with citations into your pages, and keep moving, whether they are evaluating your product, setting it up, or troubleshooting something that does not work.

Ways to build this with Kapa​

  • Deploy the Website Widget: a chat interface on your documentation site, answering with citations back into your pages. A script tag away; the widget in the corner of this page is exactly this deployment.
  • Build the interface yourself with the Chat SDK when you want the same managed agent behind a UI that is fully yours.
  • Build any public chat interface on agentic retrieval when you want to own the agent as well as the UI: your bot, grounded in your knowledge through the search tool.

What to index​

Your public product knowledge: the documentation site via Website Crawl, community threads via Discourse or Stack Overflow, API specifications via OpenAPI, and anything from GitHub, code, issues, discussions, and pull requests, via the GitHub connectors. Start with Index your first source.

Resources​