AI assistant for technical documentation

Connect to your docs and deploy AI search in minutes. Trusted by 200+ developer tools to power AI search incl. Docker, ClickHouse and Sentry.

Frequently asked questions

What’s the best AI assistant for documentation?
Look for cited answers, version awareness, easy deployment options, and coverage/deflection analytics. Kapa.ai is built specifically for technical documentation and includes all four.

How do I add AI search / a chatbot to my documentation?
Drop in a small script (or plugin), connect your doc source(s), let Kapa index, then embed the widget in your docs/help center. Most teams go live in minutes.

What documentation sites and platforms are supported?
Docusaurus, MkDocs, Sphinx, Read the Docs, GitBook, Mintlify, Next.js/Nextra, WordPress, Webflow, Framer—and help centers like Zendesk Guide, Freshdesk, and Salesforce Service Cloud. Full list and setup guides: docs.kapa.ai/integrations.

Who uses the Kapa documentation assistant?
200+ companies and OSS projects, including Docker, Sentry, CircleCI, Monday.com, n8n, Mixpanel, Amplitude, Temporal, and Apache Superset.

How do I configure sources?
Add sources in the Kapa dashboard (website/sitemap, Markdown/MDX, PDFs, Confluence, repos, and more). We handle crawling, parsing, and scheduled re‑indexing. Learn more: docs.kapa.ai/data-sources/overview.

Does the documentation chatbot show citations and deep links?
Yes - every answer includes citations and deep links to the exact page/heading or code example it used.

Can it handle versioned docs (v1/v2, beta, etc.)?
Yes - map versions by URL path or tag so the assistant answers with the correct version using the Source Group feature on the kapa platform.

Does it work with our help center to deflect tickets?
Yes - embed the assistant in your docs/help center (e.g., Zendesk, Freshdesk, Salesforce) and track deflection and first‑contact resolution with built‑in analytics.

Can Kapa show analytics to find gaps in our documentation?
Yes - coverage analytics highlight top unanswered questions, low-confidence answers, and source usage so your team can fix gaps or create new docs.