website agent
Turn your docs into a self-serve agent on your site
The answer is already in your documentation. The Kapa Website Agent helps customers find it instantly, anywhere on your site.
how it works
Connect your sources
Connect 30+ sources: help centers, GitHub code, wikis, PDFs, and SDKs.
Customize your agent
Set brand color, logo, font, layout, MCP installation link, and much more.
Deploy in 5 minutes
Copy and paste a code snippet and go live with just a few clicks.
The self-serve agent trusted to perform in front of millions of users, every day
“We have a Kapa agent live on our website, which addresses thousands of customer queries daily and helps users navigate our extensive product documentation.”
Radhika Chennakeshavula, CIO
Try Kapa with your technical content
Test and evaluate Kapa for free
Go live in less than 7 days
SOC 2 Type II & GDPR compliant
What is the Website Agent and who is it for?
The Website Agent is an AI answer widget that you embed on your website, documentation, or help center so users can ask natural language questions and get instant, grounded answers. It is built for technical and enterprise companies with complex products, and it is used by 200+ companies to deflect repetitive questions and help customers self-serve. Rather than searching through docs, users get a direct answer drawn from your own knowledge sources, with clickable links to the underlying material. Beyond answering, it can take actions like escalating to a human, opening a support ticket, or capturing a lead without the user leaving the conversation. It is a good fit for documentation, support, and developer experience teams who want to reduce ticket volume while improving the self-serve experience.
How does it stay accurate and avoid hallucinations?
The Website Agent is grounded in your own knowledge sources, so it answers from your documentation rather than relying on a model's general training. Every answer includes clickable source citations that link back to the original content, which lets users verify the response and dive deeper. You can scope answers to specific products or versions using source groups, so the agent only draws from the material that is relevant to a given page or audience. Kapa is built specifically for technical content and complex products, and its retrieval approach is designed to return reliable answers in front of large user audiences. If the answer is not in your sources, the agent is designed to say so rather than fabricate a response.
How is our data secured and kept private?
Kapa is SOC 2 Type II certified and GDPR compliant, and it maintains a public trust center for reviewing its security posture. Data is encrypted both at rest and in transit, and Kapa has Data Processing Agreements with the LLM vendors it uses so that your data is never used to train those models. For sensitive content, Kapa offers PII protection that scans documents during crawling and either removes detected personal information or replaces it with anonymized labels, storing only the sanitized version. You can also define custom entities using regular expression patterns to protect things like API keys and passwords. These controls let you deploy a public-facing agent while keeping governance and compliance requirements in view.
How much engineering effort does setup take and how fast can we go live?
Deployment is intentionally lightweight, requiring just a single script tag dropped into your site with no migration or rebuild. It works out of the box with platforms like Mintlify, Docusaurus, GitBook, ReadMe, Webflow, and Zendesk, so most teams can go live in one step. You create the integration in the Kapa app, connect your sources, customize the agent, then copy and paste the snippet to deploy. Kapa positions the Website Agent as something you can stand up in days, and the page references going live in less than 7 days. There is a free trial so you can test and evaluate before committing.
What knowledge sources can it use, and can we match it to our brand?
The Website Agent can connect 30+ source types, including help centers, GitHub code, wikis, PDFs, and SDKs, so it draws on the knowledge you already maintain. On the branding side, you can set your own brand color, logo, font, and layout, and you can customize the tone of voice so the agent feels like a native part of your product. You can also configure elements like an MCP installation link and other behavioral settings to tailor the experience. Source groups let you control exactly which content the agent can access per deployment, which is useful for multi-product or versioned documentation. The result is a fully on-brand, scoped agent rather than a generic chatbot bolted onto your site.