# Kapa documentation

If you are building agents, most of them need access to knowledge to do their job. Kapa is the agentic retrieval you build them on: it indexes your unstructured knowledge, documentation sites, PDFs, tickets, community threads, API specifications, and gives any agent the most relevant content from all of it, the moment it needs it. That goes for the agents you build into your product and workflows, and just as much for the ones your users already work in, like Claude Code or Codex.

Kapa is an ingestion and retrieval system. [Ingestion](/knowledge-sources/data-ingestion.md) assembles a unified knowledge base from [20+ types of sources](/knowledge-sources/.md) and keeps it up to date as that content changes. [Agentic retrieval](/retrieval/.md) searches across it to find what the agent needs. You hand Kapa to your agent as a search tool via [MCP](/retrieval/hosted-mcp-server.md), or call it directly through the [API](/retrieval/http-api/.md). If your agent already runs in a supported framework or workflow tool, such as n8n or LangChain, [Framework integrations](/retrieval/.md#framework-integrations) connect it there. And for the cases where you do not want to build an agent at all, Kapa offers [Prebuilt Agents](/integrations/.md) for common use cases, out of the box.

## Quickstarts[​](#quickstarts "Direct link to Quickstarts")

Set up your first project with your agent, or step by step in the dashboard.

[Set up Kapa with your agentPaste one prompt, and your agent connects a source, tests it, and deploys it](/getting-started/set-up-with-a-coding-agent.md)

[Index a source in the dashboardCrawl a documentation site and watch it become searchable](/getting-started/index-your-first-source.md)

Once your knowledge is indexed, [Connect an AI agent to your knowledge](/getting-started/connect-an-ai-agent.md) shows an agent searching it over MCP.

## Explore by use case[​](#explore-by-use-case "Direct link to Explore by use case")

These are common use cases that all rest on retrieving your knowledge. You can build any of them on Kapa's agentic retrieval, and for many of them a Prebuilt Agent ships the solution out of the box:

[Knowledge for your in-product agent](/use-cases/in-product-agents.md)Give the agent inside your product access to your documentation, whatever framework it is built with

[Ask AI on your documentation](/use-cases/documentation-assistant.md)Users answer their own questions, across evaluating, setup, and troubleshooting

[Automate support with your knowledge](/use-cases/support-workflows.md)Answer tickets automatically from your knowledge and reduce the human ticket load

[Offer a public MCP server](/use-cases/public-mcp-server.md)Let your users' agents, from Claude Code to ChatGPT, search your knowledge through MCP

[Build a company brain](/use-cases/internal-knowledge.md)Your employees self-serve answers about your product, processes, and policies

[Answer your community's questions](/use-cases/community-support.md)Your users get answers in Slack and Discord around the clock

[Competitor intelligence](/use-cases/competitor-intel.md)Index your competitors' documentation and let your team query how their products work

[Answer RFPs](/use-cases/answer-rfps.md)Draft answers to RFPs and questionnaires with Claude or ChatGPT grounded in your knowledge

## Explore by example[​](#explore-by-example "Direct link to Explore by example")

These are our most popular tutorials. Each one builds something real on Kapa from start to finish, and most come with runnable code:

[Embed an AI assistant in your app that answers questions and takes actionsBuild the agent with the Vercel AI SDK and plug in Kapa retrieval as its knowledge tool](/examples/embed-an-ai-assistant-in-your-app-that-answers-questions-and-takes-actions.md)

[Add documentation search to your MCP serverServe documentation search from the MCP server you already ship](/examples/add-documentation-search-to-mcp-server.md)

[Draft replies for your support team in ZendeskA drafted, knowledge-grounded reply on every new ticket, as an internal comment](/examples/draft-replies-for-your-support-team-in-zendesk.md)

[Connect internal knowledge to ChatGPT and ClaudeLet your team ask about company knowledge inside the assistant they already use](/examples/internal-knowledge-chatgpt-claude.md)

[Build a support triage agent with Claude Managed Agents and KapaAn agent for support that checks your knowledge base, telemetry, and your codebase to decide whether a bug report needs engineering](/examples/build-a-support-triage-agent-with-claude-managed-agents-and-kapa.md)

[Build a documentation chatbot with Mastra and KapaExtend Mastra's docs-chatbot template with your real knowledge base, citations included](/examples/build-a-documentation-chatbot-with-mastra-and-kapa.md)

Browse the full catalog in [Examples](/examples/.md).