IN-PRODUCT AGENT
Give every user an expert inside your product
Embed an agent that searches your knowledge base, runs queries, and takes action for users, inside your product. Ships on your frontend with ready-made components or via MCP. No backend or maintenance required.
how it works
Connect your sources
Connect 30+ sources: help centers, GitHub code, wikis, PDFs, and SDKs.
Embed in your product
Drop in ready-made components, or go headless with hooks.
Let it take action
Give the agent tools to query data, run workflows, and create resources.
Create an agent that resolves, not just responds
“Kapa started as an agent on our documentation. Now it's the interface layer between Port and everyone - customers, internal teams, even AI agents calling our API.”
Matan Grady, Produt Director
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 In-Product Agent and who is it for?
The In-Product Agent is an AI assistant you embed directly inside your own application, typically as a sidebar, panel, or full-page chat, so users get expert help in the moment without leaving your product. Unlike a simple Q&A chatbot, it pairs knowledge base search with custom tools, so it can answer "how do I" questions from your documentation and also run queries, inspect data, and take actions on a user's behalf. It is built for product, developer experience, and engineering teams at technical companies that want to reduce friction, guide users through setup and debugging, and deflect support load at the point of need. kapa.ai is used by 200+ technical and enterprise companies, so the agent is designed for complex products with deep documentation and real workflows.
How does it embed into our existing application?
The agent ships through the Kapa Agent SDK, which is frontend-first and designed to go live without building a backend. You can drop in ready-made React components like AgentChat and AgentPanel, or go fully headless with hooks and bring your own UI using the framework-agnostic core library. Your server mints a short-lived session token, and the SDK handles orchestration, streaming, and tool execution from there. Token management is automatic, including lazy fetching, caching, and refresh before expiry, so you do not have to manage that lifecycle yourself.
How do you keep answers accurate and prevent hallucinations?
Answers are grounded in your connected knowledge sources rather than generated from open-ended model memory, using retrieval-augmented generation tuned for technical and domain-specific content. Every answer includes clickable source citations, so users and your team can trace a response back to the original documentation and verify it. You can scope what the agent draws on using source groups, which keeps it focused on the right product or version and avoids mixed or stale citations. Event callbacks also surface what users ask and where answers fall short, so you can spot documentation gaps and i
Can the agent take actions in our product, and how do we stay in control?
Yes. Beyond answering questions, you define custom tools that call your own APIs, described with type-safe schemas so the agent knows what it can do. These tools execute client-side and inherit your application's existing authentication context, so the agent only ever operates with the permissions the current user already has. Read-only tools run instantly, while write operations require explicit user approval, so the user sees exactly what the agent intends to do before anything changes. This human-in-the-loop model lets the agent resolve and act within the guardrails you set, rather than acting unchecked.
How long does it take to go live, and is it secure enough for our data?
Because the SDK is frontend-first with prebuilt components and built-in knowledge base search, teams can typically go live within days rather than running a long build project. You connect your sources, embed the components or wire up the headless hooks, and add any custom tools you want the agent to use. On security, kapa.ai is SOC 2 Type II certified and GDPR compliant, and the architecture keeps tool execution and authentication on your side using session tokens and your existing auth. That means the agent can access live product data and act for users without you handing over standing credentials or moving sensitive operations outside your own environment.