— Slack, MS Teams, Discord, Google Chat & Self-Learning· Early access
Get early access
The best AI product feels like the agent is the app, not a chatbot with zero context. That comes down to one layer: the interface between the agent and your user. So that's the layer we're going to dig into. That's where CopilotKit comes in.
It's a complete set of building blocks for developers and teams that want to add agents to their app and assemble, debug, and ship the interface layer between your agent and your users.
It's also framework-agnostic and supports any LLM and every popular agent framework.
There are two options when it comes to chat. Pre-built components, if you want something already built or Headless UI, which allows anyone to build custom agent interfaces.
The building blocks create the foundation. It's how the agent and the UI actually collaborate beyond just chat messages.
There are three flavors of generative UI. Pre-built static components the agent can choose from schemas it emits using A2UI, or allow the agent to generate the UI on the fly using MCP apps. Simply put, the agent can actually perform tasks in your application and take real actions within the product.
Frontend tools let your AI agent directly interact with and update your application's UI in real time.
Safely shipping agents to production comes down to one thing. The moment your agent touches sensitive data like payments, customer records, or booking a flight, you don't want it acting on its own.
CopilotKit lets you drop in a checkpoint. The agent reaches that step, surfaces what it's about to do, and waits for human approval before execution continues. Human in the loop is a security measure built in.
The UI and the agent always stay in sync via shared state. If the user edits something in the app, the agent is aware of that change. If the agent updates a record on the back end, the UI rerenders automatically; the agent never acts on stale data.
Chat isn't a requirement. Headless UI is an option. You bring the interface, and CopilotKit connects to it. Whether it's guiding a user through onboarding, filling out a form, walking through checkout, or anything else in your product, the styling, components, and layout are all yours and fully customizable.
When you go to production, you will absolutely need threads to persist. This happens with CopilotKit Enterprise Intelligence. You can watch the agent stream on multiple devices at the same time, and there's automatic reconnection built in if the network drops mid-run. Threads just pick up where you left off.
Every building block is enabled through an open standard called AG-UI, the bridge between agents and user communication, started by the Copilot Kit team and adopted by hyper scalers and agent frameworks.
Getting started is extremely easy. You can be up and running in under five minutes with one command. CopilotKit is open source, and if that's something you appreciate, a star to the repo would mean a lot.
Thank you!
Subscribe to our blog and get updates on CopilotKit in your inbox.
