Generative UI

How CopilotKit lets agents drive UI. Learn the primitives it ships, what each one is for, and how to pick the right one.

Generative UI lets your agent decide what appears on the screen. The agent can render a component you built, compose a layout from a catalog, or embed sandboxed UI from an MCP server.

What CopilotKit ships

Seven primitives. Each one solves a different problem.

PrimitiveWhat it does
Components as ToolsRegister an application component as a frontend tool; the agent calls it and CopilotKit renders it inline with typed inputs.
Tool Call RenderingMap your agent's existing backend tool calls to custom UI cards showing live status, arguments, and results.
State RenderingSubscribe to the agent's streamed state and re-render UI as values arrive.
ReasoningRender the model's reasoning tokens inline as a first-class message type (default card, or fully custom).
A2UIRender UI from a declarative schema the agent emits, composed against a catalog you register. Two flavors: Dynamic Schema (LLM generates the schema) and Fixed Schema (you author the schema, agent supplies data).
MCP AppsEmbed UI that an MCP server ships alongside its tools, rendered in a sandboxed iframe. No frontend renderer required.
Open Generative UIThe agent generates sandboxed HTML, CSS, and JavaScript that streams live into the chat.

Pick one

Match the row that sounds like your situation:

You want to…Use
Let the agent render a specific application component you've already builtComponents as Tools
Brand the cards CopilotKit draws for your agent's existing backend toolsTool Call Rendering
Update UI as the agent's state changes (progress, drafts, dashboards)State Rendering
Surface the model's thinking chain in the chatReasoning
Let the agent compose layouts from a catalog you defineA2UI: Fixed Schema if the surface is known, Dynamic Schema if it isn't
Embed someone else's MCP-hosted UI in the chatMCP Apps
Let the agent generate a complete interactive UI in the chatOpen Generative UI

How the primitives compare

These primitives sit along a spectrum from author-controlled to agent-invented. All of them use the same frontend, runtime, and AG-UI protocol, so you can choose a different primitive for each feature.

  • Controlled: you wrote the component; the agent only picks which one to render and what data to pass. Highest predictability, highest engineering cost per capability. Components as Tools, Tool Call Rendering, State Rendering, Reasoning.
  • Declarative: the agent emits a structured spec; the frontend composes from a catalog you registered. Creativity inside a guardrail. A2UI (Dynamic and Fixed Schema).
  • Open-Ended: the UI is invented outside your app (by an MCP server or by the agent) and you sandbox it. Highest expressive range, hardest to guarantee accessibility / brand / security. MCP Apps, Open Generative UI.

How is this guide?