OpenIntelligentUI is CopilotKit’s open-source generative UI framework: a way for AI agents to respond not just with text but with rich, interactive interfaces rendered on the fly. Instead of a chat bubble describing a chart, form or dashboard, the agent emits structured UI that the framework renders as real components the user can interact with. The project has around 3,650 stars on GitHub and is MIT licensed.
Generative UI is the pattern behind agent experiences where the model decides what the user should see next โ a data visualisation for an analytics question, a booking form mid-conversation, a diff viewer for a coding agent. OpenIntelligentUI provides the framework layer for this in JavaScript and TypeScript applications: agent-native components, streaming render of partial UI as the model generates it, and a registry approach that keeps the model constrained to components your application actually supports. It comes from the team behind CopilotKit, the popular framework for embedding copilots into React apps, and reflects the interface patterns from their research into agentic visualisation and assistant chatbots.
OpenIntelligentUI is completely free and open source under the MIT license. You bring your own model provider and pay only for the LLM usage your agents generate.
Frontend and full-stack developers building AI copilots, analytics assistants or agent products who want responses that are visual and interactive rather than walls of text, while keeping full control over the components an agent can render.