CopilotKit AG UI: The Protocol Every AI Agent Builder Needs

The CopilotKit team presents their breakthrough AG UI protocol that's solving the critical infrastructure gap between AI agents and user-facing applications, launched just six weeks ago to hundreds of thousands of installs and emerging as the de facto standard for building agentic applications. Co-founder Atai and founding engineer Tyler demonstrate how AG UI transforms the traditional M×N integration problem into M+N by providing a lightweight, event-based protocol that connects any agent framework to any client interface. The protocol enables seamless interoperability between major frameworks like LangGraph, CrewAI, Mastra, and custom solutions while supporting everything from web applications to mobile, voice, and Slack integrations. During live demonstrations, the team showcases building blocks including frontend tool calls, human-in-the-loop collaboration, generative UI with real-time state updates, and complex applications like AI research assistants that mirror Perplexity's functionality. CopilotKit has achieved remarkable traction with 22,000 GitHub stars, tens of thousands of weekly installs, and partnerships spanning Silicon Valley unicorns to Fortune 100 companies, positioning itself as the leading infrastructure for user-facing agents. The AG UI protocol addresses the reality that truly successful AI agents require user interactivity and collaboration rather than full autonomy, enabling applications like Cursor, Windsurf, and V0 that allow users to steer agents effectively.
💥 Highlights:
AG UI protocol launched six weeks ago with hundreds of thousands of installs
22,000 GitHub stars and double-digit month-over-month growth in adoption
First-party integrations with LangGraph, CrewAI, Mastra, AutoGen, LlamaIndex, Pydantic AI
Event-based protocol supporting 16 standard events for agent-user communication
Frontend tool calls enabling agents to manipulate UI elements and render components
Human-in-the-loop collaboration with real-time state synchronization across frameworks
Generative UI capabilities for dynamic interface creation based on agent outputs
CLI and Next.js clients with instant setup using npx create-agui-app command
Open governance protocol designed for ecosystem-wide adoption and contribution
Live demonstrations of AI research assistant and multi-framework compatibility
Partnership pipeline including AWS, Google, and major agent framework providers
User-facing agent infrastructure serving Fortune 100 and Silicon Valley unicorns
🎙️ Presenters: Atai Barkai (Co-founder & CEO) and Tyler Slaton (Founding Engineer)