Agentic GUI Protocols: Google's A2UI and CopilotKit's AG-UI

The main open-source projects for the Agentic GUI protocol (also called Agentic UI / Generative UI protocol) currently revolve around Google's A2UI and the AG-UI protocol from the CopilotKit team. These are the "A2UI-like" projects mentioned in the original discussion ("AI2UI" is essentially an alias or typo for A2UI).

Both protocols address the same core problem: letting AI agents safely generate or interact with rich, cross-platform UI (beyond plain text output), avoiding the security risks of directly executing HTML/code (such as XSS), while supporting native rendering across web, mobile, and desktop.

1. A2UI (Agent-to-User Interface) — Google's Declarative UI Protocol

• GitHub: https://github.com/google/A2UI (Apache 2.0 license, actively developed)
• Website: https://a2ui.org/
• Key features:
• The agent describes UI intent using safe, declarative JSON components — a whitelisted component library of about 18 base components (rows, columns, buttons, forms, etc.) with Material-style support.
• Clients render with native components (official renderers for Lit, Angular, Flutter, React, etc.), ensuring cross-platform consistency and safety.
• Designed for "agents generating dynamic interactive interfaces"; already integrated into Google Chat, Vertex AI, and other products.
• Use cases: when an agent needs to generate complex UI such as real-time forms, dashboards, or rich-text chats.
• Community support: ready-made integration examples exist for CopilotKit, LangGraph, and other frameworks.

2. AG-UI (Agent-User Interaction Protocol) — CopilotKit's Real-Time Event Protocol

• GitHub: https://github.com/ag-ui-protocol/ag-ui (MIT license)
• Website / docs: https://www.copilotkit.ai/ag-ui
• Key features:
• Lightweight, event-driven protocol (SSE streams + 16 standard event types, e.g. TEXT_MESSAGE_CONTENT, TOOL_CALL_START, UI_UPDATE).
• Handles real-time bidirectional synchronization between agent backends and frontend apps, mapping output from agent frameworks (LangGraph, CrewAI, ADK, etc.) directly to UI state.
• Highly complementary to A2UI: AG-UI manages "communication," A2UI manages "UI content format" — they are often used together.
• Use cases: building full-stack agentic applications requiring real-time streaming UI updates, tool-call feedback, and user intervention.
• Strongest ecosystem: CopilotKit (https://github.com/CopilotKit/CopilotKit) is its official implementation, adopted by LangChain, AWS, Microsoft, and others, with React/Angular SDKs and many Generative UI examples.

Related Protocols (part of the full stack, but not GUI protocols per se)

• A2A (Agent2Agent Protocol): Google-led protocol for agent-to-agent communication (https://github.com/a2aproject/A2A), used for multi-agent collaboration; can work with AG-UI/A2UI to push results to the frontend.
• MCP (Model Context Protocol): Anthropic-led tool/context protocol (https://modelcontextprotocol.io/); some extensions support MCP Apps (interactive UI), but it primarily targets tool invocation rather than pure GUI.

Getting Started

• To try A2UI directly: clone the Google repository and run the demo with Google ADK (Agent Development Kit).
• To build a full-stack app quickly: use CopilotKit + AG-UI + A2UI (official starter repos are available, e.g. with-agent-spec).

As of March 2026, the Agentic GUI ecosystem is still iterating rapidly. These two projects are the de facto standards, and nearly all mainstream agent frameworks are aligning with them.