How Does A2UI Compare?¶

The agentic UI space is evolving rapidly. Here's how A2UI relates to the other major approaches.

At a Glance¶

	A2UI	MCP Apps	AG UI
Approach	Declarative component blueprints	Pre-built HTML via `ui://` URIs	High-bandwidth protocol connecting backends to frontends
Rendering	Native components (Angular, Flutter, Lit, etc.)	Sandboxed `iframe`	Developer-defined (any framework)
Styling	Host app controls — inherits design system	Isolated — remote server controls appearance	Developer controls — part of the host app
Security	Declarative data, no code execution	Sandboxed iframe isolation	Trusted code within your own app
Multi-agent	✅ Across trust boundaries	✅ Multiple MCP servers	⚠️ Primarily single-agent
Cross-platform	✅ Web, mobile, desktop, native	⚠️ Web-focused (iframe)	✅ Protocol is framework-agnostic
LLM generation	✅ Designed for streaming output	❌ Pre-built by server	✅ Via A2UI integration
Spec	Open protocol (Apache 2.0)	MCP extension (SEP-1865)	Open source (by CopilotKit)

A2UI vs MCP Apps¶

MCP Apps treat UI as a resource — servers provide pre-built HTML via ui:// URIs, rendered in sandboxed iframes. The remote integration controls all content and appearance, with configuration happening through tool calling. A2UI takes a declarative UI approach — agents send component blueprints, but the host application controls styling, theming, and how those components are configured and rendered. Choose MCP Apps when the server should own the full UI experience; choose A2UI when you want dynamic, cross-platform UI that fits naturally into your app.

A2UI vs AG UI / CopilotKit¶

AG UI is a transport protocol connecting agent backends to frontends with real-time state sync. A2UI is a UI format — the payload that describes what to render. They're complementary: use AG UI as the pipe, A2UI as the content. AG UI is a project by the CopilotKit team, who also contributed the A2UI Composer. AG UI has day-zero A2UI compatibility.

A2UI vs ChatKit (OpenAI)¶

ChatKit offers a tightly integrated experience within the OpenAI ecosystem. A2UI shares some design philosophy with ChatKit — both define a set of basic components and use a configurable, declarative abstraction layer. A2UI is platform-agnostic — designed for developers building their own agentic surfaces across web, mobile, and desktop, or for multi-agent systems where agents need to render UI across trust boundaries.

Using Them Together¶

These approaches are complementary, not competing:

A2UI + AG UI — AG UI as transport, A2UI as the generative UI format
A2UI + A2A — A2UI messages sent via the A2A protocol for multi-agent systems
A2UI + MCP — Upcoming bridge lets MCP servers provide A2UI blueprints alongside HTML resources