Google releases A2UI 1.0, letting AI agents show real UI

A2UI is an open generative UI protocol from Google. It lets AI agents reply with full UI screens, not just plain text. The protocol has now reached its first stable version.

Google has officially released A2UI 1.0. Developers can use it with open-source renderers for web, Android, and iOS, plus agent SDKs in Python, TypeScript, and Dart. Companies like Atlassian, Bloomberg, and ByteDance are already building on it.

Under this protocol, the agent only describes what to show, and each app draws the UI with its own native components. Whether inside a company's own app or an AI helper like Gemini or Claude that supports MCP, users get low latency and a fully native look.

Version 1.0 adds two-way function calls, plus better accessibility and metadata support. Renderers cover React, Angular, Lit, Flutter, SwiftUI, and Jetpack Compose, and some can run directly inside MCP apps.