Google releases A2UI 1.0 for native interfaces from AI agents
Google has released A2UI 1.0, the first stable version of its open protocol for generative user interfaces. The protocol lets AI agents respond with interactive screens, including forms, charts and cards, instead of relying only on text.
- A2UI 1.0 is Google's first stable release of its open protocol for generative UI.
- The protocol lets agents return interactive screens rendered with an app's native components.
- Version 1.0 adds bidirectional function calls, accessibility support and metadata extensions.
- Open-source renderers support web, Flutter, iOS and Android platforms.
- Atlassian, Alation, ByteDance and Google teams are building with A2UI.
A protocol for interactive agent responses
Google says A2UI is designed to address a limitation of current agent interactions: text is often a poor format for comparing products, reviewing plans or completing multi-step tasks. With A2UI, an agent describes what should appear, while the application renders the interface using its own native components.
The company says applications receive A2UI messages and render the content securely and at native speed. Developers can use open-source renderer SDKs for the web, Android and iOS, as well as agent SDKs for Python, TypeScript and Dart. Companies including Atlassian, Alation, Bloomberg and ByteDance are already building with the protocol.
Google describes A2UI as a way to create interfaces personalized to a user's needs while preserving a company's design system. Agents can place and configure only the components that an application registers.
What A2UI 1.0 adds
The 1.0 release adds bidirectional function calls, allowing components such as flight or reservation pickers to check availability during an interaction. It also adds accessibility support and extensions for additional metadata.
Open-source renderers are available for React, Angular and Lit on the web, Flutter on mobile, web and desktop, SwiftUI on iOS, and Jetpack Compose on Android. The Jetpack Compose renderer currently supports version 0.9.1, with version 1.0 coming soon.
Google has also released TypeScript and Dart SDKs alongside Python support. These SDKs simplify integration with agent frameworks by generating skills, translating messages and correcting errors. The company says swappable inference formats can reduce latency, while A2UI Macros in Python support programmatic component expansion and data binding.
Early applications across agents and platforms
Atlassian is bringing A2UI to Jira so agents can ask teams for missing details or decisions through interactive cards inside work items. Alation uses it for feeds containing metric cards, charts and other components, with charts able to show the SQL query and data behind each figure. The same feed can appear in Claude and in Alation's Data Products App.
ByteDance released an A2UI renderer for its Lynx cross-platform framework in June 2026. Its Lynx GenUI Playground demonstrates A2UI messages rendered natively as a Kyoto trip planner. Google also says Gemini Notebook uses cards that let users select presentation options, topics and visual styles before generating a slide deck.
A2UI screens can run inside MCP Apps, enabling them to appear in AI assistants that support MCP Apps, such as Claude and Gemini. Developers can read the documentation and quickstart at a2ui.org, choose a renderer, and discuss the project on GitHub. Google says support for Jetpack Compose version 1.0 and further design-system primitives are on the roadmap.
Google for Developers: A2UI 1.0: Build delightful, fast, reliable, agentic experiences- Google Developers Blog
