A2UI 1.0 lets agents answer with native screens

Google released A2UI 1.0 on November 11, 2026: the first stable version of its open protocol for generative UI. It is not a new chatbot surface. An agent describes an interactive screen — a form to confirm, a chart, a review card — and the host application draws that screen with components it has registered.

Vishal Motwani, A2UI PM, and the Google A2UI Team published the announcement on the Google Developers Blog. Google Cloud Tech posted the same day that 1.0 was out, calling it faster, easier, and stable. The blog is the account of what froze at 1.0, and of what is still on 0.9.x.

The agent describes; the host draws

The post frames current agent chats as a terminal era: capable, and answering almost everything in text. A wall of text is a poor way to compare products or to edit and approve a plan. With A2UI, the agent describes what to show. The application receives A2UI messages and renders them securely, at native speed, with its own components and its own design system. Agents place and configure only the components the host registered.

The security model is server-side. Google says A2UI binds to data and child components on the server, so sensitive values never reach the model, and the app streams the UI in as the agent works. Models can compose the interface with A2UI Express, or a "system one" decision model can pick components instead of writing text.

Developers build the agent and choose the platforms it works with. The protocol describes each view, can combine components from different apps in one view, and leaves drawing to native renderers. The blog's sick-day and dinner-rescheduling stories are illustrations of that pattern, not products shipped with this release. Users are meant to get the same native fidelity whether they reach an agent in a company's own app or in a third-party assistant that hosts MCP Apps, such as Gemini or Claude.

What 1.0 actually stabilizes

Version 1.0 adds bidirectional function calls, so a flight or reservation picker can check availability on the spot. The protocol also adds accessibility support and extensions for additional metadata.

Open-source renderers draw A2UI screens with React, Angular, and Lit on the web, and with Flutter on mobile, web, and desktop. Get started lists SwiftUI on iOS with those renderers, and applies "supports 0.9.1 today, with 1.0 coming soon" only to the Jetpack Compose renderer. What's next names only Jetpack Compose support for 1.0 as coming soon. Web and Flutter renderers also work inside MCP Apps, so an A2UI screen can appear in assistants that support MCP Apps, including Claude and Gemini, drawn by the host rather than parsed out of free-form text. Alation already uses both to show its A2UI feeds in Claude. A June post, referenced here, covers three ways to combine A2UI and MCP Apps.

New agent SDKs for Python, TypeScript, and Dart are meant to sit beside an existing harness or framework. They generate skills, translate messages, and correct errors. Swappable inference formats are there to reduce latency. In Python, A2UI Macros add programmatic component expansion and data binding.

Who is on 1.0, and who is still adapting

Google names Atlassian, Alation, Bloomberg, and ByteDance as companies already building with A2UI. The post describes three of them. What Bloomberg is building is not specified.

Atlassian is bringing A2UI to Jira, where teams can assign work items to AI agents. When an agent needs a missing detail or a decision, it asks with an interactive card in the work item, and the team answers there.

Alation's Intelligence Operating System builds feeds — metric cards, charts, and more — with A2UI. Each chart can show the SQL query and the data behind a figure. Ask Claude for a feed about profitable products and the feed renders in the conversation; the same information can be viewed inside Alation with a nearly identical layout. GT Volpe, vice president of product management at Alation, said A2UI provides "the interactive UI scaffolding that makes these artifacts feel like high-touch, tailored applications, whether they are used in Alation or in the AI Assistants our customers already use."

ByteDance released an A2UI renderer for Lynx, its cross-platform framework, in June 2026. Lynx keeps streamed screens smooth and ships a default component set on its lynx-ui library. In the Lynx GenUI Playground, A2UI messages render natively; the example in the post is a two-day trip planner for Kyoto. Lynx supports A2UI 0.9 today and is adapting to 1.0 with the A2UI team. Shouqun Liu, engineering lead for Lynx at ByteDance, said the protocol "separates how a model produces an interface from how an app renders it."

Google's own products are split the same way. Ask Gemini Notebook for a slide deck and it replies with a card that asks two things: presenter slides or a detailed deck, and which topics to focus on. A review card follows with format, focus, and visual style, which you can edit before it generates the deck. The same cards guide Video Overviews on the web and on mobile. Organizations can register their own agents that respond with A2UI. Gemini renders those responses from its component gallery and supports A2UI 0.9.1 today, not 1.0. Agents in Gemini Enterprise can already use A2UI, and Google says it is working on upgrading Gemini Enterprise to 1.0. Other Google teams, including some behind large products, are building A2UI features that have not launched.

The post says the roadmap covers design-system primitives that render natively on every platform and official SDK support for decision models.

If you are wiring an agent to a host UI, start with the 1.0 quickstart at a2ui.org and a 1.0 renderer: React, Angular, Lit, Flutter, or SwiftUI. Coming from 0.9, read the evolution guide, and do not assume Jetpack Compose speaks 1.0. The CopilotKit A2UI Playground is the fastest way to watch a described screen get drawn with registered components.

Sources

• A2UI 1.0: Build delightful, fast, reliable, agentic experiences developers.googleblog.com
• Google Cloud Tech on A2UI 1.0 x.com