All postsSHORT2 min read
Google's A2UI 1.0 Turns Chatbot Text Walls Into Buttons
A2UI 1.0 is Google's first stable release for agent-built screens. How it works, where it renders, and the platform versions you need to check.
Your client's chatbot answers in walls of text. Google's A2UI 1.0 lets it answer with real buttons. Google released the first stable version on Nov 11, 2026. A2UI itself dates back to its announcement on Dec 15, 2025. This week's news is the stable release.
What A2UI is
An agent can answer with an interactive screen: a form to confirm, a chart or a reservation picker. The agent describes what to show as data. Your app draws the screen with its own native components.
That's the job of A2UI, Google's open protocol for generative UI. Agents can place and configure only the components you register, so the screens use your design system. Arbitrary executable code isn't allowed in those UI messages.
Google says A2UI binds data and child components on the server, so sensitive values never reach the model. The app can stream the screen as the agent works.
It's open source under Apache-2.0 and free to use. Model and hosting costs depend on your setup.
What's new in 1.0
The first stable version adds two-way function calls. A flight or reservation picker can check availability on the spot. It also adds accessibility support and extensions for extra metadata.
If you're upgrading from 0.9, Google's post points to an evolution guide covering the changes.
Where it renders
Renderers are the pieces that draw these screens. The open-source options include:
- React, Angular and Lit for the web.
- Flutter for mobile, web and desktop.
- SwiftUI for iOS.
- Jetpack Compose for Android, currently on 0.9.1.
The web and Flutter renderers also work in MCP Apps. That lets A2UI screens appear inside assistants such as Claude and Gemini. Gemini's own A2UI support is still on 0.9.1, so check your target host's version.
Agent SDKs are available for Python, TypeScript and Dart.
Who's using it
Google names Atlassian, Alation, Bloomberg and ByteDance as companies building with A2UI.
Atlassian is bringing interactive follow-up cards to Jira work items. When an agent needs a missing detail or a decision, the team can answer in the card. Alation already uses A2UI to show feeds with metric cards, charts and custom components inside Claude.
The catch
Stable 1.0 doesn't mean every platform has reached that version. Android's Jetpack Compose renderer is on 0.9.1, with 1.0 coming soon. Gemini supports 0.9.1, and its Enterprise upgrade to 1.0 is in progress. ByteDance's Lynx supports 0.9 and is adapting to 1.0.
You still build the agent and register its components. A2UI supplies the screen protocol and renderers. Developers choose which platforms their agent works with.
Start with the docs and quickstart, the GitHub repo and Google's release post. Check the platform versions before choosing where to put the screens.
The builder angle (my opinion, no income guaranteed)
I'd find a business whose support or booking chatbot already answers in walls of text. Offer to rebuild those replies as A2UI cards using the business's own design system.
The deliverable: confirm buttons, reservation pickers and approval forms in the client's app, with screens also available in Claude through MCP Apps. Pick the components with the client and register them for the agent.
That's a proposed service. There's no income guaranteed.
Daily email
Get each breakdown in your inbox.
One email each evening, right after the Short goes live: the AI tool it covers, the full write-up with sources, and how to get paid for it. Free, and you can unsubscribe anytime.