Google has released A2UI 1.0, the first stable version of its open protocol for generative interfaces. Instead of answering only in text, an AI agent can now reply with an interactive screen, such as a form to confirm or a chart to compare. The release landed on November 11, 2026.
What A2UI actually does
With A2UI, the agent describes what to show, and every app draws the interface with its own native components. The agent never sends executable code, and sensitive values bind on the server, so they never reach the model.
The same agent output works on any screen. Ask an agent to move dinner with a friend to Friday, and it can answer with one view lining up the calendar, open tables, and rideshare prices.
What is new in 1.0
Version 1.0 makes the protocol stable. It adds bidirectional function calls, so a flight or reservation picker can check availability on the spot. The release also adds accessibility support and extensions for any additional metadata a developer needs to carry.
Open-source renderers now cover React, Angular, and Lit on the web, Flutter on mobile, web, and desktop, and SwiftUI on iOS. Jetpack Compose on Android still supports version 0.9.1, with 1.0 coming soon. Agent software kits arrive for Python, TypeScript, and Dart.
Who is already building with it
Atlassian is bringing A2UI to Jira, where an agent working on a task can ask the team a follow-up question through an interactive card inside the work item.
Alation builds its intelligence feeds with A2UI, so metric cards and charts render in conversation, and each chart can show the query behind it.
ByteDance released an A2UI renderer for Lynx, its cross-platform framework, in June 2026, and is adapting it to version 1.0. Bloomberg is also building with the protocol.
Inside Google, Gemini Notebook already replies with option and review cards for slide decks, and Gemini itself renders A2UI 0.9.1 today, with Gemini Enterprise still upgrading to 1.0. Google’s Antigravity agent environment is part of the same push.
A2UI and MCP Apps
A2UI works together with MCP Apps. The web and Flutter renderers run inside MCP Apps, so A2UI screens can appear in AI assistants that support them, such as Claude and Gemini. Alation already uses both to show its feeds inside Claude.
Our Read
The important shift is not any single demo. Agents graduate from chat bubbles to real interfaces, and they do it through an open protocol rather than one company’s design system.
The trust story is the design itself. The agent describes a view instead of shipping code, so the host app only ever draws components it registered.
That is a stronger guarantee than a policy document, and it is the detail enterprises will weigh most.
Even so, the ecosystem is mid-migration, not finished. Jetpack Compose still sits on 0.9.1, Gemini Enterprise is still upgrading, and Google published no latency or adoption figures.
So its claims about speed and reliability are its own word for now. The real test is whether assistants beyond Claude adopt A2UI as their default interface layer.
Why It Matters
For developers, the promise is to build a view once and ship it everywhere. For users, it means tapping to confirm a plan instead of retyping it. For Google, it is a standards play: if A2UI becomes the default way agents show their work, assistants render inside native apps.
Gemini 4 Hydrogen shipped the same week, showing how fast Google is moving its agent stack.
Background
A2UI began as a preview before reaching its first stable release. Google covered combining it with MCP Apps in June 2026, ByteDance shipped its Lynx renderer that month, and the project counted more than 17,000 GitHub stars by November 12.


0 Comments