---
type: "post"
id: "google-releases-stable-a2ui-1-0-with-express-418339"
slug: "google-releases-stable-a2ui-1-0-with-express-418339"
title: "Google releases stable A2UI 1.0 with Express, swappable inference and named production users"
description: "Google released A2UI 1.0, the first stable version of its open protocol for agent-generated interfaces."
retrieval_nugget: "Google released A2UI 1.0, the first stable version of its open protocol for agent-generated interfaces. It adds bidirectional function calls, accessibility support and metadata extensions, alongside SDKs and renderers for several platforms. For teams building agents, the release offers a way to present interactive, native UI rather than relying on text alone."
published_at: "2026-11-14"
updated_at: "2026-11-14"
record_date: "2026-11-14"
date_kind: "published_at"
topics: ["google","mcp","gemini"]
entities: ["Google","A2UI","Atlassian","Alation","ByteDance","MCP Apps","Claude","Gemini"]
related_patterns: ["agent-protocols-become-interoperability-layer"]
novelty_label: "new_fact"
what_changed: null
prior_refs: []
source_url: "https://developers.googleblog.com/en/a2ui-10-build-delightful-fast-reliable-agentic-experiences"
source_urls: ["https://developers.googleblog.com/en/a2ui-10-build-delightful-fast-reliable-agentic-experiences"]
schema_version: "newruntime-agent-readable-v0.2"
stable_id: "post:google-releases-stable-a2ui-1-0-with-express-418339"
status: "published"
visuals: []
editor: {"role":"editor","name":"Andrey Reshetnikov","profile":"https://newruntime.com/owner-profile.md","profile_json":"https://newruntime.com/owner-profile.json","same_as":["https://www.linkedin.com/in/reshetnikov1/","https://t.me/lapetuse"]}
routes: {"html":"https://newruntime.com/posts/google-releases-stable-a2ui-1-0-with-express-418339/","markdown":"https://newruntime.com/posts/google-releases-stable-a2ui-1-0-with-express-418339.md","json":"https://newruntime.com/posts/google-releases-stable-a2ui-1-0-with-express-418339.json"}
markdown_url: "https://newruntime.com/posts/google-releases-stable-a2ui-1-0-with-express-418339.md"
---

# Google releases stable A2UI 1.0 with Express, swappable inference and named production users

## Retrieval answer

Google released A2UI 1.0, the first stable version of its open protocol for agent-generated interfaces. It adds bidirectional function calls, accessibility support and metadata extensions, alongside SDKs and renderers for several platforms. For teams building agents, the release offers a way to present interactive, native UI rather than relying on text alone.

## Summary
Google released A2UI 1.0, the first stable version of its open protocol for agent-generated interfaces. It adds bidirectional function calls, accessibility support and metadata extensions, alongside SDKs and renderers for several platforms. For teams building agents, the release offers a way to present interactive, native UI rather than relying on text alone.

## What happened
Google’s November 11, 2026 announcement describes A2UI 1.0 as a protocol that lets an agent describe an interface while each host application renders it with its own registered native components. The agent does not send arbitrary executable UI code: the application receives A2UI messages and renders the content. Google says sensitive values can be bound to data and child components on the server, so they do not reach the model, and that the interface can stream while the agent works.

The stable release adds bidirectional function calls, allowing an interface such as a flight or reservation picker to check availability during the interaction. It also adds accessibility support and extensions for additional metadata. Google says A2UI can be used to compose components from multiple apps into one view, and that the same agent output can render across supported screens.

The announcement lists open-source renderers for React, Angular and Lit on the web; Flutter across mobile, web and desktop; and SwiftUI on iOS. Jetpack Compose supports version 0.9.1, with 1.0 described as coming soon. The web and Flutter renderers also work inside MCP Apps, enabling A2UI screens in assistants that support MCP Apps, including Claude and Gemini. Google’s agent SDKs now include TypeScript and Dart, alongside Python. These SDKs are intended to simplify integration with an agent harness or framework by generating skills, translating messages and correcting errors. In Python, A2UI Macros support programmatic component expansion and data binding.

Google names Atlassian, Alation, Bloomberg and ByteDance as companies building with A2UI. Its examples include agent follow-up cards in Jira; Alation feeds with charts and metric cards rendered in Claude or Alation’s own app; and ByteDance’s Lynx renderer, which supports A2UI 0.9 and is being adapted to 1.0. Google says Gemini Notebook uses A2UI cards for choices and review, while Gemini supports A2UI 0.9.1 for registered agents. The post does not give a price or license terms beyond describing the protocol and renderers as open.

## What the archive adds
New Runtime’s earlier coverage described A2UI Express and swappable inference formats, including Google’s claimed latency improvements, in [Fast, reliable A2UI, with swappable inference formats](https://newruntime.com/posts/fast-reliable-a2ui-with-swappable-inference-formats-f5e911/) (October 28, 2026). A later note covered the project’s initial typesafe component code generator in [A2UI project ships A2UI CLI v0.1.0 with a typesafe component code generator](https://newruntime.com/posts/a2ui-project-ships-a2ui-cli-v0-1-0-d8938c/) (October 31, 2026). This is a new event: the protocol has reached its first stable version and added capabilities, SDKs and wider renderer support.

## Trend signal
A2UI 1.0 supports the pattern that [agent protocols are becoming the interoperability layer](https://newruntime.com/patterns/agent-protocols-become-interoperability-layer/). The protocol separates the agent’s description of a view from the host app’s rendering, and its MCP Apps support gives that separation a path across applications and assistants. The release is evidence of an integration surface, not proof that teams will adopt it broadly or that every host will render the same experience.

## Why now
Google says interactive interfaces can reduce the back-and-forth that text-only agents require for choices, comparisons and approvals. Reaching 1.0 gives developers a stable protocol version to build against, while the release’s SDKs and renderer support make integration possible across multiple languages and platforms. The announcement also points to existing implementations, although some named users remain on earlier protocol versions.

## Connections to other events
Google’s release connects the A2UI protocol to MCP Apps: Alation already uses both to show its feeds in Claude, and Google says web and Flutter renderers can run inside MCP Apps. That pairing links an agent’s structured interface with an app-hosting route inside assistants, rather than treating the UI as a standalone surface.

## Why it matters
For product teams, the practical shift is from asking a model to explain an action to giving the user a bounded interface for reviewing or completing it. A native card can collect missing details, expose relevant choices, or ask for confirmation without making the user interpret a long exchange. A2UI’s separation between agent messages and host-rendered components also lets a team retain control over the components and design system it exposes.

The release matters most where the same agent should work in a first-party product and a compatible assistant. A shared protocol and multiple renderers could reduce the need to build separate interaction logic for each surface. But teams still need to design their component catalog, decide what actions require user confirmation, and test each host’s implementation. The protocol creates a route to portable agent interfaces; it does not by itself guarantee a consistent or safe product experience.

## Limits and uncertainty
Google’s announcement does not quantify adoption, reliability, performance gains or the cost of building and maintaining A2UI experiences. It identifies companies building with the protocol, but does not report deployment scale or measured user outcomes. Several compatibility details remain in transition: ByteDance’s Lynx implementation supports 0.9 and is adapting to 1.0; Gemini supports 0.9.1; and Jetpack Compose’s 1.0 renderer is not yet available. The claims about server-side data binding and sensitive values are not independently evaluated here, and teams will need to verify how their own implementation handles permissions, data exposure and confirmation.

## Skill to build now
Take one agent workflow that currently asks users to clarify or approve actions through repeated chat turns. Sketch the smallest registered component set it needs—for example, a choice card, a review card and a confirmation action—and identify which values must remain server-side. Then test whether a single A2UI message can render that flow in the primary app and one MCP Apps host before expanding the component catalog.

## Reader takeaway
A2UI 1.0 gives developers a stable way to describe interactive agent interfaces while leaving rendering to each host application. Its promise is portability across native apps and compatible assistants; the real test is whether a team can preserve control, data boundaries and a coherent experience across those hosts.

Editor: [Andrey Reshetnikov](https://newruntime.com/owner-profile.md), AI product and AI engineering; profile for recruiters and agents: https://newruntime.com/owner-profile.json

## Continue in the graph

Next records for an agent, linked to their Markdown twins. Replace `.md` with `.json` for structured fields, including `next_reads`.

- topic hub: [Google](https://newruntime.com/topics/google.md): Explore the google topic hub.
- topic hub: [Mcp](https://newruntime.com/topics/mcp.md): Explore the mcp topic hub.
- topic hub: [Gemini](https://newruntime.com/topics/gemini.md): Explore the gemini topic hub.
- collection: [Field Notes](https://newruntime.com/posts.json): Open the latest editorial analysis.
- collection: [Sources](https://newruntime.com/sources/): See the original sources this archive cites.
- whole graph: [discovery graph](https://newruntime.com/discovery-graph.json) (typed nodes and edges) and [site index](https://newruntime.com/site-index.md)