Answers from the documentation; checking which model is on duty
Hi - I answer from the OpenSmartRoute documentation: routing, the API, plans and quotas, self-hosting. Ask away, or open a support ticket if you need a person.
Grounded in the docs - follow a source before acting on it.
A2UI 1.0: Open protocol for interactive, native, cross-platform genera - OpenSmartRoute
A2UI 1.0: Open protocol for interactive, native, cross-platform generative UI
Google AI releases A2UI 1.0, an open protocol enabling agents to display interactive, native UI across platforms, improving user experience and developer flexibility.
Key points
A2UI 1.0 adds bidirectional function calls for real-time checks
Introduction - Announcing A2UI 1.0 and its purpose
Google AI has released A2UI 1.0, a new open protocol for building interactive user interfaces with AI agents. This protocol allows agents to display native, cross-platform screens that users can interact with. It aims to improve how users experience AI by making interactions faster, more reliable, and more natural. The release marks the first stable version of this protocol, opening new possibilities for developers and organizations.
A2UI 1.0 is designed to replace simple text responses with rich, interactive views. It enables agents to show forms, charts, and other visual elements that adapt to each user. This makes AI interactions more engaging and easier to understand. The protocol supports a variety of platforms and programming languages, making it flexible for many types of applications.
What is A2UI - Open protocol for generative UI with interactive screens
A2UI is an open protocol that lets AI agents generate user interfaces with interactive screens. Instead of replying with just text, an agent can describe what to show on the screen. The application then draws the interface using native components. These components can be forms, buttons, charts, or other visual elements.
The protocol is designed to be flexible and extendable. It allows the agent to describe complex views that combine data from different sources. The app then renders these views securely and at native speed. This approach ensures that the UI feels natural and responsive, matching the look and feel of each platform.
A2UI 1.0 supports multiple platforms through open-source renderer SDKs. For web development, developers can use SDKs for React, Angular, and Lit. On mobile, there are SDKs for Flutter, SwiftUI, and Jetpack Compose. These SDKs enable native rendering of A2UI screens on different devices.
Developers can also use agent SDKs for programming languages like Python, TypeScript, and Dart. These SDKs help integrate A2UI into existing AI frameworks and tools. They simplify tasks like generating UI components, translating messages, and handling errors. The SDKs are designed to make it easy to build cross-platform AI agents that deliver consistent user experiences.
Google AI Edge Gallery now supports accurate, multilingual speech recognition and natural speech synthesis entirely on device. These models run offline with low latency.
New features in version 1.0 - Bidirectional calls, accessibility, metadata
A2UI 1.0 introduces several new features to enhance its capabilities. One key addition is bidirectional function calls. This allows the UI to check real-time data, such as availability or status, during interactions. For example, a flight booking screen can verify seat availability instantly.
The protocol also adds support for accessibility. This ensures that screens are usable by people with disabilities. Developers can include metadata to describe UI elements, making them easier to navigate with assistive technologies. These features help make AI-generated interfaces more inclusive and functional.
Renderer libraries are the tools that draw A2UI screens on different platforms. For web applications, open-source renderers for React, Angular, and Lit are available. These libraries convert A2UI messages into native web components. On mobile, Flutter supports drawing screens on Android, iOS, and desktop.
For native app development, there are renderers for SwiftUI on iOS and Jetpack Compose on Android. The current version of Jetpack Compose support is 0.9.1, with version 1.0 expected soon. The web and Flutter renderers can also work inside MCP (Multi-Channel Platform) apps, allowing A2UI screens to appear in AI assistants like Claude and Gemini.
These renderer libraries enable developers to build consistent, native interfaces across platforms. They ensure that users see familiar components, whether on a phone, tablet, or computer.
Use cases and examples - Atlassian Jira, Alation AIOS, ByteDance Lynx
Many organizations are already using A2UI to improve their AI interactions. Atlassian has integrated A2UI into Jira, where teams can assign work items to AI agents. When a follow-up is needed, the agent shows an interactive card within the work item. Teams can respond directly in Jira, streamlining collaboration.
Alation uses A2UI in its Intelligence Operating System (AIOS). It builds feeds with metric cards, charts, and data insights. These visual elements are rendered using A2UI, making data more accessible and understandable. Users can see detailed information, such as SQL queries behind charts, directly within conversations.
ByteDance released an A2UI renderer for Lynx, its cross-platform framework. Lynx supports streamed screens that are smooth and responsive. It ships with a default set of components based on its lynx-ui library. Developers can use Lynx to build trip planners, booking interfaces, and other interactive apps that render natively on any device.
These examples show how A2UI can be used in real-world applications to create engaging, native user experiences. They demonstrate the protocol’s flexibility across industries and platforms.
Integration and registration - How organizations can adopt A2UI
Organizations interested in adopting A2UI can register their own agents that respond with A2UI messages. Google provides documentation on how to do this, including hosting agents on cloud platforms like Cloud Run. Once registered, agents can send interactive screens to users across different apps and services.
Developers can start experimenting with A2UI using the Playground provided by CopilotKit. This environment allows testing and prototyping A2UI messages before deploying them in production. The protocol is designed to be easy to integrate into existing AI systems, with SDKs and libraries that simplify development.
By registering their agents, organizations can deliver richer, more interactive AI experiences. They can also leverage the protocol’s extensibility to add custom components and metadata, tailoring interfaces to their needs.
Why it matters - Benefits for speed, trust, native fidelity
A2UI 1.0 improves AI interactions by making them faster and more reliable. Native rendering ensures that interfaces look and feel like the platform they run on. This increases user trust and comfort, as interfaces match familiar design patterns.
The protocol also supports real-time data checks and accessibility features. These enhancements make AI interactions more trustworthy and inclusive. Developers can build interfaces that adapt to each user’s needs, providing a more personalized experience.
Overall, A2UI helps bridge the gap between AI-generated content and native user interfaces. It enables AI agents to deliver visual, interactive experiences that are seamless and trustworthy across platforms.
How OpenSmartRoute helps - Routing requests to A2UI-enabled agents
OpenSmartRoute is an open-source router for AI requests. It directs each request to the best-fit model, agent, or tool from a catalog defined by the team. The router scores candidates based on quality, cost, speed, and safety. Teams can set weights for each factor to prioritize their needs.
When a request arrives, OpenSmartRoute pins it to rules such as data privacy or cost caps. It learns from outcomes, giving more traffic to well-performing models and less to poor ones. New models can join the catalog easily, competing on the next request without changing the app.
For teams using A2UI, OpenSmartRoute can route requests to agents that respond with interactive screens. This setup ensures users get fast, reliable, and native-like interfaces. It also helps control costs and maintain security by adhering to rules and constraints.
How it compares - what existed before, what this changes and what stays the same
Before A2UI 1.0, AI agents mostly answered with text. Users had to read long paragraphs or ask follow-up questions. This made interactions slow and hard to use for complex tasks.
Developers used different methods to build interfaces. Some relied on custom code or simple chat windows. These solutions often lacked consistency across platforms and did not feel native. They also required extra work for each app or device.
A2UI changes this by providing a standard protocol for interactive screens. Instead of just text, agents can now send UI descriptions that apps render natively. This makes interfaces faster, more reliable, and more familiar to users.
What stays the same is the core idea of AI agents answering questions. The difference is how they present information. A2UI adds a layer of interactivity and visual design that was missing before. It also supports multiple platforms with the same interface description.
A2UI is compatible with existing AI models and tools. It does not replace the models but enhances how they communicate. Developers can continue using their favorite models and frameworks. They just add A2UI to make the output more engaging and usable.
This approach builds on the progress made in text-based AI. It extends the capabilities to include rich, native interfaces. It also sets a foundation for future features like real-time updates and accessibility support.
Questions this leaves open - what the source does not say and how a reader can check it
The source does not specify how widely A2UI is adopted. It mentions some companies using it, but not the total number of users or deployments. It also does not detail how easy or difficult it is to integrate A2UI into existing systems.
Another open question is how well A2UI performs in terms of latency and responsiveness. The source states it is fast and reliable, but does not provide benchmarks or performance metrics. Readers interested in speed should test A2UI in their own environments.
The source mentions support for accessibility and extensions, but does not specify what features are included. For example, it does not say if screen readers or keyboard navigation are supported out of the box. Developers can verify this by reviewing the protocol documentation and testing with accessibility tools.
It also does not detail security considerations. Since A2UI involves transmitting UI descriptions, it is important to understand how data privacy and security are maintained. Teams should review the security model and conduct their own assessments.
The source highlights SDKs for web, Android, and iOS, but does not mention other platforms like desktop or embedded systems. It is unclear if A2UI can be extended to these environments. Developers can explore the open-source renderer libraries to see if they fit their needs.
Finally, the source does not specify how future updates or extensions will be handled. As the protocol evolves, compatibility and backward support will be important. Teams should follow the development of A2UI and participate in community discussions to stay informed.
To check these aspects, developers can review the official documentation, join community forums, and run pilot projects. Testing A2UI in their specific use cases will reveal its strengths and limitations.
