A2UI is a JSON-based streaming protocol for agent-generated user interfaces. It lets an agent describe UI surfaces, components, data models, updates, and renderer-to-agent events in a structured stream that a compatible renderer can validate and display.
A2UI is part of the broader generative UI ecosystem. Instead of sending only Markdown or prose, an agent can progressively create and update interactive UI elements. This is useful for dashboards, forms, multi-step workflows, approval screens, and rich task outputs.
The protocol separates several concerns:
- the agent emits component and data-model updates;
- the renderer validates and displays the UI;
- user actions are sent back to the agent through event messages; and
- transport bindings can map A2UI over systems such as AG-UI, A2A, or MCP.
A2UI should not give a model arbitrary control over a frontend. Renderers need component allowlists, schema validation, event constraints, permission checks, and clear provenance so users understand which agent produced the interface.
The A2UI v1.0 candidate specification describes A2UI as a JSON streaming protocol for dynamically rendering agent-provided user interfaces.
The LLM Knowledge Base is a collection of bite-sized explanations for commonly used terms and abbreviations related to Large Language Models and Generative AI.
It's an educational resource that helps you stay up-to-date with the latest developments in AI research and its applications.