Google open-sources A2UI protocol: agents design native UIs

Google open-sourced A2UI (Agent-to-User Interface). The spec and companion libraries let AI agents declaratively describe native user interfaces for client apps to render. The project is published at a2ui.org and was covered in several outlets during Dec 20–22, 2025. (sources: 2, 3, 4, 5)

A2UI defines a simple schema and client libraries. An agent emits component descriptions, events, and interaction contracts instead of plain text. Clients render those components and handle inputs and security policies. Google presents A2UI as a cross-platform open standard. A Google Cloud update that week also described stronger governance for Vertex AI Agent Builder, showing alignment on agent governance. (sources: 2, 1)

Expect A2UI to reduce per-integration UI plumbing and enable richer agent-driven experiences on web, mobile, and desktop. Adoption depends on client-library maturity and platform security reviews. Early adopters and third-party tool vendors are likely to add renderers and integration guides as the open project gains contributors. (sources: 2, 3, 4)

Why It Matters

• Replace ad-hoc UI glue with declarative component schemas to shorten agent integration time.
• Client teams keep rendering control and enforce security policies because A2UI separates interface description from execution.
• Open client libraries speed prototyping across web, mobile, and desktop and enable vendor-agnostic agent UIs; expect third-party renderers and guides to appear quickly.