Google’s A2UI 1.0 ships native agent UIs

A2UI is a serious attempt to move agent experiences beyond chat, but its real-world success will depend on renderer coverage and the quality of each app’s trusted component catalog.

• –Declarative rendering creates a stronger security boundary than executing agent-generated frontend code.
• –Bidirectional function calls make A2UI useful for interactive workflows, not just generated screens.
• –Native rendering preserves platform conventions while giving developers a shared protocol across clients.
• –A2UI Express and decision-model support target the biggest weakness of generative UI: latency and token-heavy JSON generation. [Performance update](https://developers.googleblog.com/fast-reliable-a2ui/)
• –The protocol’s Apache 2.0 license and integrations with agent ecosystems give it a credible shot at becoming shared infrastructure. [Project repository](https://github.com/a2ui-project/a2ui)