Weekly Google AI Roundup: Agent Standards and Tooling Updates
AI with Suryago watch the original →
the gist
Google released four open agent standards, transitioned Gemini CLI to Antigravity CLI, and reported significant real-world AI applications in medical diagnostics and UK housing development.
Open Agent Standards Suite
Google released four major updates to the agent protocol stack, focusing on interoperability and resource discovery.
- A2A (Agent-to-Agent): The protocol for cross-company agent communication turned one year old, with over 150 organizations now utilizing it for production workflows.
- A2UI + MCP Integration: Google detailed three architectural patterns for combining Agentic User Interfaces (A2UI) with Model Context Protocol (MCP) servers: rendering UI directly from MCP tools, embedding MCP apps within A2UI components for stateful interactions, and packaging A2UI renderers inside legacy MCP apps.
- Agentic Resource Discovery (ARD): A new specification enables agents to programmatically discover tools, data sources, and other agents within an ecosystem based on intent.
- Open Knowledge Framework (OKF): A new standard designed to facilitate the discovery and access of knowledge stored across disparate data sources.
Tooling and Real-World Applications
- Antigravity CLI: The Gemini CLI has been officially deprecated and replaced by the Antigravity CLI. The migration process automatically detects and imports existing Gemini CLI configurations, including MCP servers, agent profiles, and memory data.
- Medical AI Performance: Google’s AMIE model achieved physician-level performance in disease management, matching or exceeding 21 primary care physicians in plan preciseness and guideline alignment according to research published in Nature.
- Housing Development: DeepMind is partnering with the UK government to automate planning decisions, resulting in a 50% reduction in processing times for housing approvals.