A2A protocol just moved to a neutral foundation — worth knowing if you’re building anything that talks to other agents

Worth flagging since it’s directly relevant to anyone here building multi-agent or cross-agent setups. Google’s A2A protocol, the standard for agent-to-agent communication, just moved from the Linux Foundation’s general umbrella into the Agentic AI Foundation (AAIF), a foundation specifically focused on agentic AI standards. MCP, the protocol most of us already use for connecting agents to tools and data, is already in that same foundation. So now both halves of the stack, tool access (MCP) and agent-to-agent communication (A2A), sit under one neutral governance body instead of A2A being Google’s thing off on its own. Why this actually matters practically: AAIF went from under 40 members at its December 2025 launch to over 250 now, with Google, Microsoft, Amazon, Anthropic, OpenAI, Bloomberg, Shopify, and Block all backing it. That’s a wide enough coalition that A2A has a real shot at becoming the actual standard rather than one vendor’s protocol other people grudgingly support. The practical upside if you’re building agent-to-agent workflows: less custom glue code between agents built on different models or platforms, and more freedom to swap providers based on cost or performance without rebuilding your integration layer every time. The AAIF’s exec director put it as “companies don’t want just one protocol; they want the whole stack to be open,” which tracks with what a lot of us have been hacking around manually. Worth noting this is coming from the companies’ own framing in interviews, not independently verified outcomes, so take the “reduces integration work” claims as a goal they’re pushing toward rather than a guarantee. Still, worth watching if you’re deciding what to build your interop layer on right now. Anyone already building directly against A2A? Curious how mature the actual spec feels in practice versus how it reads on paper.