Agent-to-Agent Protocol (A2A): What It Is

The Agent-to-Agent Protocol (A2A) is an open interoperability standard for AI agents, launched by Google on April 9, 2025 and contributed to the Linux Foundation in June 2025. Google reported more than 150 supporting organisations by late 2025. Version 0.3 (July 2025) added gRPC transport and signed agent cards.

It solves a different problem from ACP (the ChatGPT feed and checkout) or UCP (the Google commerce journey). A2A sits a layer below: how do two agents from different vendors talk to each other when they share a multi-step task? A2A standardises that conversation.

In commerce, A2A is what would let a shopper's personal agent talk to a retailer's catalog or loyalty agent through one shared integration. UCP and AP2 both reference A2A in their interoperability design.

A2A standardizes three things across agent vendors:

Discovery. An agent advertises its capabilities through a standardized agent card (a JSON descriptor of what it can do, what it expects as input, what it returns). Other agents read agent cards to find collaborators without prior integration.

Message format. Agents exchange typed, versioned messages. v0.3 added gRPC support alongside HTTP for higher-throughput agent-to-agent flows. Tasks have explicit lifecycle states (pending, in-progress, completed, failed) so calling agents can track progress.

Security card signing. Agent cards are cryptographically signed. A receiving agent can verify the calling agent is who it claims to be before delegating sensitive operations.

The Linux Foundation handover in June 2025 turned A2A from a Google project into a vendor-neutral standard that cloud providers, enterprise software vendors and AI companies could govern together.

The four open protocols sit at different layers:

Example: a shopper's assistant needs to check a retailer's loyalty balance held by an agent on a different vendor stack. A2A is the common ground for that handoff, so the retailer integrates once for every customer-side agent vendor.

Most retailers receive A2A through the agent platforms they already use. Their agents (customer service, internal tools, procurement) get A2A support through the platform they run on, the way they get HTTP through their web framework.

What matters at the retailer level:

1. Lower lock-in. A retailer keeps its choice of AI vendor open. A2A lets the underlying agent vendor change without rebuilding integrations.
2. Cross-surface consistency. Internal agents (for store associates, for example Walmart's associate AI training) and external shopping agents can interoperate without custom bridges.
3. Foundation for AP2 and UCP. Both reference A2A primitives, so retailers adopting either inherit compatibility.

A2A is worth tracking and rarely worth building. What every agent on top of it will read is the retailer's product information, and that is the part the retailer controls. If you want to know how your products look to the assistants that already use these protocols, start with the free AI readiness report or see how Paz handles the product data work.