What is the Agent2Agent Protocol (A2A)?

The Agent2Agent Protocol (A2A) is an open standard through which AI agents from different vendors and platforms communicate with one another and hand tasks over. Each agent describes its capabilities in a so-called agent card; other agents can use it to discover the agent, assign it work, and track the task’s progress. Google introduced A2A in April 2025 and handed it over to the Linux Foundation later that year.

Why does this matter?

Organizations are building agents in many systems at once – in the ERP, the CRM, analytics platforms. Without a shared standard, each one remains an island or needs bespoke interfaces. A2A makes agents combinable across platform boundaries and reduces dependence on a single vendor.

What does this look like in practice?

An agent in the service system notices complaints piling up for one product. Via A2A it asks a process analysis agent on another platform whether the affected orders share a delivery route. The analysis agent reports progress and returns the result — without either platform having built a dedicated interface for the other.

Not to be confused with

Not to be confused with the Model Context Protocol (MCP): MCP connects an agent to tools and data sources, A2A connects agents to each other. The two complement each other — an agent can query data via MCP and delegate to another agent via A2A.