Agent2Agent (A2A)
standardA2A · Agent2Agent Protocol
An open protocol that lets agents from different vendors discover each other, exchange messages and hand off tasks, instead of one bespoke integration per pair of agents.
If MCP standardises the vertical connection -an agent to its tools and data- A2A standardises the horizontal one: agent to agent. Google Cloud introduced it in 2025 and donated it to the Linux Foundation, which launched the Agent2Agent project at Open Source Summit North America in June 2025 with AWS, Microsoft, Cisco and Salesforce among the founding members, and a technical steering committee drawn from AWS, Cisco, Google, IBM Research, Microsoft, Salesforce, SAP and ServiceNow. In 2026 it moved to the Agentic AI Foundation, the same neutral home that hosts MCP. What that changes operationally: interoperability between agents stops being a vendor decision and becomes a layer you can govern. What it does not change: two agents being able to talk says nothing about who authorises the action, who pays for the tokens and who answers when it goes wrong.
How it differs from
- Model Context Protocol (MCP)
- MCP connects an agent to tools and data (vertical). A2A connects agents to each other (horizontal). Complementary, not competing.
- Agent orchestration
- A2A is the shared language that lets them talk. Orchestration decides which one acts, in what order, and what happens when it fails.
Examples
- A purchasing agent handing the compliance check to another vendor's agent, with no bespoke integration in between.
- An agent discovering that another agent in the organisation already does the task, and handing it over instead of rebuilding it.
FAQ
- Does A2A replace MCP?
- No. They solve different axes: MCP standardises how an agent reaches tools, data and services; A2A standardises how two independent agents discover and coordinate with each other. Since 2026 both live under the Linux Foundation's Agentic AI Foundation.
- Do I need A2A to run multiple agents?
- No, you can coordinate agents with your own integrations. A2A matters when the agents come from different vendors, or when you want to swap one out without rewriting the others. It is a decision about lock-in, not about capability.