AI_SLANG_ENTRY
What Does A2A Mean in AI?
Agent2Agent protocol: an open standard that lets independent AI agents discover capabilities, exchange messages and results, and coordinate tasks without exposing all of their private internals.
What does A2A mean?
Agent2Agent protocol: an open standard that lets independent AI agents discover capabilities, exchange messages and results, and coordinate tasks without exposing all of their private internals.
A2A is a communication pattern for AI agents. It gives agents a common way to discover capabilities, exchange task information, and coordinate without sharing all of their private internals.
Origin and usage
Introduced by Google as Agent2Agent, then donated to the Linux Foundation and adopted in agent interoperability discussions alongside MCP.
Source type: product-term. Last checked: 2026-08-26.
Official open protocol term with strong builder-community usage. The entry treats A2A as shorthand for agent-to-agent interoperability rather than a guarantee that two arbitrary agents will cooperate safely.
What A2A stands for in AI
In AI-agent discussions, A2A stands for Agent2Agent. The protocol gives separate agent systems a shared way to describe capabilities, send tasks, provide updates, and return results.
A2A has other meanings in business and technology, so context matters. This glossary entry covers the open agent-interoperability protocol, not every expansion of the acronym.
What the A2A protocol does
- Publishes an Agent Card that describes an agent's identity, endpoint, skills, capabilities, and supported input or output modes.
- Lets a client send messages and create or follow tasks handled by a remote agent.
- Supports updates and results without requiring one agent to reveal its internal memory, prompts, or tools to another.
- Provides a common interaction model for agents built with different frameworks, languages, or vendors.
A2A versus MCP
MCP standardizes how an AI application connects to tools, resources, and context. A2A standardizes how one independent agent or agentic application communicates and collaborates with another.
They can be used together: an agent may use MCP to access its tools and A2A to delegate a task to a remote agent. Neither protocol automatically supplies trust, authorization, data governance, or a good workflow design.
What A2A does not guarantee
- Two arbitrary agents do not become useful collaborators merely because they speak the same protocol.
- A2A interoperability does not remove the need for authentication, authorization, privacy controls, and audit logs.
- An Agent Card describes capabilities; it does not prove that every claim is safe, accurate, or currently available.
- Protocol compatibility does not decide which agent should own a task or whether its result can be trusted.
Examples
- Use MCP for the tools and A2A when the workflow needs another agent to take a task.
- The demo said A2A, but the hard part was still deciding which agent was allowed to touch customer data.
FAQ
What does A2A stand for in AI?
A2A stands for Agent2Agent, an open protocol for communication and task coordination between independent AI agents or agentic applications.
What is the A2A protocol used for?
It is used for agent discovery, exchanging messages, managing collaborative tasks, sending progress updates, and returning results across different agent systems.
What is the difference between A2A and MCP?
MCP connects an AI application to tools and data. A2A connects independent agents to one another so they can communicate and delegate work.
Does A2A make agents secure?
No. A2A defines an interoperability protocol, but implementers still need authentication, authorization, privacy rules, validation, monitoring, and limits on what each agent may do.