When to use Agent-to-Agent (A2A) protocol
Key Takeaway
Use A2A when an Agentforce agent must interoperate with an agent on Google, Microsoft or any other A2A-compliant platform, in either direction.
The Agent-to-Agent (A2A) protocol is an open standard that lets AI agents from different vendors or platforms find each other, hand off tasks, and work together. In Agentforce, you use A2A when an Agentforce agent needs to collaborate with an agent on another platform, such as Microsoft or Google. A2A sits beside MCP (Model Context Protocol): MCP connects an agent to tools and APIs, while A2A connects an agent to other agents.
Protocols, architectures and Agent Gateway
Multi-agent terms fall into three separate groups:
| Terms | Group | A simple way to remember it |
|---|---|---|
| MCP and A2A | Protocols: the language agents speak | How agents talk |
| SOMA (Single-Org Multi-Agent) and MOMA (Multi-Org Multi-Agent) | Architectures: how agents are organized | Where agents live |
| Agent Gateway | The governance control plane | The gatekeeper |