MCP lets your ServiceNow AI agent call a tool and get one answer back. A2A is different, it allows your agent to hand over a whole task to another agent and let that agent run with it. In this episode, Sayan walks through Agent2Agent (A2A) protocol: what it is, how it's built for agent-to-agent collaboration rather than a single function call, and how ServiceNow uses it in both directions: as the agent reaching out, and as the agent being reached. Here's what's covered: • What A2A Actually Is: Google's open standard for direct, peer-to-peer collaboration between AI agents, and why it hands over structured context instead of a single request/response. • A2A vs MCP: What is difference between MCP and A2A and when to use one. • ServiceNow as the Primary Agent: how ServiceNow discovers an external agent, pulls its capabilities, and delegates the task. • ServiceNow as the Secondary Agent: registering an AI agent in Agent Studio so outside agents can find it and put it to work. If you've been following the Action Fabric series and wondering exactly where A2A fits next to everything, this is the episode that answers it. Complete Action Fabric Blog: https://www.servicenow.com/community/developer-advocate-blog/servicenow-action-fabric-a-detailed-walkthrough/ba-p/3581348 0:25 - What A2A Actually Is 0:47 - MCP as a Function Call 1:19 - MCP vs. A2A: When to Use Each 1:52 - MCP Example: Claude Pulling Knowledge Articles 2:34 - A2A Example: Planning a Trip Across Agents 3:36 - Diving Into How A2A Works 4:02 - ServiceNow as the Primary Agent 4:55 - ServiceNow as the Secondary Agent