Welcome to A2A, the Agent2Agent protocol, built in partnership with Google Cloud and IBM Research. In this course, you'll build and run agents that communicate through the A2A or Agent2Agent protocol. There used to be multiple competing protocols for agent-to-agent communication. But since IBM and Google joined forces, A2A is emerging as the leading industry standard. In this course, you configure sequential and hierarchical workflows of multiple agents that collaborate through a unified interface provided by A2A. I'm delighted that the instructors for this course are Holt Skinner and Ivan Nardini from Google Cloud, and returning instructor Sandi Besen from IBM. Thanks Andrew. We're excited to work with you on this course. To build a team of agents that collaborate through A2A, you can wrap your agent in an A2A server and publish an agent card, which is a JSON document that describes the agent's capabilities, endpoint, and authentication requirements. an A2A client, which can itself be part of another AI agent or any other software application, then discovers the A2A server and reads that contract to determine how to structure requests to the agent to execute. This standardized approach helps teams build agents independently and connect them seamlessly. Take for example, a multi-agent system for customer support. Maybe one team will build a logistics agent to answer questions related to order status, and a different team might build a RAG agent to answer general questions about the products. Each of these agents could even be built with a different framework and then exposed to an A2A server. An orchestrator agent can process customer inquiries and route them to the specialized agent using A2A. Since the communication happens through a standardized client-server interface, each team can update their agents without the others needing to also change their code. This standardization also makes it easy to integrate existing agents into a multi-agent workflow. Many frameworks like Google Agent Development Kit, Microsoft Agent Framework and BeeAI have built-in A2A support, making it simple to expose or connect to A2A agents. A2A is an open source and community governed under the Linux Foundation. Google launched A2A in April 2025, and IBM's agent communication protocol joined forces with A2A shortly after. And while A2A is decoupled from MCP, A2A and MCP are complementary protocols. MCP standardized how an agent connects to its tools, APIs, and resources to get information. A2A standardized how an agent connects to peer agents. In this course you'll build a healthcare multi-agent system consisting of three remote agents. an insurance agent to answer coverage questions using Claude on Vertex AI. a health research agent to search for health-related information using Google Agent Development Kit and a doctor matching agent to find doctors using LangGraph. You'll learn how to wrap each agent into an A2A server and how to build a simple A2A client that you can use to communicate to any A2A server. You'll also see how to connect to A2A servers using clients from Microsoft Agent Framework and Google ADK. You'll then use Google Agent Development Kit to chain the insurance and research agents sequentially and connect the doctor matching agent to an MCP server for doctor data. After that, you'll orchestrate all three agents into an agentic system using the BeeAI framework. Instead of hard coding and chaining the agents, you'll dynamically hand off tasks to specialized agents using the requirement agent in the BeeAI framework. And finally, you'll learn how to deploy, run, and import A2A agents using Agent Stack, which is open source, self-hostable infrastructure for teams needing fast deployment of agents, framework flexibility and data control. Many people have worked to create this course. I'd like to thank the A2A Technical Steering Committee from IBM Research, the BeeAI and Agent Stack team, and from DeepLearning.AI, Hawraa Salami. A2A standardizes communication between agents that could be built by the same developer, by different teams within the same organization or even by teams across different organizations. It helps us to standardize multi-agent collaboration, and that's what you'll learn more about in the first lesson with Holt. Let's go to the next video to get started.