Short CourseIntermediate

A2A: The Agent2Agent Protocol

Instructors: Holt Skinner, Ivan Nardini, Sandi Besen

Google Cloud logoIBM Research logo
  • Intermediate
  • 14 Video Lessons
  • 8 Code Examples
  • Instructors: Holt Skinner, Ivan Nardini, Sandi Besen

What you'll learn

  • Expose agents built with frameworks like Google ADK, LangGraph, or BeeAI as A2A servers to make them A2A-compliant.

  • Create A2A clients from scratch or by using A2A integrations to connect to A2A-compliant agents.

  • Orchestrate A2A-compliant agents into sequential and hierarchical workflows.

About this course

Join this new short course on A2A: The Agent2Agent Protocol, built in partnership with Google Cloud and IBM Research, and taught by Holt Skinner, Ivan Nardini, and Sandi Besen.

Connecting agents built with different frameworks typically requires extensive custom integration. A2A provides an open protocol that standardizes how agents discover each other and communicate. Launched by Google Cloud in April 2025 and donated to the Linux Foundation, with IBM’s Agent Communication Protocol merging shortly after, A2A is emerging as the industry standard for agent collaboration.

In this course, you’ll build a healthcare multi-agent system with three specialized agents using different frameworks. You’ll wrap each agent in an A2A server, build A2A clients to communicate with them, and orchestrate them into sequential and hierarchical workflows. You’ll see how A2A complements MCP: while MCP connects agents to external data systems, A2A enables agents to collaborate with each other.

In detail, you’ll:

  • Understand why A2A standardizes agent communication and explore its client-server architecture, including agent cards and the agent lifecycle.
  • Build an insurance policy agent using Claude Haiku 4.5 on Vertex AI, wrap it in an A2A server using the A2A Python SDK, and create an A2A client to communicate with it.
  • Build a health research agent using Google Agent Development Kit and Gemini 3 Pro that performs research with Google Search.
  • Chain agents sequentially using ADK, where one agent’s output feeds into the next through A2A.
  • Create a healthcare provider agent with LangGraph, connect it to an MCP server, and interact with it using Microsoft Agent Framework’s A2A client.
  • Orchestrate multiple agents dynamically with BeeAI Framework, using the Requirement Agent to delegate tasks to specialized agents as needed.
  • Deploy your A2A agents to BeeAI’s Agent Stack , an open-source infrastructure for deploying and sharing A2A agents.

Whether you’re orchestrating agents within your organization or collaborating across different teams, this course gives you hands-on experience with the protocol standardizing how agents work together.

Who should join?

AI developers building multi-agent systems or working with agentic workflows. Familiarity with Python and basic understanding of AI agents recommended.

Course Outline

14 Lessons・8 Code Examples

Instructors

Holt Skinner

Holt Skinner

Developer Advocate, Cloud AI at Google

Ivan Nardini

Ivan Nardini

Sandi Besen

Sandi Besen

Additional learning features, such as quizzes and projects, are included with DeepLearning.AI Pro. Explore it today

Want to learn more about Generative AI?

Keep learning with updates on curated AI news, courses, and events, as well as Andrew’s thoughts from DeepLearning.AI!