Short CourseIntermediate1 Hour 41 Minutes

ACP: Agent Communication Protocol

Instructors: Sandi Besen, Nicholas Renotte

IBM Research's BeeAI
  • Intermediate
  • 1 Hour 41 Minutes
  • 12 Video Lessons
  • 8 Code Examples
  • Instructors: Sandi Besen, Nicholas Renotte
    • IBM Research's BeeAI
    IBM Research's BeeAI

What you'll learn

  • Build ACP-compliant agents by wrapping them in an ACP server, launch the server to activate the agents, and make them discoverable by ACP clients to enable easy integration within multi-agent systems.

  • Chain ACP-compliant agents in linear and hierarchical workflows; use a router agent to delegate tasks to the specialized agents.

  • Import ACP-compliant agents into a registry to make them easy to discover and share across teams.

About this course

Introducing ACP: Agent Communication Protocol, a short course built in partnership with IBM Research’s BeeAI and taught by Sandi Besen, AI Research Engineer & Ecosystem Lead at IBM, and Nicholas Renotte, Head of AI Developer Advocacy at IBM.

Building a multi-agent system with agents shared across teams and organizations can be challenging. You may need to write custom integrations each time a team updates their agent design or changes the agent’s framework. The Agent Communication Protocol (ACP) is an open protocol that addresses this challenge by standardizing communication between agents. It provides a unified interface through which agents can collaborate regardless of their frameworks, making it easy to replace an agent with a new version without needing to refactor the entire system.

In this course, you’ll learn to connect agents through ACP. The protocol is based on a client-server architecture: you host an agent built with any framework inside an ACP server, and send requests to the server through an ACP client. You’ll learn how to wrap an agent inside an ACP server and set up an ACP client to connect to the server. You’ll build sequential and hierarchical workflows of agents hosted inside ACP servers, and learn how to manage this workflow on the client side through a process or another agent. 

In detail, you’ll:

  • Learn the underlying architecture of ACP and how it enables agents built with different frameworks to work together through a common interface.
  • Understand the lifecycle of an ACP Agent (configuration, activation, discovery, execution), and how it compares to other protocols, such as MCP (Model Context Protocol) and A2A (Agent-to-Agent).
  • Build a RAG agent with CrewAI and wrap it inside an ACP server.
  • Define an ACP Client to make calls to the ACP server you created.
  • Define another ACP server, built with Smolagents, and sequentially chain it to the RAG agent.
  • Build a hierarchical workflow using a router agent that transforms a user’s query into tasks and delegates the tasks to agents available through ACP servers.
  • Build an agent that uses MCP to access tools and ACP to communicate with other agents.
  • Import your ACP agents to the BeeAI platform, an open-source registry to discover and share agents easily. 

By the end of the course, you’ll know how to create ACP-compliant agents that can communicate regardless of their frameworks and collaborate to address queries.

Who should join?

This course is perfect for AI builders who want to easily reuse and connect multiple agents built with different frameworks in a single system, or for anyone curious to learn about the Agent Communication Protocol. Some experience with Python is recommended.

Course Outline

12 Lessons・8 Code Examples
  • Introduction

    Video4 mins

  • Why Agent Communication protocol

    Video5 mins

  • ACP Core Principles

    Video8 mins

  • Building a RAG Agent with CrewAI

    Video with code examples13 mins

  • Wrapping the RAG Agent into an ACP Server

    Video with code examples8 mins

  • Calling an ACP Agent using the Client

    Video with code examples5 mins

  • Wrapping a Smolagents Agent into an ACP Server

    Video with code examples8 mins

  • Sequentially Chaining the Agent Calls

    Video with code examples7 mins

  • Hierarchically Chaining the Agent Calls using a Router Agent

    Video with code examples12 mins

  • Adding MCP to the Hospital Server

    Video with code examples17 mins

  • Managing ACP Compliant Agents

    Video9 mins

  • Conclusion

    Video1 mins

  • Quiz

    Reading1 min

  • Appendix – Resources, Tips, and Download

    Code examples1 min

Instructors

Sandi Besen

Sandi Besen

AI Research Engineer and Ecosystem Lead at IBM Research

Nicholas Renotte

Nicholas Renotte

Head of AI Developer Advocacy at IBM

Course access is free for a limited time during the DeepLearning.AI learning platform beta!

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!