An Illustrated Guide to AI Agents

Chapter 8. Multi-Agent Systems

A Note for Early Release Readers

With Early Release ebooks, you get books in their earliest form—the author’s raw and unedited content as they write—so you can take advantage of these technologies long before the official release of these titles.

This will be the 8th chapter of the final book. Please note that the GitHub repo will be made active later on.

If you’d like to be actively involved in reviewing and commenting on this draft, please reach out to the editor at [e-mail].

Thus far, we have discussed the agent in isolation. A single entity that autonomously creates plans, executes them, and reflects on these processes. These single-agent systems are solely responsible for achieving the initial goal without interaction with intelligent systems other than humans. As powerful as these entities can become by themselves, a system where multiple agents work together to reach a common goal has even more potential. This is the multi-agent system (MAS), which may feature one or more environments in which many agents, some differently than others, can interact with each other.

Each agent in such a system will have a level of autonomy, but may have different sets of tools, (sub)goals, or even personalities. Regardless of their potential differences, these agents will need to collaborate to reach common goals through their own skill sets. Multi-agent systems are quite adept at handling intricate tasks that require balancing multiple complex ...