liveBook · Manning

Multi-agent system

As explained in AI Governance

A multi-agent system is a system in which multiple agents interact or collaborate to perform tasks.

A multi-agent system uses multiple agents that interact or collaborate to solve tasks. Agents may run on separate model instances or share a common model backbone. section 2.6.1 section 2.6.1

Code generation and review

One agent generates code while another reviews or tests it. Collaboration can catch mistakes, but coordinating agents adds complexity and failure paths. section 2.6.1

Agent interactions can be fast and hard to inspect. As autonomy, tool access, and workflow involvement increase, so can the reach and scale of harm; individually acceptable steps may combine into an outcome no reviewer would approve. section 6.7 section 6.7

Figure 2.6 A high-level architecture of how a multi-agent system works[22]. Agents may use in separate model instances or share a common backbone model.

Use the architecture overview to see how agents may be arranged around separate or shared model instances. section 2.6.1

How AI Governance covers it

How deep it goes
Defines agent collaboration and gives a code example, while noting orchestration complexity and failure paths.
Examples
Illustrates code generation paired with review or testing, and IT support agents that run diagnostics and collaborate.
What you can do after
Understand how agent collaboration can catch mistakes while adding orchestration complexity and failure paths.
How it connects
Connects multi-agent systems with agentic workflows, autonomy, and tool use.

Also explained in

Compare how 13 books explain multi-agent system →

Related concepts

This explainer was generated by Manning's knowledge graph from the text of AI Governance. The section links point to the parts of the book it draws on.