Glossary · Updated July 2026

What is Multi-agent system?

A multi-agent system is a setup in which several AI agents — each with its own role, tools, and goals — work together, or negotiate, to accomplish something no single agent handles well alone. The agents divide labor, exchange results, and coordinate through an orchestration layer that routes work between them.

Common shapes include supervisor-and-workers (one agent delegates and assembles), peer collaboration (agents pass work laterally), and blackboard patterns (agents read and write a shared workspace). The premise is the same one that makes human teams work: a scoped specialist with the right tools beats a single generalist stretched across everything.

What a multi-agent system multiplies, though, is the management problem. Every additional agent is another set of permissions to scope, another writer to the shared memory, and another actor whose steps must land on the record. Coordination scales the capability; management is what keeps the whole system accountable as it grows.

How it relates to agent management

A multi-agent system is the thing AI agent management exists to run — many agents made legible and accountable at once, not just orchestrated.