In plain English
Instead of one AI doing everything, the work is split up. One agent plans, others research or write, and another checks the result, passing work between them like a small team.
Example
For a market report, a lead agent sends three research agents to look into competitors, pricing and customer reviews at the same time, then combines what they find.
In short
A team, not a solo act. Several AI agents splitting up the work.
Why it matters
Splitting the work lets agents take on bigger tasks, work in parallel and check each other. It also costs more and adds new ways for things to go wrong, so teams use it when a task really needs it.
Common confusion
Multi-agent system vs. agentic AI. Agentic AI describes AI that pursues a goal on its own. A multi-agent system is one way to build it, using several agents instead of one.