What it means
A multi-agent system uses several AI agents, each with a specific role, that work together on a task, for example one gathering information, one drafting and one checking. A coordinating component manages how they hand work to each other.
Real-world example
A research team at a public broadcaster tests a multi-agent setup for preparing fact-check briefings. One agent collects claims from a transcript, another searches archives and trusted databases, and a third drafts the briefing with sources. An editor reviews the briefing before any use, and the team logs where agents disagree.
Why it matters for AI adoption
Splitting work across agents can improve structure and specialisation, but it adds complexity, cost and new failure modes, such as errors passing between agents unnoticed. For many use cases, a single well-designed workflow is simpler and more reliable.
Questions to ask
- Why does this task need multiple agents rather than one workflow?
- How are errors detected when they pass from one agent to another?