Multi-Agent
Several specialized agents working together on one task instead of one agent doing everything.
Think of it like
A film crew — director, editor, sound — each expert at one job, coordinating toward the same movie.
Example
A research task uses a planner agent to outline, several worker agents to gather sources in parallel, and a writer agent to stitch the report.
How it actually works
Splitting a job across agents lets each hold a focused role and prompt, and lets some run in parallel. The cost is coordination: agents talk past each other, duplicate work, or compound each other’s mistakes. Often a single well-prompted agent beats an elaborate committee — multi-agent earns its keep mainly when the subtasks are genuinely independent.
For product teams
Tempting for hard problems, but more agents means more ways to fail — reach for it only when the work truly splits.
For engineers
Multiple agent instances with distinct roles and contexts, coordinated via a shared orchestrator or message passing.
Related
- Orchestration — The coordination layer that keeps the team in sync.
- Supervisor Agent — The boss agent that assigns and reviews.
- Worker Agent — The agents doing the actual pieces.
Read anything AI without the jargon
Look up any term in plain English, or save terms as you read with the free Chrome extension.
Open DecoderAdd to Chrome