A multi-agent system is an architecture where multiple LLM-backed agents collaborate to complete a task — rather than one agent doing everything. Common patterns: an orchestrator agent that plans and delegates, plus specialist agents (researcher, writer, coder, reviewer) that execute specific subtasks.
The appeal is that specialization works: a focused agent with a tight system prompt and limited toolset makes fewer mistakes than a general-purpose agent trying to do everything. Complex pipelines — generate a report, verify facts, format output, check compliance — benefit from dedicated agents at each step.
The gotchas: inter-agent communication costs tokens (each handoff passes context), failures compound (one bad output propagates), and debugging is harder because state is distributed. Build single-agent first. Only add agents when you've hit a specific ceiling a second agent would solve.
Bring this to your business
Knowing the term is one thing. Shipping it is another.
We do two-week AI Sprints — one term, one workflow, into production by Day 10.