Coordinate teams of specialised agents — orchestrators, sub-agents, critics — to tackle problems no single agent can solve alone.
Five passes over the same idea, each from a different angle. Do them in order, or jump to whichever you need.
Multi-agent systems break complex tasks into specialised subtasks assigned to purpose-built agents. An orchestrator decomposes goals and routes work; sub-agents execute; a critic evaluates results. Managing communication, shared state, and failure modes across agent boundaries is the key challenge. Patterns like supervisor-worker, debate, and mixture-of-agents each handle different trust and scale constraints.
