Multi-agent system

A multi-agent system coordinates several specialized AI agents on one objective, each handling part of the work and passing results between them. Coordination covers planning, delegation, and reconciliation, so the system completes tasks that cross functions rather than confining each agent to a single narrow job.

Specialization is the argument for the pattern. An agent built for document extraction and an agent built for policy checking need different tools, prompts, and evaluation criteria, and combining them into one degrades both.

The cost is coordination. Failures in multi-agent systems are usually failures of handoff rather than of individual reasoning, which is why observability across the whole run matters more than the accuracy of any single step.