Multi-Agent System

A multi-agent system is an architecture in which multiple AI agents collaborate or compete to complete tasks that are too complex for a single agent. Agents may act as orchestrators, specialists, critics, or executors — passing context, delegating sub-tasks, and aggregating results. Multi-agent systems require governance at both the individual agent level and at the interaction layer, where data flows, permission grants, and action chains cross agent boundaries.