Tagged: orchestration

Multi-agent systems assign specialized roles to separate AI agents that coordinate to complete tasks no single agent could handle reliably. The key architectures — supervisor, pipeline, and peer-to-peer — each trade control for flexibility in different ways.

The three dominant agent orchestration patterns are supervisor (central coordinator delegates to workers), sequential pipeline (agents pass output forward in a chain), and swarm (agents communicate peer-to-peer). Each trades control, debuggability, and flexibility in different proportions.