Autonomous Event-Driven Multi-Agent Orchestration for Enterprise AI at Scale

The article presents a study on multi-agent orchestration for enterprise AI, introducing a Task Manager that enhances continuous operation through priority inference and event merging. Evaluating DAG Plan and Execute and ReAct across 208 enterprise scenarios reveals that orchestration performance degrades at larger scales due to agent discovery noise, with DAG Plan and Execute showing higher precision at smaller scales but increased overhead at enterprise levels. The Task Manager significantly reduces high-priority queue latency by 14-75% and improves event correctness by over 20 percentage points, making it a crucial advancement for practitioners managing large-scale AI systems.