Multi-Agent System
Definition
An architecture in which multiple autonomous AI agents collaborate to accomplish tasks that exceed the capability, context window, or specialization of any single agent. Agents in a multi-agent system may run in parallel (fan-out), sequentially in a pipeline, hierarchically (supervisor dispatching to workers), or as peers negotiating via a protocol like A2A. Each agent typically has a specialized role, a bounded context, and a defined interface with other agents. Coordination mechanisms include: shared memory/state stores, message passing via queues, event-driven triggers, and explicit orchestration by a supervisor agent. Multi-agent systems are most valuable when tasks are naturally decomposable, require diverse specialized knowledge, or exceed a single context window.
Builder Context
Multi-agent is often the wrong starting point. Start with a single agent and add agents only when you hit a real constraint: context window exhaustion, quality degradation from role overload, or genuine need for parallelism. When you do build multi-agent systems, the hardest problem is not the agents — it's the coordination layer. You need: (1) a clear contract between agents (what each receives, what each returns); (2) shared state management (where do agents write results that other agents read?); (3) failure handling across agent boundaries (what happens when a worker agent fails mid-task?); (4) observability across the full trace. The most common failure mode is agents that talk to each other instead of doing work — prefer direct data passing over agent-to-agent conversation.