Multi-Agent System
A multi-agent system is a system in which two or more LLM-driven agents collaborate on a task, typically by exchanging messages, handing off ownership, or being orchestrated by a higher-level coordinator. Patterns include sequential pipelines, hierarchical supervisor-and-workers, peer-to-peer collaboration, and graph-routed flows. Frameworks like LangGraph, CrewAI, the OpenAI Agents SDK, and Mastra all provide multi-agent primitives. Multi-agent systems trade single-prompt simplicity for specialization and explicit control flow — useful when the work has genuinely separable sub-tasks, expensive when it does not and a single well-prompted agent could have done the job.
Example
A content-production multi-agent system might pair a "Researcher" (gathers sources), a "Writer" (drafts the piece), an "Editor" (rewrites for clarity), and a "Fact-checker" (verifies claims against retrieved sources) — each implemented as its own agent with its own instructions and tools. The pipeline composes through either explicit sequential handoff, a manager-led hierarchy, or a graph routing decisions across them.
Put this into practice
Build polished, copy-ready prompts in under 60 seconds with SurePrompts.
Try SurePrompts