A subagent is a smaller agent that a larger agent delegates a specific task to, instead of doing everything itself.
Subagents help break a complex job into manageable pieces. In practice, teams reach for them when one agent would otherwise have to plan, execute, check, and summarize all at once.
They are useful when you want:
A parent agent decides that a task should be delegated. It sends a subagent a focused instruction, often with a limited context window, tools, or permissions.
The subagent works on that slice of the problem and returns an output to the parent agent. The parent then decides whether to accept it, revise it, or ask for another pass.
In many systems, subagents are just implementation patterns rather than a formal standard term. Different frameworks may call them workers, helper agents, child agents, or agent instances. The key idea is still delegation under supervision.
Scenario: a product team asks an AI agent to prepare a release note.
