AgenticGlossary
Supervisor-Worker Pattern
The supervisor handles reasoning and coordination. Workers handle execution. The economics: a frontier model can cost many times more per token than a small one, so you want the frontier doing the hard thinking and cheap models doing the grunt work.
Concrete example: a customer-service supervisor (a frontier model) decides what each ticket needs, then dispatches to workers: one summarizes the ticket, one pulls the customer history, one drafts a response, one checks for compliance. Supervisor stitches the outputs together. The savings depend on how much of the work the workers can carry; measure them with your own evals.
Example
You are a supervisor agent. Given a user request:
1. Decompose into 2-5 sub-tasks.
2. For each, choose a worker tool: `summarize`, `extract_entities`, `draft_response`, `check_policy`.
3. Specify the input each worker needs.
Output JSON of the form: [{worker, input}, ...]
Do NOT attempt any of the sub-tasks yourself.When to use it
- Pipelines with mixed complexity (some easy steps, some hard)
- Cost pressure where running everything on the supervisor is wasteful
- Tasks that parallelize across workers (huge latency win)
When not to use it
- Tasks that don't decompose cleanly, where splitting adds overhead with no benefit
- Workers are tiny enough that their errors dominate (quality regression)
- Single-step tasks where supervisor overhead beats the savings
Articles that use this
- When to Use Subagents: Delegation Prompts That Keep a Long-Context Agent FocusedAgentic
- Agent Handoff Protocols: How to Transfer State Between Specialized Agents Without Losing ContextAgentic
- The Supervisor-Worker Pattern: One Strong Model Plans, Cheaper Models ExecuteAgentic
- When to Use a Reasoning Model, and When Thinking Just Costs MoreTechniques

