Prompt Chain

A prompt chain is a sequence of LLM calls where each call's output feeds the next — used to break complex tasks into reliable steps.

One big prompt asking for everything tends to do everything badly. A chain — extract → classify → draft → critique → finalize — often beats it because each step is narrow enough for the model to nail. Chains are the foundation of agentic workflows; agents are chains where the AI picks the next step.

Example

Refund triage chain: 1) extract order info, 2) classify request type, 3) check policy, 4) draft response, 5) verify against guidelines.

How OpenLabor uses it

Most OpenLabor missions are prompt chains under the hood, with the right model picked per step.

Won't a chain be slower than one call?

Usually yes by ~2-3x. The win is reliability — fewer wrong outputs that cost more downstream.

Related: agentic-workflow, tool-use, planner-executor.

AI Labor Glossary