Prompt Chaining Workflow

Definition

A workflow that decomposes a task into a sequence of steps, where each LLM call processes the output of the previous one.

Key points

• Can add programmatic checks ("gates") on intermediate steps to keep the process on track.
• Trades latency for higher accuracy by making each call an easier subtask.
• When to use: "tasks that can be easily and cleanly decomposed into fixed subtasks."

Why it matters for the exam

• "Prompt chaining" is a named workflow pattern; know that it is a fixed sequence with optional gates, distinct from dynamic decomposition.

Common gotchas

• Start simple; only add agentic complexity when it demonstrably improves outcomes — chaining fits when subtasks are known in advance, not open-ended work.

See also

• Workflows vs Agents
• Routing Workflow
• Task decomposition (chaining vs adaptive)
• Augmented LLM

Sources

• blog-building-effective-agents