Technique: Prompt Chaining (Task Decomposition)
Prompt chaining breaks a complex task into sequential phases where the output of one phase feeds into the next. Instead of asking the AI to do everything at once (“write me a story”), you guide it through distinct stages: brainstorm plots, develop characters, write an outline, then draft each section.
This technique mirrors how professional creators work and produces dramatically better results for complex tasks. Each phase can use a different prompt optimized for that specific sub-task. In production AI systems, prompt chaining is implemented as pipelines where each step is a separate API call, allowing you to review and adjust between steps.
When to use: Long-form content creation, multi-step analysis, research reports, code generation projects, and any task that benefits from breaking into “think → plan → execute → review” phases. If the task would take a human multiple working sessions, use prompt chaining.