Prompt chaining is the practice of using a series of connected prompts to build up a complex output, where each step uses the output of the last. It produces dramatically better results than trying to get everything in one giant prompt.
Example: building a personalized outreach sequence.
Instead of: "Write a 5-email cold outreach sequence for a VP of Finance at a SaaS company about our expense management software."
Chain it: Step 1 — "Based on this job description and LinkedIn profile, identify 3 pain points this person likely has." Step 2 — "Using these pain points, write 3 different opening hooks for a cold email." Step 3 — "Using hook #2, write the full email." Step 4 — "Write 4 follow-up emails to this sequence, each with a different angle."
Each step is easier for AI to do well, and the compounding specificity of the chain produces output that's much harder to get in a single prompt.