Prompt chaining: stack steps into a workflow
Single prompts are powerful. Chained prompts are transformative. Learn how to feed the output of one prompt into the next to tackle complex tasks that no single prompt could handle.
What prompt chaining is
Prompt chaining means using the output of one AI prompt as the input for the next. Instead of asking AI to do everything in one go — which leads to unfocused, mediocre output — you break the task into stages and handle each one deliberately.
A typical chain looks like: (1) Extract → (2) Transform → (3) Package. For example: extract the key points from a document → transform them into a narrative → package as a one-page brief. Each step improves the quality of the next.
When to chain vs. when to use a single prompt
Use a single prompt when the task is self-contained and well-defined — "rewrite this in plain English," "give me five headline options." Use chaining when:
- The task has multiple distinct stages that require different instructions
- You want to review and edit between steps
- A single prompt would need to do too many things at once to do any of them well
- The final output needs to be built up from intermediate components
The feed-forward pattern
The most important rule in chaining: each step should produce something the next step can work with. Don't chain steps that don't naturally connect. A good chain has a logical feed-forward: Step 1 produces a summary → Step 2 uses that summary to write an argument → Step 3 uses that argument to write a persuasive email. Each step adds a transformation, not just repetition.
Your review checkpoint between steps
One of the advantages of chaining is that you can review and edit between steps. Don't skip this. A small correction at Step 2 prevents a larger problem at Step 4. The chain is faster than doing the whole task manually, but quality control at each node is what makes the output actually usable.
Write and run a 3-step chain
Choose a piece of long-form content you have on hand — a report, an article, a meeting transcript, a set of research notes. Run it through the chain below.
Prompt: "Read the following [document type] and extract the 5 most important points. Be specific — include any key numbers, decisions, or recommendations. [PASTE CONTENT]"
Read the 5 points. Remove any that are wrong or trivial. Add anything important that was missed. Edit the wording so it's accurate. This is your input for Step 3.
Prompt: "Using these 5 key points, write a 150-word executive summary suitable for a senior manager who hasn't read the original. Lead with the most important finding. [PASTE YOUR EDITED POINTS]"
What to remember
- Chaining breaks complex tasks into stages — each step does one thing well.
- The pattern: Extract → Transform → Package. Each step feeds the next.
- Review and edit between steps. Quality control mid-chain beats fixing problems at the end.
- Don't chain steps that don't naturally connect — it adds complexity without benefit.