Most developers’ first reaction when I tell them “you can prompt AI to write a prompt” is laughter. It sounds ridiculous—too meta to be useful.
But when they try it, the laughter usually shifts into wide eyes and open jaws. The “aha” moment lands. Suddenly they see that AI isn’t just a code generator. It can be a partner in sharpening the very questions we ask.
The Back-and-Forth Shortcut
A fellow coach once shared a story with me: they had been iterating with Copilot through several rounds before finally getting a solid answer. Frustrated, they asked, “How could I have prompted differently to get here faster?”
The response from the AI unlocked something bigger than just a fix. It revealed that instead of diving straight into solution requests, you can stop and ask AI to help you frame the problem first. For new or tricky challenges, this shift—prompting AI to guide your prompting—often leads to faster, clearer, and more accurate results than brute-forcing through trial and error.
Beyond Developers: Refining Stories with AI
This works just as well for product people. In story refinement, I’ve used prompts like:
“Given the attached specification, examine it like you’re a software engineer, and tell me anything that’s missing or unclear. Anticipate questions an engineer might ask. Then, give me a prompt to evaluate this story using the INVEST model.”
The AI’s response often surfaces gaps no one had noticed—assumptions hiding in plain sight. It helps product managers see how a developer will approach their story, and helps developers push back on requirements before those requirements become costly mistakes.
Why It Works: Context = Accuracy
I tell teams a simple mantra:
The more complete the context, the more accurate the outcome.
When you ask AI to generate prompts, you’re effectively telling it to mine context from you. It flips the script—the AI asks the questions, you supply the missing details. With that richer context, the outputs become sharper, whether it’s a code snippet, a test plan, or a story breakdown.
Pitfalls to Avoid
Most new users take the shallow route:
“Give me a function that does X” or “Where is Y in the code?”
That’s like asking my toddler to complete a task with no detail—you’ll rarely get what you hoped for. Success with AI isn’t about magic incantations, it’s about probability. If you don’t give enough context, you lower your success rate. And when people don’t understand this, they blame the tool instead of their technique. That’s how skepticism grows, and why coaching matters.
A Practical Recipe
Here’s a 5-step approach teams can try tomorrow:
- Make a plan. Write out the steps you’d take to solve a task (like adding an endpoint or writing tests).
- Bolster with AI. Ask AI to refine that plan, adding detail and context from your codebase or stack.
- Critique the plan. Use AI as a judge: “What’s missing? What are the weaknesses?”
- Finalize. Update the plan based on its critique.
- Execute. Now run with Copilot or another tool in Agent Mode, pausing to review and understand at each step.
This isn’t just “getting AI to code faster.” It’s teaching teams to treat AI as a collaborator in thinking, planning, and validating before execution.
The Full Product Developer Connection
Prompting-for-prompts ties directly into the Full Product Developer mindset.
As AI takes over more of the how, the advantage shifts to those who can own the why and the what. Developers who understand the product, can frame context, and know how to orchestrate AI effectively will always deliver more value than those who just execute tasks.
A Full Product Developer doesn’t see AI as magic, or as a crutch. They see it as a cognitive power-up. They experiment, refine, measure, and build habits that turn AI into a force multiplier. Most importantly, they never delegate their critical thinking.
At the end of the day, AI should enhance our ability to solve problems—not atrophy it. Prompting AI to prompt is one of the sharpest tools we have to make that enhancement real.
Stop asking AI for outputs. Start asking it to help you ask better questions. That’s where the real leverage is.