For decades, product managers have been writing requirements. This foundational skill has evolved from formal PRDs (Product Requirements Documents) to user stories and lightweight acceptance criteria.
However, the core of the job has never changed: we still need to define who the user is, what they need, and why it matters. The challenge has always been to get this right—concise, clear, and actionable—so developers can translate it into functional software.
Now, AI is involved. Suddenly, there’s a rush to redefine everything under the banner of “prompt engineering.” The reality? It’s simply another way of writing requirements.
AI Is the New Developer
Think about the best developers you’ve worked with. You don’t hand them a spec filled with implementation details. You give them the problem to solve, constraints to consider, and outcomes to achieve. The best ones ask questions, clarify ambiguities, and sometimes challenge your assumptions. The conversation is iterative, and the output improves as a result.
AI should be no different.
Instead of treating prompt engineering as some mystical new discipline, we need to approach AI the way we do with lead developers. Clear inputs, well-structured problems, and an openness to refining our requests based on the responses we get.
Requirements, Not Prompts
The rush to master prompt engineering is a temporary distraction. We need to develop the fundamental skill of writing structured, outcome-driven requirements—just as we always have.
AI doesn’t change the need for clarity; it amplifies it.
Rather than crafting elaborate prompts, we should be focusing on:
Who are the customers? Define personas, their problems, and their constraints.
What do they need? Share the desired outcome, not the implementation details.
Why does it matter? Articulate the value to customers and to your business.
When we shift our thinking from “engineering prompts” to “defining requirements,” we create a natural, structured way of working with AI that seamlessly integrates with how we already build products.
The Future Is a Conversation
Just as we collaborate with developers through discussions, refinement, and iteration, we must learn to engage with AI in the same manner. AI is not a tool that simply takes perfect prompts and produces flawless results; it is an entity that requires interaction, feedback, and adaptation.
Good product managers don’t just write; they engage in conversation. They navigate ambiguity, clarify intentions, and refine based on feedback. That skill is more crucial than ever—whether you’re communicating with a developer or an AI model.
Developers—and AI tools—don’t need details; they need context: who, what, and why, not how and when.
The tools will change. The interfaces will evolve. But writing good requirements? That’s forever.

Want to upskill your team with enhanced skills? Contact us about The AI-Empowered Product Manager and other learning programs.