Wayne Gretzky said, "Skate to where the puck is going, not where it has been." Right now, most organizations are skating toward prompt engineering. Smart move. The puck is there. But it is already moving.
Two companies, two different bets. Company A rolls out a prompt workshop. Two hours, 500 employees, and by Friday everyone can write a decent prompt. Productivity ticks up.
Company B starts the same way. Then they ask a different question: which of our strategic problems should AI actually solve?
Six months later, Company A has a growing prompt library and scattered efficiency gains. Company B has transformed one domain, built scaffolding around it, and learned more in six months than Company A will learn in two years. The difference is not the tool. It is the thinking.
Why prompting is necessary but not sufficient
Prompt engineering earns its moment. It democratizes access for anyone, regardless of technical background. It produces real results quickly. It builds momentum.
But democratization is not differentiation. Your competitors have the same AI tools you do. Research from GitHub showed AI assistants made developers 21% to 55% faster at coding tasks. Speed matters. But once everyone in your industry can prompt, speed alone stops being a differentiator.
Have you ever gotten a technically correct AI response that completely missed the point? That is the gap. Not the prompt. The thinking behind it.
From prompts to problems
The skill that determines AI's real impact is problem formulation: finding the real problem beneath the symptoms, breaking it into addressable parts, and defining boundaries that guide AI toward useful outputs.
Here is the difference in practice.
A prompter says: "Write a marketing email for our Q4 launch."
A strategic thinker asks: "What is actually causing our customer retention to drop? Is it product, messaging, or competitive pressure? Which segment should we prioritize before we say anything publicly?"
The first gets a competent email. The second might discover the launch should be delayed because retention data reveals a positioning problem the email campaign would only mask.
Prompting delivers your thinking. Problem formulation sharpens the thinking first.
The skills that compound as AI improves
Here is the counterintuitive part: as AI gets better at execution, the capabilities that determine its impact become more human.
Reckitt did not deploy AI everywhere. They focused on one domain (marketing), analyzed 300 tasks systematically, and achieved 60% faster product development. The strategic choice of where to focus determined their results. AI did not make that choice.
Kodak and Fujifilm had the same access to digital technology. Kodak optimized film. Fujifilm reimagined itself around materials science and healthcare. Same tools. Fundamentally different thinking. AI generates options. Humans decide which futures to build.
Stop and think: what percentage of your AI investment goes to prompt engineering versus the thinking skills that determine what your people do with a prompt?
The bottom line
Teach prompting. Everyone in your organization should be able to work with AI tools. But if prompt engineering is the only capability you are building, you are skating to where the puck was. The organizations that win will use prompting as the gateway to something harder and more durable: precise context and intent, problem formulation, and the judgment that determines whether AI produces insight or noise.