Wharton professor Ethan Mollick recently wrote a piece called "Management as AI Superpower" that got me thinking. In it, he describes an experimental class where he challenged executive MBA students—doctors, managers, leaders, most of whom had never written a line of code—to build a startup from scratch in four days using AI tools. What he saw in a few days was, by his estimate, an order of magnitude further along than what students typically produce over an entire semester.

What’s interesting to me here isn't that they succeeded, it's why these students succeeded. They didn't learn some special AI technique, or really know much more about AI than you or me. They succeeded because they had management skills. Or, as Mollick put it: "They had hard-earned frameworks from classes and jobs, and those frameworks became their prompts."

In other words, their experience was the prompt. They just had to use it.

The Skills You Already Have

There's a misconception that getting value from AI requires a fundamentally new skill set. That you need to learn 'prompting' or become 'AI-fluent' before these tools become useful. But that's not really the case. The skill set you need to get real value from AI is the same one you've been building for years in your career: the ability to define a problem clearly, to scope work, to explain what 'done' looks like, to evaluate output and give useful feedback when something misses the mark. These aren't AI skills. They're management skills — the skills you develop over an entire career, by doing the work, over and over, until you build the judgment to know what good looks like.

Mollick points out that this challenge — communicating what you want clearly enough for someone else to execute on it — is so universal that nearly every industry has invented its own paperwork to solve it. Product requirements documents. Shot lists. Five Paragraph Orders. Engagement scopes. All of these translate what's in one person's head into someone else's actions. And all of them work just as well as AI instructions. Because the challenge has never been "how do I talk to AI?" It's "how do I communicate what I need clearly enough that someone else can deliver it?"

What's Actually Changed

So if the skills are the same, what's different? I think it’s the leverage.

Before AI, your expertise lived in your head and came out through your own labor—your own writing, your own analysis, your own execution. The bottleneck was always your capacity. You could only do as much as your time and energy allowed.

Now, that same expertise can be deployed at a completely different scale. You can delegate to something that's fast, inexpensive, and increasingly capable—as long as you can tell it what good looks like. The knowledge you've built over years doesn't just inform your own output anymore. It directs an engine that can produce, iterate, and refine at a pace that wasn't possible before.

The thing that's changing isn't what you need to know. It's what you can do with what you already know.

Start Where You Are

I wrote recently about the importance of naming the hard problem—finding the one constraint that, if solved, lifts multiple metrics at once. That idea matters here too. If you haven't named the hard problem, AI will happily help you produce an impressive volume of work that doesn't matter. If you have named it, AI becomes a force multiplier—compressing time-to-insight and accelerating execution in the direction that actually counts.

So if you've been on the sideline, waiting until you feel "ready," I'd encourage you to reconsider. The entry point isn't technical. It's experiential. Start with what you know best. Take a task you've done a hundred times—something where your judgment is sharp and your standards are clear—and delegate it to AI the same way you'd delegate it to a capable new team member.

The people who get the most from AI won't be the ones who learn the most about AI. They'll be the ones who know their own work deeply enough to direct it.