Everyone’s making lists right now of the “new skills” for AI-native work. Context framing. Clear intent. Structured delegation. Critical evaluation of what comes back. Iterating through feedback instead of specifying everything upfront.
I’ve been reading these lists with a strange sense of recognition. Because I’ve seen this skill set before — for twenty-five years, in every truck and control room and product review that ever went well. These aren’t new skills. They’re the oldest skills in the building, finally getting billing.
What actually changed
Here’s the honest version of the shift. When your own hands did the work, you could get away with fuzzy intent — you’d figure it out as you went, because the conversation was happening inside your own head. The moment the work runs through something else — a model, a system, an agent — fuzziness gets expensive instantly. Vague intent in, confident nonsense out, at remarkable speed.
So the leverage moves to a different set of muscles: Can you state what you actually want, and why, and what “good” looks like? Can you provide the context that makes the difference between a generic answer and the right one? Can you evaluate output critically — catch the confident wrong answer — instead of accepting whatever came back polished? And can you give feedback that improves the next attempt rather than just rejecting the last one?
Now read that list again
Read it again, and tell me that isn’t also a description of every great director you’ve ever worked for. Every great EIC. Every product leader whose teams would run through walls.
Clear intent is what a director gives a truck before the show: not a keystroke-by-keystroke script, but a picture of what tonight should feel like — and what matters if it all goes sideways. Context is what separates the leader people trust from the one people merely obey: the why behind the ask. Critical listening — hearing what came back, testing it against reality, catching the confident nonsense — is exactly the signal-path discipline I wrote about a few posts ago: staying calm enough to actually hear what’s coming back instead of reacting to the noise of it. And good feedback has never once been “wrong, do it again.” It has always been “closer — here’s what’s off and here’s why it matters.”
The unexpected gift
And here’s the part I find genuinely hopeful, in a conversation that usually runs anxious. An entire generation of professionals is now getting daily, high-repetition practice at intent, context, evaluation, and feedback — the four muscles most careers never deliberately train. Every session with these tools is a rep. And the reps transfer, because the muscles don’t care whether the other end of the conversation is a model or a colleague.
The people getting genuinely good at this work are, almost by accident, becoming better delegators, better briefers, better listeners, better leaders. The practice field changed. The game was always people.
So let’s drop a waypoint: don’t train your team on tools. Train them on intent, context, evaluation, and feedback — and let the tools be the practice field. What you’ll get back is not just better output. It’s better engagement with every human in the building, because it was never two skill sets. It was always one.
Leave a Reply