The most common conversation I have with executives starts the same way. They've just read about a new AI capability, or they've seen a peer use it, and they ask: "Can we use AI here?"
It's a natural question. And it's almost always the wrong one.
"Can we?" is a question about possibility. It's about whether the technology can do the job. And the answer in 2026 is almost always yes. AI can write, analyze, summarize, reason through problems. It can do a remarkable number of things.
But possibility is not the same as worth doing.
The question that matters is "Should we?" Should we use AI in this workflow? At what stage? With what guardrails? And crucially: how am I going to direct it so it does what I actually need, not just what it's capable of doing?
That's the leadership question. And it's where most leaders hesitate.
It's tempting to hand that question to someone else. A consultant can map the landscape and say, "These are the possibilities." A vendor can say, "Here's what our tool can do." A colleague can experiment and say, "Look what I built." But none of that answers the question that matters to you: Should we do this, and if we do, how do I stay hands-on enough to direct it?
Here's what I see. A leader gets a roadmap with forty use cases. It looks comprehensive. But six months in, half of them don't fit the way the team works. A quarter create more friction than they solve. The ones that do work are fragile, because the leader never got close enough to understand the judgment calls that matter.
The difference between the teams that get value and the teams that don't is not better consultants or better tools. It's that the winning teams have a leader who stays hands-on. Someone who uses the AI themselves. Someone who knows what good looks like and can catch when it falls short. Someone who can say, "That's close, but it's missing the context. Here's what I need you to look at instead."
That's what "should we" really means. It means you own the judgment. You don't outsource it. You live it.
So the next time you see a new AI capability, skip "Can we use it here?" Ask instead: "Should we? And if yes, how am I going to stay hands-on enough to direct it?"
That question changes everything.