Hi Reader!

Last month I ran a few team sessions helping companies build their AI governance structures. And honestly, I think it's a great place to start the conversation about what's right and what's wrong about how we use AI, before the full company rollout.

Because when something becomes a norm, it's extremely hard to reverse it and put new rules in place. But when you start having these conversations early on — when you don't have anyone fully on board yet, when there are more questions than answers, when you have the people who are excited about AI and the ones who are afraid of it in the same conversation — you create a culture of co-creation. You figure out together how you want to move forward.

The question everyone asks — and the one they should

In every one of these sessions, the same question comes up: how do we train people to use AI?

It's a fair question. But really, it's not the hardest one to answer.

The bigger question is how a leader's job should change when the team uses AI to produce the results. When AI and a person's expertise lead to different conclusions, how do you navigate that? When two people are on the same role and one is using AI and the other isn't, how do you evaluate their individual performance? And, more importantly, how do you know that while they were delivering the results, your team applied human judgment and critically evaluated AI's output to make sure it's free of hallucinations and bias?

Training doesn't get you there. The leadership job does.

The shape of the new job

I wrote about this in my latest Forbes Coaches Council article. The short version: the AI shift isn't happening to the workers anymore. It's happening one level up, to the leaders.

For years, a leader's value came from managing people one by one — reviewing work, giving direction, helping each person get better at their craft. AI has taken over a lot of that. So the job moves up a level, from managing individuals to designing how the whole thing works.

And it shows up in three places.

1. Work Design

The old job was to get more out of each person. The new job is to decide what the freed-up time is for.

When AI takes over part of the work, the easy move is to pour more of the same into the gap. The better move is to redesign the work itself — rebuild the role so the saved time goes toward higher-value work, not just more volume.

Greta Stahl, VP of Organizational Learning and Development at Workday, goes deep on exactly this on my Built by People Leaders podcast — how they redesigned roles and process rather than just layering AI on top. Worth a listen if this is live for you right now.

Listen to the episode

2. Decisions

The old job was to be the answer. People brought you the hard call, and you made it.

That doesn’t hold when AI is producing work across the whole team at once. You can’t be the final read on everything.

So the new job is to build the thing that makes good decisions happen without you — who owns what, where a concern goes, which work gets a second set of eyes before it counts.

3. Psychological Safety

The old way to shape a team was informal — set the tone, build trust, keep good relationships. That still matters, but it’s no longer enough.

Here’s what I see in my consulting work: people don’t fully trust what AI gives them, so they check it by hand and say nothing. There’s nowhere to raise the concern.

Safety used to mean people could push back on each other. Now it also means they can push back on the tool — say “this looks wrong” out loud, without it reading as if they’re slow or behind.

Final thought

The value has moved toward the human and organizational skills. What's left to decide is whether you run the old job a little longer, or start building the new one.

Talk soon,

Daria

P.S. If you're heading into an AI rollout this year, the best time to shape how your team uses AI is before it becomes the norm. If you'd like to work through what AI governance should look like for your team or your organization, book a call with me — we'll map where to start and what the early conversations need to cover.