The role of middle leadership in AI adoption
Scaling AI adoption means moving from individual to collective use. Middle leadership is the key piece, and there are four conditions they must create before expecting results.
One of the main barriers organizations face when scaling adoption is moving from individual to collective use — that is, translating individual productivity into organizational productivity.
And here, middle leadership —team and unit leaders— plays a fundamental role.
On one hand, they're the closest to the team's climate and culture. They can almost sense who's open to experimenting with AI, who's more resistant, and who doesn't use it not out of resistance but out of paralysis in the face of uncertainty or unfamiliarity. They're also the ones who'll have to manage their teams' doubts and concerns about AI. On the other hand, they live and breathe day-to-day operations. They know their teams' operational friction in detail, because they're the ones who surface it every time they hit a challenge.
That's why it's essential to involve them in any initiative that aims to scale AI. They must not only be properly trained in how to use these tools — they must model a behavior. They must make visible what's expected of them, empower their teams, and manage change in a way that creates the safety needed to experiment and move forward.
There are four conditions leaders have to create before expecting results.
The first is to genuinely get involved. Leaders set priorities almost subtly — not so much by what they say as by what they do. If in every weekly meeting you ask about sales but never ask what the team did with AI that week, by the time the meeting ends you already know where the focus will be. Don't get me wrong: it's not about no longer asking about sales. It's about making room to show that using AI matters.
The second is to give clarity on the why. Without it, 80% of your team will come up with their own hypothesis — which isn't advisable, considering that scaling AI often brings fear, uncertainty and anxiety. The team needs to collectively understand why the organization is adopting AI and how that connects to real objectives. Without clarity, uncertainty turns into noise.
The third is to use it yourself, too. If all you do is ask people to use AI but they never see you using it, your team will complete the trainings just to get them out of the way, not to genuinely absorb what they can do with it. If you don't do it, it must not be that important or valuable.
The fourth is to create psychological safety to experiment. This is different from giving clarity. Here the goal is to show that getting it wrong is part of the process. If you don't show how you tried something that didn't work, how you iterated and got nowhere, how you spent a whole day attempting something new… how do you expect your team to feel comfortable doing the same? Or, better yet, to say it out loud so you can help and support them toward success.
Culture change isn't decreed. It's modeled.
Which of these four conditions is the hardest to sustain in your organization?