Your leadership team agreed in the room. That is not the same as agreeing.

If you are leading this, the commitment is probably real. You have seen the case. You want to move. That is rarely where it stalls.

What slows it tends to sit a layer down, in the leadership team, and it often does not look like opposition. Nobody argues against it. What you get instead is a steady stream of reasonable concerns. More governance before we proceed. More consultation first. More certainty before we commit. Each one makes sense on its own. Together they create drag.

That drag is hard to address because it has never been named. You cannot challenge a position no one has stated. So months pass. Pilots stay pilots. The organisation keeps spending on AI without scaling it, and the gap between what was agreed in the room and what happens after it goes unspoken.

This is not a rare pattern. McKinsey's research on AI points the same way: the organisations that fail to scale it mostly fail for organisational reasons, not technical ones, and the leadership team is the most common place it comes undone. The technology is rarely what stalls.

Most of what slows an AI programme is not a disagreement about AI.

When you look at what actually drives the hesitation, it is rarely bad faith. It is more human than that.

For some, it is their own position. AI raises a question earlier technology changes did not: does my expertise still matter? For a leader whose standing has always rested on what they know, that is not a small question, and it does not get said aloud.

For others, it is load. AI does not always arrive as a once-in-a-generation opportunity. Sometimes it arrives as one more complex, unpredictable thing landing on top of everything they are already carrying. Accountabilities do not shrink when a new priority shows up. They stack.

And there is the leader two or three years from the end of their career, managing the distance between what is being asked and the time they have left. Not obstructing. Managing. None of these responses is unreasonable. All of them shape the outcome of an expensive programme if they go unnamed.

Most of the preparation goes into the business case. Almost none goes into the human one.

Most leaders have a vivid, detailed picture of success. They have refined it and built the case around it. The picture of failure is usually thinner, and almost never shared across the team. That gap is where programmes come apart. When a team has no common view of what they are trying to avoid, they cannot make coordinated calls when things get hard.

The other question worth sitting with is about the person least ready to move. Not how to manage them. What they would need in order to feel they had a real stake in this. Asked that way, the most reluctant person stops being an obstacle and becomes part of the design.

The teams that scale AI are not the ones that avoided any of this. They are the ones that started here.

In practice, that comes down to a few things.

Listen before you go further. Not a communication cascade. A real conversation with the full leadership team about what they are navigating, including the concerns nobody has said aloud. This is the most useful thing you can do, and it is rarely too late to do it.

Meet people where they are. The person who cannot move at the pace being asked is not a problem to manage. They are a design brief. What would make them feel invested, rather than consulted?

Agree what happens when progress stalls, before it does. Who names it, who steps in, what happens next. Without that, the whole thing drifts into polite agreement and slow delay.

That conversation is hard to run on your own team. The questions are uncomfortable to ask, and harder to answer honestly when you are the one in the room. That is the part worth getting help with.