The consensus view on AI and organisations goes something like this: AI democratises capability. It compresses the gap between strong performers and weak ones. Everyone gets a co-pilot.

I don't think that's what's actually happening. From where I sit — across my portfolio and the companies I advise — AI is an amplifier. It takes whatever is already there and makes more of it. Good operators are using it to move with more precision and at higher speed. Bad operators are using it to produce more output without improving outcomes. More decks. More roadmap items. More hiring specs. The illusion of progress at higher velocity.

The gap between strong and weak operators isn't closing. It's widening.

Here's why. Good operators already know what problem they're solving. They have organisational clarity — a shared understanding of what matters, what doesn't, and why. When you give that kind of team AI, they use it to go deeper and faster on the things that already work. The leverage is real because the foundation is there.

Bad operators don't have that clarity. What they have is ambiguity dressed up as strategy, and AI gives them industrial-scale capacity to produce more of it. The problem isn't the tool. The problem is that the tool doesn't ask the hard question: do you actually know what you're trying to do?

I saw a version of this play out at Revolut when we were scaling fast. The teams that had genuine clarity about their objectives used every new resource — people, process, budget — to compound what was working. The teams that didn't had the same resources and created more noise. The constraint was never capability. It was always clarity.

AI makes that dynamic more visible, not less. Before asking "how do we use AI," the more useful question is: do we have the organisational clarity to use it well? Clear ownership. Real accountability. The discipline to cut what isn't working, even when AI makes it cheap to keep it running. That clarity is what separates the operators who will benefit from those who will be buried in better-looking mediocrity.

The operators winning with AI right now aren't the ones who moved first. They're the ones who moved with precision.