What football managers can teach us about AI and the developers who won’t adapt
Every generation of managers who dismissed the last wave got left behind. The same thing is happening right now with AI — and the developers who won’t adapt are already losing ground.
When Arsène Wenger arrived at Arsenal in 1996, the reaction from parts of English football was somewhere between scepticism and contempt. A Frenchman. Ideas about nutrition, sports science, training methods that didn’t look like training. Players who’d never heard of him. A league that had been doing things the same way for decades and saw no particular reason to change.
We know how it went.
What’s less talked about is what happened next. Alex Ferguson didn’t dismiss it. He watched, adapted, brought in support, evolved. You don’t win what Ferguson won across four decades by being precious about how things have always been done. Then came Pep Guardiola and Jürgen Klopp, each generation raising the bar further, each one building on what the last had unlocked.
And the managers who had looked at Wenger and decided it was a fad? Who had protected their methods and their status, done it the old way because the old way used to work? They found themselves managing in the Championship, or not managing at all.
The game didn’t wait for them.
I’ve been thinking about this a lot lately, because I keep having a version of the same conversation about AI-generated code.
The reaction from some developers is familiar. The output isn’t very good. It doesn’t understand context. A real engineer would never write it that way. There’s something almost territorial about it, a profession that has spent years becoming expert in something difficult, suddenly facing a tool that can do a version of that thing quickly and cheaply, and deciding the tool must be flawed rather than asking what that means.
I understand the instinct. It’s the same instinct that told English football the foreign coaches were wrong.
But here’s what I notice. The developers who are thriving aren’t the ones who dismissed it. They’re the ones who did what Ferguson did, brought the new thing in, worked out where it adds value, kept their own expertise for the parts that actually need it. AI writes the scaffolding. They build what matters.
The rest are still explaining why the code isn’t very good.
I should say: I’m not a developer. So this isn’t a lecture from someone inside that world. It’s an observation from someone watching from outside, who has his own version of the same story.
I have ideas. I’ve always had ideas. What I don’t have anymore is a large development team to support them. What I do have is Claude Code.
Since our new website went live, I’ve been building dashboards. Student recruitment tools. Things that would have required a development sprint, a backlog, a prioritisation meeting, and several weeks of waiting. Now they’re getting built quickly, tested quickly, iterated quickly. Not perfectly, but well, and fast, and in service of what actually matters.
The ideas didn’t stop when the resource changed. The support just changed shape.
That’s the thing about the managers who didn’t adapt. They weren’t short of ideas either. They were short of the willingness to let someone, or something, help them deliver those ideas differently.
The game doesn’t wait. It didn’t wait for them. It won’t wait for any of us.