Spinning the wheel

I’ve worked at a startup with twelve people and a listed company with thousands. A German firm that got acquired, restructured, and split from its holding group before I’d been there two years. A growth-stage company that hadn’t decided yet what it wanted to be.

Every single one of them was at a different point on the same wheel. It goes something like this.

When you’re small and broke, you build. Not because you want to — because you have to. Off-the-shelf software costs money you don’t have, and the generic solutions don’t fit your specific problem anyway. So someone hacks something together. It’s messy, it lives in one person’s head, and it breaks on Friday afternoons. But it fits. It does exactly what you need.

Then something works. You grow. More people, more complexity. The thing built at the kitchen table starts creaking under the weight of the company it helped create. Maintaining it is expensive. Scaling it is a project in itself. And so you buy. You pick a platform, sign a contract, migrate. For a while, it’s a relief — someone else handles the maintenance, the updates, the infrastructure. You focus on growing instead of fixing.

But platforms are built for the average customer. And the more you grow, the more you stop being average. The out-of-the-box product that saved you starts to feel like a suit that doesn’t quite fit. You add integrations. You write workarounds. You customise so heavily that what you bought barely resembles what it was on the demo. And then one day someone asks: why are we still paying for this? We’ve rebuilt half of it ourselves.

So you start building again.

I’ve watched this happen in different industries, different countries, different company sizes. The details change. The motion doesn’t. What I’ve never seen is a clean exit — at least not for most companies. The ones that seem to break the cycle share one thing: they were successful enough that their particular solution became general. Amazon didn’t set out to build a cloud platform — AWS emerged from years of scaling, APIs, web services, the whole Web 2.0 moment. Airbnb didn’t plan to become an experimentation reference — they grew until no off-the-shelf tool could keep up with what they needed. Both found their way out. Both made it profitable. But neither exit was a strategy. It was what happens when success collides with a problem the market hasn’t solved yet.

I keep thinking about this now that AI has entered the picture, because AI is doing something strange to the wheel. It’s compressing it. The cost of building custom things has dropped sharply. A small team can now build what would have taken a full engineering department five years ago. The barrier that used to push companies toward generic solutions — cost, complexity, time — is getting lower. At the same time, large companies are suddenly acting like startups again. Building internal tools. Experimenting. Moving faster than their size usually allows.

But here’s what I find most interesting: the same duality is playing out inside AI itself. You can build a custom connector that links an AI tool directly to your data, shaped around your team’s specific questions and context. But the product you’re connecting to has likely built their own version of the same thing — trained on data from all their customers, optimised for the general case. So which one wins? Yours knows your particularities. Theirs knows the pattern across an entire industry. General versus particular, again. Same wheel, different layer.

I’m not sure AI breaks the cycle. I think it might just spin it faster — and replicate it at every level.

What I know is that every company believes its situation is unique. The startup that thinks it moves too fast for any system. The enterprise that thinks its scale makes it incomparable. The company mid-transition that mistakes complexity for exceptionalism. They’re all partly right. But they’re all on the same wheel.

I used to think the right question was where the exit is. I’m less sure now. Maybe the wheel isn’t a trap to escape — it’s just the shape of how things move. You’re either spinning it or it’s spinning you.

George Michael sang about not wanting to watch someone spin a wheel they couldn’t see. I keep coming back to that. The companies that found their way out didn’t leave the system — they became the ones turning the wheel for everyone else. The rest of us keep going around, hoping the next revolution lands somewhere better than the last.

Maybe that’s just how it goes.