Tip
Andrew Wilkinson talks about his biggest mistakes in business: “The biggest mistakes I’ve made have been going into business models where other people have repeatedly failed and thinking, I can do this better.”
Turns out strategic decisions about AI adoption work the same way.
You’re evaluating whether to build an AI-powered customer service system. Dozens of companies have tried automating customer service. Most failed. Your CTO says “but we’ll do it better—we have better data, better prompts, better integration.” That sounds familiar.
Here’s what you know from watching colleagues chase sexy ideas: the friend who thought their restaurant would be different (it wasn’t), the exec who thought their consulting firm would crack the code (it didn’t), the startup that thought they’d solve healthcare workflows (they couldn’t). The pattern isn’t execution failure—it’s structural problems in the business model itself.
Younger managers see the opportunity: “AI finally makes this possible!” You see the graveyard. You’ve watched enough smart people with good resources fail at structurally broken models. That’s not pessimism—it’s pattern recognition from seeing which obstacles are about execution versus which are about fundamental market dynamics or regulatory constraints or unit economics that can’t work.
This judgment—distinguishing between “we can execute better” and “the model is fundamentally broken”—comes from watching both repeatedly. The temptation to think “but we’re different” is universal. The wisdom to resist it when the pattern is clear? That only comes from seeing colleagues burn themselves repeatedly on the same models.
Context
Andrew Wilkinson has started or been involved with 75+ businesses as co-founder of Tiny, a holding company. He’s seen nearly every business model play out—the ones that work despite weak execution and the ones that fail despite strong execution.
The distinction between execution problems and structural problems only becomes clear after watching enough attempts. For experienced leaders evaluating AI investments or strategic pivots, this isn’t academic—you’ve watched peers chase broken models for decades.
That pattern recognition is the advantage.