Build the Twin

Tony's first real attempt to "modernize" almost soured him on the whole idea, and the way it failed is the most useful lesson in this book.

Flush with the win from the engine, he decided to fix the part of his business he hated most: quoting. Every quote meant Tony driving to the site, eyeballing the job, going home, digging through supplier prices, doing the math on a legal pad, typing it up, and emailing it two days later — by which point half the customers had already hired someone faster. So he bought some AI tool, bolted it onto his existing quoting process, and told it to speed things up.

It was a disaster.

Because the tool just made his bad process faster. His quoting was slow for reasons that had nothing to do with typing speed — it was tangled up in how he priced, which suppliers he checked, the back-and-forth with the customer, the way it all lived in his head and nowhere else. Pour AI into a tangled process and you don't get a clean process. You get a tangled process that's now also confusing and automated. Faster mess is still mess.

Tony almost quit there. A lot of people do. In fact, most corporate AI projects — the big, expensive ones with consultants and steering committees — fail for exactly this reason. They take a process built for humans, full of human hand-offs and approvals and workarounds, and they jam AI into the gaps. It's like the early days of television, when they put radio announcers on camera and had them read the radio news into a microphone. New medium, old habit. They hadn't figured out what TV was for yet. Tony hadn't figured out what AI was for yet either.

What it's for is not making your old business faster. It's building a new one. ✱ ✱ ✱

"You never change things by fighting the existing reality. To change something, build a new model that makes the existing model obsolete." — Buckminster Fuller

Let me tell you about a chocolate company, because it explains everything Tony did next.

In 1976, Nestlé — already a giant — invented a clever little product: a machine that made perfect espresso from a sealed pod. They called it Nespresso. And for about ten years, it went nowhere, because they tried to run it inside the mothership. Wrong fit at every turn. The existing sales force sold groceries to supermarkets; Nespresso needed to sell a luxury experience to individuals. The existing supply chain, the existing brand, the existing everybody — all of it quietly rejected the new thing, the way a body rejects a transplant. The antibodies again.

So eventually Nestlé did the only thing that works. They took Nespresso out. Separate unit, separate building, separate brand, free from the immune system of the parent company. And freed from the old machine, it became one of the most profitable things Nestlé has ever made. There's a Nespresso machine in half the hotel rooms in the world because somebody finally stopped trying to grow it inside the old company and let it grow beside it.

That's not a fluke. It's the pattern. Amazon didn't build its cloud business inside the retail operation — it built AWS at the edge. Apple built the Macintosh with a small team pulled off to the side and told to go disrupt the company's own products. The pattern has a name, from thinkers who study this: disruption happens at the edge, never at the core. You cannot rebuild the engine of a plane while you're flying it and carrying passengers. You build a new plane.

For a giant, building at the edge is a major undertaking. For you, it's almost unfair how easy it is — because you're small, and you have no immune system to escape from. You just need to know to do it.

So here's what Tony did the second time, and here's the method, in plain English.

---

First, he pictured the finished thing. Before touching a single tool, Tony sat down — with help — and described what Tony's Plumbing Co. would look like if it were built today, from scratch, AI-native, by someone who'd never carried the baggage of "how we've always done it." Not the current business with gadgets bolted on. A clean-sheet version. How would quoting work if it were perfect? How would scheduling, dispatch, follow-up, hiring? You start from the destination and work backwards to today. (It's called backcasting, and it's the easiest thing in the world to do in an afternoon of conversation with an AI — picture where you want to be in three years, then ask what has to be true in one.)

Second, he picked one piece. Just one. Not the whole business — that's how you blow yourself up. One workflow. He picked the thing that hurt most: quoting. The rule is to choose a single, well-understood, repeatable piece of the business and rebuild only that, cleanly, in the new world.

Third — and this is the part that keeps you safe — he copied it, he didn't move it. The old quoting process kept running for the real business, untouched. The cash cow was never at risk. Off to the side, in a lean little setup of its own, Tony built a brand-new AI-native version of quoting: redesigned from scratch (not the old mess, sped up — a new process), with AI doing the heavy lifting underneath. It ran in parallel. If it broke, nothing real broke. The lights stayed on.

Fourth, he let it prove itself, then let it take over. The new quoting ran alongside the old one until it was clearly, obviously better — faster quotes, out the same day, more of them turning into jobs. Once it was winning, Tony retired the old way and the twin took over quoting for the whole company. Then he picked the next piece — scheduling, say — and did it again. Copy, prove, switch, grow. One workflow at a time, the new AI-native company grew up beside the old one, organ by organ, until one day Tony looked up and realized the twin was the business now, and the old exhausting version had quietly faded away.

He never bet the company. He grew a better one inside its own backyard.

---

Iwant to name the one decision that made this work, because it's the one most people get wrong, and it sets up everything that comes later.

Tony did not try to do this alone, and he did not hire a consultant. He found a builder.

There's a world of difference. A consultant studies your business, writes you a beautiful report about what you should do, hands you an invoice, and leaves you exactly where you started — now with homework. A builder rolls up their sleeves and actually makes the thing: builds the twin, wires up the AI, gets the new quoting running, stays until it works. When execution is the whole game, a stack of recommendations is worthless. You don't need someone to tell you to build a new plane. You need someone who builds planes.

Hold onto that distinction. It's going to matter a great deal at the end of this book, when we talk about how you actually pull all this off without quitting your day job or losing your mind.

But first, something happened to Tony's company once the twin was running that he genuinely did not expect.

It became worth something. Not "good income" worth something — asset worth something. The kind of worth that makes a stranger call you up and ask, carefully, whether you'd ever consider selling.

And the day Tony got that call, he finally understood the difference between owning a job and owning a company.