The most important weeks of my year were not the ones when I shipped code.
Last summer, I stopped building with AI for six weeks. Not because I burned out. Not because I ran out of ideas. Not because I took a holiday. I stopped because my teenage son wanted a go-kart and I had a third salon location to build out, a 300-square-foot satellite inside a plastic surgeon's med spa in the southwest part of Las Vegas.
Normally I would hire that kind of work out and get back to whatever was moving the business forward. This time I did something different.
Where I Was With AI Going Into That Summer
To understand why the pause mattered, you need to know where my head was in June 2025.
I had genuinely hit my stride. I had spent the spring rebuilding our salon website from scratch, migrated the code to GitHub, spun up a proper Supabase and Vercel stack, and started working closely with AI tools in a way that finally felt like real leverage. For the first time in over twenty years of shipping software, ideas were landing faster than I could queue them. Every Sunday afternoon I could feel exactly what the coming week would produce.
On Father's Day, I was at a staff party thrown by our salon team and I could not stop prompting from my phone. Physically present, mentally buried inside a build that felt like it could not wait. I was the person ignoring the room around him. That was the level of obsession I had reached going into summer.
Then July arrived and forced a different kind of decision.
The Build That Needed Doing
Our third location was opening, a small satellite space inside a medical practice. Three hundred square feet. A real partnership with a real business outcome, but a limited buildout by any commercial standard.
A job that size is exactly where a sensible founder hires out, stays out of the way, and gets back to the work that actually scales. I had quotes. I had a contractor ready to run it. That was the smart move.
Instead, I said no. Because at exactly the same moment, my son was asking for a go-kart.
The Go-Kart Move
We are deliberate about our three kids understanding the relationship between work and what they have. They live a comfortable life and my wife and I are determined that none of them grows up assuming that comfort is simply the default.
So when he asked for the go-kart, I did something slightly unconventional. I bought it for him first, upfront, before he had done anything to earn it.
Most parents would structure it the other way around, earn it first, then receive it. I reversed the order intentionally. Here is the go-kart. Now come and earn it by helping me build a salon. With your hands. Alongside me. For six weeks.
The reason the reversal mattered is that once the go-kart was already sitting in the garage, the dynamic between us quietly shifted. I was not supervising him to perform labour for a prize he was still chasing. He already had the thing. We were just two people building the same space together. The arrangement stopped being transactional almost immediately. It became something more like presence. I did not fully understand what was going on. The go-kart-first move had bought me something I had not known I was buying.
What We Actually Built
The work was deeply unglamorous and I loved almost every minute of it.
We spent evenings at the house assembling flat-pack cabinets, the storage units that would live in the back-of-house of the new location. We hauled them piece by piece from the garage to the space. Installed them together. Screwed together frames, aligned shelves, worked through instructions that ranged from perfectly clear to genuinely maddening.
Drywall. Mirrors. Paint. Stylist stations. Lighting. The hundred small decisions that make a finished salon look like a salon rather than a converted medical office. None of it was efficient. A proper commercial contractor would have done it faster and cleaner with considerably less back pain. That was not the point.
The point was that my son got to watch a real space come together from nothing. He was present for the decisions you would never notice on a finished walk-through, which surfaces get caulked and which do not, how lighting angles change what a stylist sees when they work on hair colour, why a mirror positioned three inches differently changes the entire feel of a room. You do not learn that from a book. You learn it from being in space while it is still being made.
What He Actually Learned
My son came out of those six weeks with something I could not have transferred any other way.
It was not how to install drywall. He will almost certainly never need that skill again. It was the underlying understanding of how businesses actually get built at the unit level. How the thing you walk into as a finished, polished experience is the product of hundreds of decisions, most of them made under mild time pressure, most of them imperfect, all of them necessary and specific.
That kind of knowledge does not transfer through conversation. It transfers when you are holding the other end of a cabinet at nine in the evening, trying to work out why the alignment is slightly off and what to do about it.
He is interested in the marketing side of what we do as a family business. The path we are walking him through slowly, entry-level marketing, then deeper into operations, eventually toward the CEO role over the coming decades, has to start somewhere real. By the time he is looking at a profit and loss statement or writing campaign copy, I want him to have stood inside a space we built with our hands and understood what went into making it exist. That summer was the first real day of that longer education.
What the Break Taught Me
I came back to AI the day after the new location opened in late September and did not surface properly until the following January.
The six weeks did not slow me down. If anything, they clarified something I had not been able to see clearly while I was moving fast.
Compounding gains from AI tools are real. The speed is real. The leverage is real. None of that is wrong or worth dismissing. But the six weeks away from the code taught me something no AI-for-founders article had managed to: the gains are a means. They are not the thing.
Whatever you built the speed to give you more time for, that is still the actual work. The business, the people in it, the next generation learning to carry it forward. You do not optimise those things. You do not A/B test them or iterate faster on them. You spend time on them, in person, with your hands when that is what the moment needs. And then you go back to the tools, and the tools serve something that is still intact.
I did not know I was over-indexed on velocity until I had to set velocity down for a season. The pause showed me where the real work actually was.
The code ships so you can go and do that. The go-kart was just the trick that made it obvious.