Twenty-two years I spent in this industry's operations. Since January, I've built eight live products with AI as the workforce.
That's the split I'm writing from. Most of what gets written about AI in the trades comes from one side of it: the platform, the analyst, the consultant who never stood a schedule board up at six in the morning. I've spent real time on both sides now, and the view changes completely depending on where you're standing. The view from the deck sees a market. The view from the work sees a foreman with four minutes before the crew loads out, an owner still paying off last year's software mistake, and a schedule that has to hold regardless of what the demo promised.
I'm not writing this from the deck. I'm writing it from twenty-two years on the ground, plus seven months building the thing I wished existed the whole time I was down there.
Here's where the flip actually sits. The assumption is that the hard part of AI is the AI: the model, the tool, the subscription. It isn't. You and I have the same models. The hard part is the structure around them, and almost nobody selling AI into this industry has ever run a truck, a crew, or a Friday payroll. That gap is the whole story, and it's the reason so much of what gets built for the trades gets built wrong before anyone even opens the laptop.
The industry's AI conversation is loud right now, and it is happening almost entirely over the heads of the people doing the work. Webinars about transformation. Panels where the word gets used forty times and defined zero. Vendor pitches promising a copilot for everything from estimating to payroll. None of it is aimed at the person actually running the crew, and underneath the noise sits a plainer question almost nobody asks out loud: who owns what gets built while you use the tool. Not who holds the license. Who owns the thing the software is learning about your business, every single day you run it. That's the question the noise is drowning out, and it's the one that decides whether AI compounds in your operation or just adds another line item to the stack.
Here's a piece of this I've lived through myself, not just built around. One operator I know is four years into LMN and still uses maybe a fifth of what he's paying for, to this day. I've been on the other side of it too. I personally used Aspire, and watched a company fail to adopt it. Not for lack of trying. For lack of the time, staff, and training a rollout like that actually takes. Entire consulting markets exist around this exact pain point, because the platform's own staff and onboarding get you through the initial launch and the surface-level integrations, and then they stop. I've seen multi-year efforts end in abandonment. I've also seen companies succeed, after they put in the real time, the real resources, the staff, the training, and the commitment it takes to get there.
I don't need to rant about this, and I'm not going to. None of this is a knock on any one platform. But I'm not just working off what I've watched. McKinsey's own survey research, tracked consistently since 2012, puts transformation success at under 30 percent. Their more recent work found only 16 percent of digital transformations both improved performance and made the improvement stick. Pendo's 2019 Feature Adoption Report, built from usage data across 615 software products, found that 80 percent of the features in the average product go rarely or never used. That's not one company's bad luck. That's the shape of the whole category, documented, not just my war stories.
Here's what I actually see, building it, not reading about it. One of my early adopters, a landscape company in growth mode, is running a platform built to how their workflows actually are, not the other way around. When something in it doesn't fit their crews, the fix lands the same day, or the next one, not in a quarterly release note six months out. There's a direct line to the person making that adjustment, and that direct line creates its own momentum. When a ticket comes in, it gets read and processed immediately, by someone who already knows their build. And their staff has an AI agent trained on their own build, one that knows their product only, intimately, the way a good foreman knows his own truck before he knows anyone else's. It looks, sounds, and feels like their company. Their language is in it. Their internal names are there. Ask it a question about the operation and it answers about the operation, not about the category.
That's not a small difference. Most software relationships in this industry run one direction: you adapt to the platform, forever, because the platform was built for ten thousand companies and you're one of them. This runs the other direction. The platform adapts to you, and keeps adapting, because someone is actually watching your business use it.
The part worth sitting with is what's underneath that: the data. A small business in growth mode, building this way, right now, is creating its own proprietary data at exactly the stage when that data is worth the most. They're owning it, not renting it. It isn't tied to the fate of a big platform, the LMNs and the Aspires and the LeanScapers of the industry, real tools, built on a real and different bet: that you rent the system and reshape your business around it. This is the other bet. You own the system, and it reshapes itself around your business. I'm not here to tell you the big platforms are wrong for every shop; plenty of operators are well served by exactly that trade. I'm telling you what ownership buys the operator who's still growing: the data compounds toward the business that made it, not toward whoever's platform it happened to live inside that year.
I don't have a clean way to close this out, and I'm not going to manufacture one. This is a new frontier being explored, and it has a long way to go before anyone can honestly say it's been defined. We are early. The possibilities are only growing. The ideas are becoming solutions, one build at a time, one operator at a time.
If any of this sounds like your shop, the practice page is on my site. Everything else is a conversation, same as it's always been.