I spent the first part of my career inside enterprise AI — running implementations with organizations like Citizens Bank, Fidelity, Optimizely, and Jabra. Machine learning pipelines, personalization engines, predictive models. I know how AI projects get scoped, how they get sold internally, and how often they fail to deliver what was promised.
A few years ago, I started working with a high-end residential general contractor in Boston. They wanted to bring AI into their operations — faster estimates, better documents, less time on administrative work. What I found when I got inside the business was a company that ran entirely on institutional knowledge. Nothing was written down. Pricing logic lived in the owner's head. Every document started from scratch. I'm also a licensed builder — which means the Construction OS metrics aren't theoretical. I understand the work the tools are replacing.
That's when the framing shifted. The problem wasn't that they needed better AI. The problem was that the business wasn't legible — not to AI, not to new hires, not to the owner when he wanted to step back. You can't automate what hasn't been documented. You can't delegate what hasn't been described. The gap wasn't technological. It was structural.
What I do now is Operational Architecture. I work with service businesses to document how they actually operate, structure that knowledge into the right containers, and build AI on top of something solid. The result isn't just faster workflows — it's a business that can scale without the founder in every conversation, onboard a new person on day one, and use AI tools that actually work because the knowledge is there for them to work from.