Seeing the Forest and the Trees

June 3, 2026

I am not an engineer. I have run service businesses for decades, and a year ago I could not have told you how a knowledge graph works or what a vector index does. So the part that still surprises me is that I ended up designing AI infrastructure, and that the design holds up when people who build these systems for a living look at it. I have thought about why that happened, and the answer is in where I started.

I did not start at a model, or at any single tool or feature. I started with a question. Why doesn't AI work better for everyone? Then a second one came right behind it. Why can't AI just be the system that runs the productivity tools, instead of one more tool sitting beside them? I sat with that for a long time before I let myself reach for any piece of an answer.

That second question is what sent me down the road. If AI was going to run the tools instead of being one of them, something underneath had to hold what all of those tools were actually about. The work kept pointing at the same place. Meaning has to be held in one spot and owned by the person or the organization, instead of rebuilt from scratch every time a model is asked to do something. That became the persistent layer, and I did not set out to design it. The question led there.

Once the foundation was in front of me, it explained things I had not been asking about, like how agents should actually deploy and why AI breaks down across so many different workflows. The same missing piece was underneath all of it. I had gone looking for why AI would not run a person's productivity, and I had found the layer the whole field had been building on top of without ever naming it.

I started with the whole thing, and it pulled me down to the ground it grows from. Most of the people in this conversation start at a tree. They start at the model, because the model is the thing in front of them, and they try to build outward from it. It is a reasonable instinct and it is the wrong place to stand, because the model is one part of a system that has no ground underneath it. You cannot find the foundation by climbing a tree. You find it by looking at the whole forest and asking what it grows out of.

I will say this plainly, because it is true and I have stopped apologizing for it. The architecture was not luck and it was not a guess. I reasoned it through from the question down, and it has held up to every challenge I have put to it and every one others have brought.* The engineers know things I never will, and I still saw something they had walked past. Both of those are true at once. I also could not have done it alone. Working through it with AI as a reasoning partner let me push the thinking past where my own technical knowledge would have stopped it, and the irony is not lost on me that the thing I was trying to understand is the same thing that helped me understand it.

So when people ask how an operator with no engineering background landed on an AI architecture, this is the honest answer. It was not about out-engineering anyone. I started with the forest, and the trees came after. The foundation was only ever going to be visible from there.

*The companion papers and architecture documents supporting this work are available on my LinkedIn profile.