I run my work like a company — and AI is the team.

For two decades I led engineering and product across e-commerce, logistics, supply chain, and marketplaces — most recently as a Director of Engineering at a big-tech company — building systems that move millions of orders and decisions a day. I spent that time learning how large, messy, real-world machines actually hold together.

Now I am pointed at a different problem: running my own work the way I would run a company, with AI as the labor and me as the founder who reviews, audits, and decides. Less "open a chatbot and ask," more "design the machine that does the work, then check it." Most of what I am learning does not come from prompts — it comes from building the systems around the model.

I write here to force myself to understand that machinery deeply — context, memory, agents, the harness around the model — and then to say it plainly. The bar I hold myself to: if I cannot explain a thing with an everyday analogy and one honest diagram, I do not understand it yet.

The model was never the bottleneck.
The harness was.