About

I've spent twenty-plus years working across software engineering, user experience, and product management. These days, my work has three pieces. I'm increasingly helping solo experts build real things with AI, I act as a fractional product leader for business clients, and I build my own products with AI in the cracks. They all feed into each other, where the learnings from one inform the others.

I started in software engineering, self-taught and database-heavy, building Windows products right as web software, SaaS, and the cloud emerged. Learning those technologies from the hardware up taught me how software gets built. It also taught me that we can build whatever we want, and users will do whatever they do. That's what pulled me toward the human side of software. Those first years ran through big shops, Packard Bell's support centers, American Express, and a startup, before I went out on my own.

Skybend was two guys freelancing: a business partner who was a traditional graphic designer and me. We helped companies build software and built some things of our own. One of those clients was Amirsys, which built clinical decision support software for radiologists and pathologists. The UX work there made me want to understand the business side too, so joining Amirsys to lead product was an intentional shift. I still ran user experience and design.

Elsevier acquired Amirsys a few years later, and I stayed eight years, working on several large medical products. Toward the end, I moved back closer to the technology and stood up the platform product group. My team of product managers built the shared services that a $500M portfolio of 20+ global products ran on, health record integration, search, analytics, and I directly owned the drug information products, a $25M line. Everything since has sat at the intersection of the engineering, the human factors, and the business of product. Those are the pieces I enjoy, and the work I do now uses all of them at once.

AI has made building software accessible to a whole new set of people, and it has collapsed the time between having an idea and putting something real in front of a user. What it hasn't supplied is the judgment that makes software work for anyone other than its author. That judgment—melding the user, the business, and the technology into one cohesive product—is what I've done for twenty years and what I write about now.

Some of what that looks like in practice. One of my fractional clients has a small team of engineers building a new SaaS platform, and we're adding a second product line that follows most of the patterns of the first. The UI design was in good shape, detailed mockups screen by screen, stories roughed out against them. And we hit a wall anyway. The business goal was clear, and the engineers could have forged ahead, but I could see it wasn't going to come together, because the new work exposed gaps in the edge cases and business rules, and in today's world that means nobody could tell the AI coding harness what to build and what not to break. Five years ago, an engineer would have caught those gaps mid-build, hit something weird, raised it, and moved on. Now the AI just makes something up and brings the whole thing to a halt. The cycles are so much faster today. So I called the stop, ran the refinement calls with the leads and engineers, and drove the edge cases into writing before anyone wrote code. Catching that disconnect before the build hits it is the real work.