The Operator Story
For 25 years, I've run Bostig, a VW engine conversion systems business. I designed, built, or augmented the internal systems the operation runs on. Procurement, planning, inventory, fulfillment.
Alloy Oven is my newer venture. It's an outdoor pizza oven company, and I'm building it from the ground up with AI. Product R&D, engineering validation, operational architecture. AI isn't bolted on. It's the foundation.
DeftX is where I get to build or improve systems for other people's businesses.
The Burner Spec
Last year I needed a validated propane burner specification for an outdoor pizza oven I'm developing. Normally you'd hire a specialized engineering firm. It would take months and cost tens of thousands of dollars. I had zero experience with computational fluid dynamics.
I used AI to learn CFD fundamentals, build simulation models, and run chemistry validation. The entire process took 10 days of fractional work. No outside engineering firm. No manual code. The output was a complete, validated burner and subsystem specification.
A contact at Insight Numerics, a firm that uses CFD as part of their flame detection products, reviewed the work. He said getting to that point should have taken months or years of specialized experience.
That's what I mean when I talk about compressing expertise.
How I Manage the Work
Both businesses are heavily automated by AI systems I built. That's not a conflict with consulting. It's the proof that the consulting works. Consulting clients get my full attention.
How I Think About AI
Most companies approach AI the wrong way. They hire a consulting firm that spends three months on an assessment, produces a PDF with recommendations, and then leaves. The company is exactly where it started, minus the consulting fee.
I do the opposite. You bring me a problem. I build a system that solves it. We test it together. If it works, we keep building. If it doesn't, we know in a week, not a quarter.
I don't specialize in one industry or one type of tool. If I understand the problem well enough, I can build a solution for it. I did it for computational fluid dynamics without ever having touched the field. I did it for materials planning. I did it for structured AI reasoning. The pattern is the same every time: learn fast, build fast, validate fast.
That's Knowledge Compression. It's not a framework I sell. It's how I work.
If this is how you want AI built for your business, I'd like to hear what you're working on.
See How I Work