AI integrated into real-world problem solving — architectures, specifications, and the conviction that AI expands what humanity can achieve
In the previous post, I argued that we ship one mental model to every user — not because we lack the knowledge to do better, but because we could not afford to. That post approached...
How many times have you had to learn a new app from scratch? A new project management tool, a new calendar, a new way to do the same thing you already knew how to do....
Part 1 mapped the problem. Most enterprises are context poor — they have the artifacts but not the reasoning that produced them. By 2027, the gap between the organizations that captured why-data and the ones...
Most enterprises are context poor. They do not know it yet — and many never will. But by late 2027, the gap will be visible to everyone else. This is not a doom piece. It...
NLSpec tells agents what to build. But building is only half the problem. A true dark factory — agents running autonomously, producing artifacts, shipping to production — needs a specification for how to operationalize and...
We are entering the era where the spec is the product. I wrote NLSpec — a specification system for AI coding agents — and then threw it out and rewrote it from scratch the next...
Context The five levels of AI coding Dan Shapiro wrote the definitive framing of the five levels of AI coding — modeled on the NHTSA’s five levels of driving automation: Level Analogy What you are...