Breadth was the training. This is the point.
My career doesn't fit on one line. Different industries, different rooms, different problems — for a long time that looked like restlessness. It turned out to be training. Every field runs on the same expensive grind: the reading, the checking, the sorting that nobody wants and everybody pays for.
That grind is exactly what AI is now absurdly good at — if you point it hard at one job and refuse to let it wander. Not AI as a feature. Not AI as a demo. AI aimed at a specific outcome a professional would pay for, with the receipts to prove it did the work.
Agents work my night shift at throwaway prices. Work that used to need a team now needs one bench and electricity.
When the cost of a night's work collapses to almost nothing, the old rules about scale stop applying. You don't need a platform, a headcount, or a roadmap slide. You need a small number of tools built exceptionally well — each one doing one job, priced in the open, maintained by the person who built it. That's what I make. The measure is simple: real value, real impact, or it doesn't ship.