I decide what's worth building, then build it, then teach the next 1,500 or so people how to do the same thing with AI.

Every project starts with a kickoff, and everyone who touches it is in the room, even the person who shows up once and never comes back. Every. Single. Involved party. We walk through the requirements as a group until we agree on the facts, then we make decisions from those facts, together. I borrowed the approach from how solutions architects build a solution, and I keep using it because it works: people who helped define the facts buy into the solution.

I use Claude for research synthesis and first-draft speed, on prototypes and on the training curriculum I manage at Agile Defense. I still make the final call on what ships.

I'll take the boring answer that's right over the clever one that needs a follow-up meeting to explain, every time.