Articles
Strategy
AI cut man hours to almost nothing, but token spend didn't shrink with it, and none of it is worth anything if nobody goes back to confirm the user's problem actually got solved.
Operating Model
AI made it fast to build things once called 'small,' but small was always a measure of risk, not build time, and those two stopped being the same thing.
Spec-driven development promises efficiency, but a long spec written before anyone validates the problem is the same mistake as waterfall, just running at agent speed instead of committee speed.
Engineering Leadership
As AI shrinks engineering teams, the ones who stay need to do more than execute tickets. They need to sit close enough to the customer to know when a feature solves the right problem.
Product Thinking
AI agents made shipping fast. They didn't make validating whether we solved the right problem any faster, and that gap is becoming the real bottleneck.