Lessons, opinions, and post-mortems from shipping production AI in messy, regulated, high-stakes domains.
Tech is cutting juniors at the fastest rate in history while spending $700B on AI. The senior engineers companies still depend on take 7–10 years to grow. The pipeline is being unbuilt — and the price shock is mathematically inevitable.
Read post →By early 2026, 85% of developers use AI daily and individual output is up 25–55%. Team productivity hasn't budged past 10%. The bottleneck isn't the generator — it's the human context window. A proposal: Maximum Effective Token Usage.
Read post →A B2B platform doubled its member base from ~1,300 to ~3,200 companies in a year — same six-person team, weekly releases throughout. The lever wasn't headcount; it was embedding AI across the development workflow.
Read post →Building an AI agent that touches real money, on real chains, with weekly releases and a mid-flight migration between chains. What it took, from an engineering-leadership angle.
Read post →Programmatic advertising is a black box. Budgets flow through dozens of intermediaries and nobody can tell which part of the supply chain actually earned the conversion. Inside the Media Economics Platform we built from zero — and the production results.
Read post →