AI

Everyone says the job turned into project management. It's the opposite — the artifacts took over the translating, and the role stopped being inhabitable.

Everyone's writing obituaries for SaaS and for open source. They're the same obituary — and software isn't the body.

Organizational ecology has spent decades documenting how industries are born, shake out, and consolidate — and the AI industry is following the script, with a few structural twists worth naming.

AI dissolves accidental complexity at speed, but essential complexity — the question of whether it all adds up to something good — remains stubbornly human.

AI can reproduce substantial software systems. The moat against substitution scales with the blast radius of failure — but the line between what's safe and what's vulnerable keeps moving.

Four scenarios for how AI gets absorbed into the real world — and why all four are happening at once.

I've always used writing to think. This time I'm using AI to write — not to produce text faster, but to think harder.

Cycles are getting shorter everywhere — in careers, in technology, in how organizations work. A decades-old Marine Corps doctrine saw it first, and its insights about coordination under pressure feel uncomfortably relevant to building software with AI.