Everything I have written.

Kent Beck's explore/expand/extract framework reframed how I was reading my team's uneven AI adoption. I'd been quietly treating it like a ladder with laggards and leaders. It's actually parallel experiments, and the unevenness is the best data I have.

If your team's AI coding sessions feel like re-teaching the model from scratch every time, this is for you. We built a constitution → auto-scoped instructions → specialized agents → external skills → /remember stack and hardened it before rolling it out.

2,334 shows, 441 songs, 593 venues, all playable in-browser alongside the setlist. If you want Dead stats, you go to one site. If you want to actually listen, you go somewhere else. Nobody had put those two things in the same place, so I built it.

It's blind debugging: the ability to mentally step through complex state changes without reaching for a chat window. The research is more specific than most people realize, and frontier teams are actively fighting back.

The pushback on "don't let your skills atrophy" didn't come from me. It came from my engineers, who wanted that language explicit and fleshed out. What I actually walked away with was something I didn't expect: my team is thoughtful.

I've been telling you to embrace AI agents. Turns out I left out the part where they can become insider threats. My first LinkedIn article covers the security risks I kept quiet about while championing adoption.

Not because I got faster at writing code. Because the tools got precise. She was in tears over spelling practice with other apps. They were all about catching mistakes. Mine separates practice from judgment.