Every year or so, AI shifts underneath us in a way that makes the previous six months of habits obsolete. In 2023, that shift was chat. In mid-2025, it was Agentic AI — moving from “ask a question, get an answer” to letting AI actually carry the work forward. Right now, we’re in the earliest days of the next one: Self-Organizing AI, where a single prompt can spawn thousands of agents that run, test, and improve their own work for weeks without anyone watching.
This month’s Blockbuster Live is my attempt to make sense of that shift while it’s still happening — not in hindsight, but live and in real time. I walk through what Self-Organizing AI actually looks like inside Claude Code, why “evals” are quietly becoming the highest-leverage skill in this new paradigm, and how I’m rethinking thought leadership itself now that AI can stress-test an idea the way we used to stress-test code. I also go deep on a concept most people have never encountered — cognitive artifacts — and why it may chan…
