Recent Themes
A summary of what I've been writing about lately, generated from my 10 most recent essays using Claude on AWS Bedrock. Updated every time the site is published.
Organizational readiness matters more than AI capability. The binding constraint on AI's impact has shifted from what technology can do to how quickly institutions can absorb it. Organizations that treat AI as a posture rather than a project stay ahead; those that standardize too early scale the wrong thing. (Occam's Inversion, The Field and The Frontier)
Making process visible accelerates collective learning. Organizations circulate finished work while hiding the inquiry behind it. When leaders show how work was made — including AI's role — they give others the reference points needed to experiment with confidence, and norms shift through observed behavior, not policy. (How This Was Made)
Compounding requires cheap evaluation, not just cheap generation. AI makes producing work nearly free, but value compounds only when the cost of judging whether work improved is also low. The durable asset is the evaluator — the rubrics, objection sets, and feedback loops — not the output. (The Cost of the Answer)
Assumptions expire faster than organizations notice. Every consequential decision has a half-life. AI is accelerating how quickly the territory changes, and organizations need a "Notice to Mariners" system — recording why decisions were made and the conditions under which they should be revisited — not just better forecasting. (The Half-Life of an Assumption, The Barcode Bargain)
As systems run more autonomously, human judgment concentrates rather than shrinks. When AI handles execution, the human role shifts from doing to directing — setting purpose, noticing what matters, and making the calls no system can make for itself. Removing scarcity-as-governor forces leaders to name the tradeoffs they were previously able to avoid. (What The Garden Is For, The Silent Governor)
Productive tension is an organizational capability, not a problem to resolve. Healthy institutions in rapidly changing environments aren't the most stable or the most chaotic — they hold themselves far enough from equilibrium to adapt while retaining enough coherence to be trusted. Standardization applied too early freezes the wrong answer. (Beautiful Tension)