Description
This is a free preview of a paid episode. To hear more, visit natesnewsletter.substack.com Sam Altman has a confession. Despite being the CEO of OpenAI, despite having the best access to the most capable AI models on the planet, despite his own internal data showing that AI now beats human experts on three-quarters of well-scoped knowledge tasks — he still hasn’t really changed how he works. “I still kind of run my workflow in very much the same way,” Altman admitted in a recent conversation with Alex Kantrowitz, “although I know that I could be using AI much more than I am.” Something fundamental shifted in December 2025. The people closest to the technology are calling it a phase transition, a threshold crossing, a break in the timeline. Andrej Karpathy — who helped build OpenAI and has been writing code professionally for two decades — says his workflow inverted in a matter of weeks: from 80% manual coding to 80% AI agents. Ethan Mollick, the Wharton professor who tracks AI adoption, put it bluntly: projects from six weeks ago might already be obsolete. And yet most people, including the CEO of the company driving much of this change, haven’t caught up. The capability is here. The adoption isn’t — and the gap between those two things is where all the leverage lives right now. Here’s what’s inside: * The December convergence. Three frontier models in six days, two viral orchestration patterns, and a platform feature that made both obsolete within weeks. * The browser that shouldn’t exist. What it means that agents generated 3 million lines of working code in a week — and why “existence proof” undersells it. * The self-acceleration loop. Anthropic’s engineers stopped writing code. OpenAI is slowing hiring. The internal benchmarks explain why. * The capability overhang. The growing gap between what AI can do and what most people are doing with it — and why that gap is a temporary, closeable arbitrage. * The specification shift. Why the developers getting the most value aren’t the ones prompting fastest — they’re the ones who learned to describe what they want precisely enough that agents can build it. Let me show you what happened — and what to do about it. Subscribers get all posts like these!