Borrowed Intelligence
A 4-part series · best read from the top, in order.
-
Sovereign AI: The Friday the US Pulled a Model, and Every Country Started Building Its Own
What sovereign AI actually means. Starting from the day the US cut off Anthropic's Fable 5 and Mythos 5 to foreigners, a look at the three layers of sovereignty — data, model, compute — where the real bottleneck sits, and how Korea's state-backed 'national champion AI' fits in. How dependence on AI turned into a security question.
-
Why Flat-Rate Pricing Breaks Down Against AI Agents: Anthropic Pulled Out Usage-Based Billing, Then Pulled It Back
Why does a flat-rate subscription break down against an AI agent that runs on its own? Starting from the day Anthropic shelved its plan to split Agent SDK and usage-based credits out of the subscription — on the very day it was due — and reading the structure through average pricing, marginal cost, and unit economics. With the OpenAI price war and an IPO as market variables.
-
Writing for a Web That Stopped Clicking: Zero-Click, HTTP 402, and GEO
What 'zero-click' actually means. As search and AI hand over the answer on the spot and the click disappears, what is left to reward the people who write? A look at the bot toll (pay-per-crawl) that woke a 30-year-dormant HTTP 402 code, the opposite exit of GEO (generative engine optimization), and citation as the new currency a writer can still hold.
-
The One-Person Unicorn: What Remains When AI Replaces the Team
What is a one-person unicorn? The phrase began as a bet in Sam Altman's group chat and now names an era of solo founders running entire teams of AI agents. The data that upends the conventional wisdom against going solo, the three functions of a team AI still can't fill — execution, verification, and buffering — and the research showing that people who succeeded alone were never really alone.