Blog
The Fable 5 Takedown Smells Less Like AI Safety and More Like a Frontier Oligarchy
The Anthropic and Washington showdown that pulled Fable 5 and Mythos 5 dark three days after launch. The official reason is national security. It looks a lot more like consolidation in the making, and the only real hedge is staying model agnostic.
Who Types /skill: June 15, Claude Code Subscriptions, and Why It's Mostly Semantics
Anthropic starts metering programmatic Claude usage on June 15. They have a business to run, but their incentives and yours point in opposite directions: you want cheap tokens, model-agnostic workflows, and zero vendor lock. They want the opposite.
How Curiosity Turned Into Digital Infrastructure
A career note on curiosity, early bets, Zengar, and the practical digital infrastructure work that came from filling high-value gaps.
I Cut Tax Prep From 3 Hours To 30 Minutes With A Local LLM
I ran Qwen 3.6 35B locally on a 3090 to turn 150 messy expense invoices into a clean accountant package. Real ai tax preparation work: no documents uploaded to a cloud LLM, 3 hours of admin compressed into 30 minutes.
Best LLM For Coding As Of May 2026: Opus 4.7 And GPT-5.5, The Copilot And The Cracked Dev
A field report on Claude 4.7 Opus vs GPT-5.5 Codex in real autonomous coding loops, including failures, role fit, and where each model breaks.