Claude Code, AI-native engineering, and building the future with LLMs.
AI
I burned a month of allowance in five days, so I measured where every token went. Almost all of it is the agent re-reading what it had already seen. Every meter Claude and Codex run, which multiplier owns which share of the bill, and each fix priced against the real distribution.
AI
We made the most dangerous module in the system harder to change, and throughput did not move. The gate got stricter. The humans just stopped being the thing it runs on.
AI
A friend sent me obra/superpowers. I read the whole ecosystem instead — nine frameworks, what they mechanically do, and the nine conventions they all independently converged on. Then the honest case against building your own, and what I'm stealing anyway.
AI
Every review said great hardware, no apps, wait. Then in May 2026 Meta opened the in-lens display to developers and the loudest criticism stopped being true. The full review: real cost, what breaks, what people returned it for, and whether it's worth it now.
AI
35% of enterprises have already replaced a SaaS product with software they built themselves. If you can do that to a vendor, someone can do it to you. A re-audit of the Seven Powers for an era where implementation is free — and eight things that still hold.
AI
Multi-millionaire software engineers are coming. Not because AI writes code — because a system that turns intent into verified work while you sleep changes what one person can own. The mechanism, the numbers, and the part where the skeptics are right.
AI
Anthropic shipped Claude Opus 5: near-Fable-5 intelligence at Opus 4.8's exact price ($5/$25), plus an effort dial that trades cost for capability. On Anthropic's own numbers it's within 0.5% of Fable at half the cost. One-stop-shop — what shipped, what's rumor, what to switch to.
AI
Prompt, context, loop — and now graph engineering. Half the label is marketing; the shift under it is real: stop searching a flat pile of documents by similarity, start walking a typed graph of what's connected. ELI5 to the arithmetic that kills it in production.
AI
Two independent model passes consolidated into one PR comment, review queues that run on a schedule, and human judgment moved earlier — to the spec. How I actually review code now.
AI
OpenAI's models escaped a sandboxed evaluation and reached Hugging Face's production systems — apparently to obtain a benchmark's solution. What happened, how the attack chained together, and what it changes for anyone building with agents.
AI
Anthropic said Fable 5 was included in my Max plan starting July 20th. My terminal spent every session since sending me to a credits page instead. The bug behind it, the pattern behind that bug, and why Opus 4.8 is my default again.
AI
China's answer to Claude and Codex is a 2.8-trillion-parameter open model that topped the coding leaderboard two days after launch — and priced itself like the frontier it just joined. Here's whether Kimi K3 belongs in your stack.