Long-form articles on building for the web in the AI era — architecture, careers, tooling and the patterns behind the projects we ship. Written here, cross-posted to Medium and Substack.
Benchmark leaderboards for large language models contradict each other badly enough to be unusable. Published prices do not. This is a structured comparison built on the numbers each lab commits to contractually.
Every Claude skill you install occupies context on every message you send, whether or not it fires. This is a short overview of what that costs, which skills are worth the space, and where the ceiling sits.
