A field guide for using AI agents after the demo ends.
Subscribe for practical, first-person notes on running agents in real work: context rot, tool failures, local models, evals, automation loops, and the small patterns that make agents useful instead of theatrical.
Real receiptsWorking artifacts, screenshots, failed runs, and fixes — not generic AI takes.
Operator judgmentWhen to chain agents, when to code it, and when the whole workflow is fake productivity.
Readable systemsPlain-English maps of the infrastructure, economics, and UX pressure shaping agent products.
Start here:AI Compute Landscape·Prompt Tax·About Peter