The full library

Useful resources for building with AI.

Reader-first explanations, primary sources, practical decisions, and reproducible experiments—without the hype.

All resources

9 practical resources, with sources and working files available for inspection.

Browse by topic →

Inspect every model on one bench — price, benchmark spread, and the context question — then read the full research.

Point Codex at a $0.14 model in one command, prove it is actually routing there, and know the three things that break.

Investigate effort, conciseness, and reliability with a reproducible experiment and an honest HTML report.

Compare token price with total task cost by running the same app build across a controlled model matrix.

Look beyond one benchmark score and test where low-cost models hit a context cliff in real terminal work.

Turn headline token pricing into a real invoice test that includes retries, routing, and total task cost.

Build a cinematic scroll website from concept to generated footage, smooth playback, and a live deployment.

A practical path from zero to a Claude-powered AI consulting offer, including proof, delivery, and pricing.

Test the free Veo routes and turn Claude Code into a repeatable pipeline for creating video content.