What we read to keep up.

Model launches, research drops, and product announcements. Watch for Claude versions and capability shifts.

What changed in the API: new models, new params, deprecations. The version log we check before bumping a model string.

Daily-ish stream of OpenAI platform changes — new models, API tweaks, dashboard updates.

Google's official log for the Gemini API. Cross-checked when comparing model behavior across labs.

Notebooks for tool use, retrieval, classification, agents, and Claude-specific patterns. The reference we copy from when teaching.

Long-running collection of working examples for the OpenAI API. Strong on evals and structured outputs.

The CLI itself. Releases tell you what new commands, hooks, and skills surfaces just landed.

The single best public running log of how the LLM space is actually evolving. Daily reading.

Wharton professor writing about how AI changes work in practice. Strong on classroom and white-collar use cases.

Evals, fine-tuning, and the engineering side of shipping LLM products. Where we go for serious-eval thinking.

Long-form conversations with the people building the AI infrastructure layer. The podcast plus the writing.

Where to start with Claude Code. We send these to every cohort student in week 1.

The TypeScript toolkit that powers most of our shipped tools. Releases are the signal for new model integrations.