Simon Willison is the practitioner's practitioner — the developer whose daily blogging on LLMs has become one of the most reliable signals of what actually works in agent and model usage. His tools (llm, datasette, dozens of small CLIs) and the experiments he documents publicly have outsized influence on how indie devs and small teams adopt new AI primitives.

Coverage on this page tracks Simon's tool releases, his read on new model capabilities, and the patterns he surfaces from running real workloads against frontier models in the open.