LLM tooling pioneer, the most prolific AI blogger
Simon Willison
Profile
Simon Willison is the developer most other developers read when they want to know what LLMs actually do. A British engineer who co-created the Django web framework with Adrian Holovaty in 2003 at the Lawrence Journal-World newspaper, he spent a career building: Yahoo! Technology Development, early Fire Eagle geolocation, software architect at The Guardian, co-founding the conference directory Lanyrd with his wife Natalie Downe (Y Combinator, 2011; acquired by Eventbrite in 2013). He has been an independent open-source developer since 2020.
His two flagship projects are Datasette, a tool for exploring and publishing SQLite data as an interactive website, and LLM, a command-line utility and Python library for running prompts against every major model — OpenAI, Anthropic, Gemini, Mistral, local models via Ollama and llama.cpp — with a plugin architecture, SQLite logging, and, more recently, tool use. LLM is the cleanest way to put an API key in your shell and start actually using these models like a Unix citizen.
What makes him unavoidable is the blog. simonwillison.net was the most popular blog on Hacker News three years running, with 1,000+ posts a year — a mix of link commentary, deep technical write-ups, and meticulously annotated talks. He coined the term prompt injection in September 2022 and has been the clearest voice on LLM security ever since, later coining the lethal trifecta (private data + untrusted input + external communication) that describes why agentic systems keep getting owned. He has no vendor allegiance, runs every new model through his pelican-on-a-bicycle SVG benchmark, and explains what works in public. He also serves on the board of the Python Software Foundation.
For developers learning AI in 2026, Simon’s blog is probably the single highest-signal resource on the practical side of the field. He shows, not tells — every claim comes with a working example, a terminal transcript, or a one-liner you can run yourself.
Key Articles & Papers
Prompt injection attacks against GPT-3 Things we learned about LLMs in 2024 Here's how I use LLMs to help me write code The lethal trifecta for AI agents Pelicans on a bicycle Imitation Intelligence — PyCon US 2024 keynote The last six months in LLMs, illustrated by pelicans on bicycles Large Language Models can run tools in your terminal with LLM 0.26Spotify Podcasts
![[Ride Home] Simon Willison: Things we learned about LLMs in 2024](/images/people/simon-willison/spotify/72WjmOtEexIxYBwplwJpLx_hu_a4836e4e4bb8fccc.webp)
