Simon Willison — Independent developer and writer documenting practical experiments, tools, and deep analysis around large language models, generative AI, web development, security, and emerging programming workflows through detailed posts and daily TILs.
86 articles from this blog
Introduces GLM-5, a massive new open-source AI model, and discusses the emerging concept of 'Agentic Engineering' for software development with LLMs.
Introduces cysqlite, a new SQLite driver for Python written in Cython, with improved transaction handling and WASM support for Pyodide.
Introduces Showboat and Rodney, two new CLI tools designed to help AI coding agents demonstrate and test the software they build.
Research paper analyzes LLM performance on large SQL schemas, comparing 11 models across 4 data formats for structured context engineering in agentic systems.
A study finds AI tools increase cognitive load and work intensity, leading to potential burnout despite productivity gains.
Anthropic's Claude AI reportedly discovered 500 zero-day vulnerabilities, sparking debate on AI's role in security research.
Mitchell Hashimoto introduces Vouch, a system to combat low-quality AI-generated PRs in open source by requiring user vouching.
Anthropic introduces a faster 'fast mode' for Claude Opus 4.6 at a significantly higher cost, with a temporary discount.
A programmer shares his joy and reduced fear about AI coding agents, quoting David Crawshaw on how they enhance programming.
StrongDM's AI team describes a 'Software Factory' where AI agents write and test code without human review, exploring fully automated development.
Tom Dale discusses the mental health impact on software engineers due to rapid AI-driven change and cognitive overload.
Explains how to run Pydantic's Monty, a sandboxed Python subset written in Rust, in WebAssembly for secure, browser-based execution of LLM-generated code.
Heroku announces a shift to a 'sustaining engineering model,' focusing on stability over new features, prompting user migration concerns.
An OpenAI researcher describes using Codex AI to automate code exploration, experiment setup, and hyperparameter tuning for unfamiliar codebases.
Mitchell Hashimoto shares unconventional tips for effectively integrating AI coding agents into a developer's workflow to boost productivity.
Anthropic releases Opus 4.6 and OpenAI releases GPT-5.3-Codex, with analysis on their incremental improvements over previous models.
Mistral releases Voxtral Transcribe 2, a new family of audio-to-text models, including an open-weights real-time transcription model.
Explains how to distribute Go CLI tools like sqlite-scanner via PyPI using go-to-wheel, making them easily installable with pip/uv.
Deno Sandbox is a new hosted sandbox product from the Deno team, part of Deno Deploy, allowing code execution via Python and JavaScript libraries.
OpenAI releases a new macOS app for its Codex coding agent, featuring a UI, Skills support, and scheduled Automations.
