Simon Willison’s Newsletter

Cadence

Weekly

Latest

2026-10-28

By

Simon Willison

Based on

12 recent issues

• Covers: Developments in LLMs, AI agents, Python, web engineering, open source tools, and software development practices.
• For: Software engineers, data scientists, and developers working with AI and LLM technologies.
• Best for: Engineers and technical teams integrating LLMs into software products and workflows.
• Skip if: Non-technical business professionals looking for general corporate strategy or non-technical business news.

Start with

• Understanding ChatGPT Work: Explains features and practical mechanics of ChatGPT Work for professional workflows.
• Conceptual integrity and counting lines of code: Discusses software engineering principles and code measurement concepts for developers.
• Jev introduces a new shape of LLM - System One, aka Decision Models: Analyzes emerging LLM architectures and decision models for technical implementation.

Sponsors seen

• WorkOSSponsor

  Sponsor message: An agent’s job isn’t known until runtime, so permissions written in advance end up too tight or too loose. WorkOS Airlock derives them from the task instead.

  The October inflection point for vibe game development · 2026-10-28
• PomeriumSponsor

  Sponsor message: Static OAuth scopes and leaky sandboxes aren’t enough to protect autonomous AI agents. Legacy playbooks and threat models fail when applied to dynamic agentic workflows. Pomerium’s fine-grained access control

  I built a new feature for my blog using my voice · 2026-10-12
• WorkOSSponsor

  Sponsor message: Agents are the new front door to your app. auth.md by WorkOS is an open protocol that lets them register users on the fly

  Navier–Stokes, RubyGems attacked, GIS and Blender with GPT-6 Astra · 2026-09-14
• PortnoxSponsor

  Sponsor message: AI agents are your fastest-growing security risk. Portnox delivers Zero Trust for AI, partnering with CrowdStrike, SentinelOne, and MS Defender

  GPT-6 Astra, Claude Fable 5.1, and yet more rogue agent attacks · 2026-09-07
• GreptileSponsor

  Sponsor message: AI agents that review and test pull requests with full context of the codebase. See Greptile catch bugs in some of the most popular open source repos

  2026 in LLMs (so far) · 2026-09-28
• TeleportSponsor

  Sponsor message: Pressure wash your codebase with LLMs to find vulnerabilities. Human bandwidth is no longer the issue when it comes to finding software security bugs – but keeping experienced human experts in the loop matters. LLMs are now exceptional at finding security vulnerabilities with simple prompting. Teleport dedicated 13 of their best devs

  Jev introduces a new shape of LLM - System One, aka Decision Models · 2026-09-21
• TeleportSponsor

  Sponsor message: What’s scarier than an engineer with prod access? An AI agent with the same access. All jokes aside – agents spawn sub-agents, run a thousand sessions over lunch, and can delete prod before dessert. The old identity problems are the same, just faster and at a bigger scale. Teleport shows how ephemeral, least-privileged access keeps agents in check.

  Battle Drones - a vibe coded RTS with audio-based multiplayer · 2026-10-21
• TeleportSponsor

  Sponsor message: Constrain AI agent abilities with trusted runtimes and Agentic Classifiers AI agents run 24/7 and will try anything to achieve their goal – at 100x the scale of a human workforce. Teleport helps you deploy AI safely

  Conceptual integrity and counting lines of code · 2026-08-23
• DynatraceSponsor

  Sponsor message: When agents enter the SDLC, observability becomes the enabler to move from code generation to scalable engineering. Dynatrace’s engineering blog outlines a framework

  Now we have a timeline of the OpenAI accidental attack against Hugging Face · 2026-08-08
• WorkOSSponsor

  Sponsor message: Agents are the new front door to your app. auth.md by WorkOS is an open protocol

  Qwen 3.8 27B is excellent, but it defaults to wildly overthinking things · 2026-08-17
• Flux TTSSponsor

  Sponsor message: Voice agents sound off when TTS has no idea what was just said. Flux TTS carries context across turns

  OpenAI DevDay 2026 live blog · 2026-10-05
• GreptileSponsor

  Sponsor message: Static code review has a ceiling. It can reason about what the code says. It can’t tell you what it does. TREX (which stands for “Test, Run, Execute”) is Greptile’s response to that ceiling

  Understanding ChatGPT Work · 2026-08-31

Visit

Summarised by AI from 12 recent issues. Audience size is not included. Prices are not included. Report a mistake.