Simon Willison's Blog

Who writes it

Simon Willison has blogged about web development since 2002. He is a co-creator of the Django web framework, the creator of Datasette, an open source tool for exploring and publishing data, and the author of LLM, a command-line tool and Python library for working with language models. He was previously an engineering director at Eventbrite and now works as an independent open source developer.

What you will find there

The site mixes several formats:

• Entries, longer posts such as first impressions of a new model release or a detailed write-up of a technique.
• Links with commentary, which make up much of the daily output.
• Quotes from other people worth keeping.
• Notes, guides and TILs (short write-ups of small things learned).
• Release notes for his own tools.

The balance has shifted heavily toward AI. As of September 2026, the site shows nearly 2,000 posts tagged generative AI and more than 600 tagged LLM, and posts often appear several times a day when new models launch.

Why it is worth following

The signal-to-noise ratio. The field produces enormous volumes of confident commentary from people who have not run the thing they describe. Here, claims get tested, results are reported including the disappointing ones, and limitations are stated plainly. Recurring themes include:

• Prompt injection, why it remains unsolved, and what that means for agents with access to your data.
• Running models locally and comparing them with hosted ones.
• Coding agents and agentic engineering, with practical accounts of what they do well and badly.
• Model pricing and comparisons, including his informal test of asking every model to draw a pelican riding a bicycle.

How to follow it without drowning

The full feed is a lot. Options, from lightest to heaviest:

• Paid monthly newsletter: a short catch-up sent to GitHub sponsors at $10 a month or more, pitched as a ten minute read.
• Free weekly-ish newsletter of the blog's content.
• Entries-only feed: subscribe to just the longer posts through the Atom feed.
• Tag feeds: every tag has its own feed, so you can follow only topics you care about.
• Everything, for people whose job is keeping up.

Students are usually best served by the entries feed plus reading whatever gets linked in class or at work.

Disclosures

The about page states that he is not paid to write about specific topics, that weekly sponsors (a banner on the site since February 2026) have no editorial influence, and it lists his other paid work. That transparency is part of why the blog is widely trusted.

Common questions

Is it too advanced for a student? Some posts assume experience, but many are accessible walkthroughs of trying a new tool. Skip what does not make sense yet.

Is it only about AI? Mostly now, with data tools, SQLite and web development still appearing.

Best for: engineers and students who want to stay current on language models and coding agents without the hype.