Some posts here carry a note at the top saying they were co-authored with AI. This page explains what that means, so you can read them knowing how they were made.

Every post starts with something I have done. A decision I got wrong, a pattern I have seen at more than one company, an argument I had with someone at work. The model does not supply any of that.

What it supplies is the second draft. I write rough notes or a full first pass, then work through it with Claude Code. This site has a style guide checked into the repository: short sentences, plain words, active voice, British spelling, no filler. The model edits against that guide. I push back when it flattens a line worth keeping.

It also argues with me before I write anything. I get questioned on the idea itself: which claim I cannot back up, where the argument is thin, what the obvious objection is. Most of the value sits here, before any prose exists. Plenty of what I planned to write has not survived it.

I read every sentence before it publishes. If a sentence claims something I have not seen, it goes. If it sounds clever and says nothing, it goes. My name is on the post, so the argument has to be one I will defend in a room.

I label it because I write a lot about using AI at work. It would be odd to argue for it and then hide that I use it. The note is not a disclaimer. The judgement is still mine, and so are the mistakes.

Read those posts the way you would read anything an editor worked on with me. Someone helped with the words. Nobody helped with the opinion.