How I write

Where the notes come from

A note starts from something I actually ran into: a decision that was hard to reverse, a system that behaved differently from its documentation, a trade off I got wrong the first time. It is not written to rank for a search term, which is also why there are so few of them.

What AI does, and what it does not

I use AI tooling to draft, to restructure, and to argue against my own first version. I also build AI systems for a living, so pretending otherwise here would be strange. What AI does not do is decide what is true: every claim, number and date is checked by me before publication, and anything I cannot verify comes out rather than getting hedged.

Nothing is published that I have not read line by line and would not defend in a meeting.

Sources and attribution

When a note leans on someone else's work, it says whose and links to it. If a claim has no source I can point at, it is written as my experience, in the first person, so you can weigh it accordingly. An authority I cannot produce is not an authority.

Client work stays confidential

A note never contains client code, client data, or anything told to me in confidence. Where a story needs context, it is generalised until it could be any of several systems.

Corrections

If something here is wrong, write to contact@sheepit.eu and it gets fixed. Substantive corrections are noted in the note itself rather than quietly edited away, because a page that changes silently is worth less than one that admits it was wrong.