How These Notes Are Made

Most of what’s published here goes through the same pipeline: research, drafting, fact-checking, and editing done through a human-in-the-loop workflow that includes AI tools. I use AI the way I write about using it elsewhere on this site — as a research and drafting layer, not as a replacement for judgment.

Here’s what that actually means in practice:

• Research first. Claims about specific numbers, deals, product releases, or events are checked against primary or authoritative sources — company statements, regulatory filings, reporting from outlets with a track record on the beat — before they go into a draft.
• AI drafts, I decide. AI tools help me move from research to a first draft faster, and help me catch things I’d otherwise miss. They don’t get final say on what’s true, what’s worth publishing, or how something should be framed.
• Every post is reviewed before it goes live. I read the whole thing, check the sourcing, and approve it before it’s published — not after.
• Corrections happen in place. If something turns out to be wrong, I fix it in the original post and add a dated correction note rather than quietly editing history. If you spot an error, tell me — contact info here.
• Fast-moving stories get flagged as such. Posts about breaking news — acquisitions, breaches, litigation, anything still developing — carry a disclaimer noting what’s confirmed versus reported, and a last-updated date when the underlying story is still moving.

I write a lot about AI governance, data handling, and the gap between what AI tools can do and what’s actually safe to let them do unsupervised. It would be strange to write about that and then run this site the opposite way. So this page exists partly as a practical disclosure and partly as a stake in the ground: if I’m going to argue that AI needs review, sourcing discipline, and a human who’s accountable for the output, this site should visibly practice that — not just recommend it to other people.

That’s the whole system. No hidden layer, no ghostwriting, no unreviewed auto-publishing. If something reads wrong, that’s on me, not the tools.