How This Work Gets Made

The Earned Trust page explains the marks after my name — the disclosure a reader can check. This page is the other half: the way the work is actually done, so that the disclosure has something honest to point to. It comes from a full published guide, linked at the bottom; this is the ten-second version.

It was written for people like me — doing serious work without the credential their field treats as a license to be taken seriously. The trade is simple and demanding: where a credential buys trust in advance, you earn it afterward, in the open. But the method serves anyone who would rather have their AI use be legible and checkable than assumed and suspected.

Everything rests on one rule:

1. Set a base, add specialists. Pick one model you genuinely think with — that’s your base. Then add different models as critics and verifiers. Architectural diversity matters: your critics shouldn’t share your drafter’s blind spots. Use the strongest reasoning tier you can get.

2. Draft with your base model. Feed it your half-formed thoughts and let it help you find clear prose. Ask what your idea resembles in the existing literature. And steer — if it drifts into writing a different, blander piece, pull it back. It’s your thesis.

3. Attack your own work. Take the finished draft to a different model and instruct it to try to break it — a hostile reviewer who wants to reject it. Do this with at least two independent models. This is the step most people skip, and it’s the one that earns trust.

4. Verify every source yourself. Non-negotiable. AI fails in quiet ways that look like real research until you look closely — a real institution named, the wrong finding attributed. Go read what the original actually says.

5. Integrate honestly. For every criticism, make a ruling with a reason: accept it, adapt it, or decline it. Keep the objections — a review you only agreed with is a review that taught you nothing.

6. Keep the receipts. Transcripts, reviews, rulings, version history — retained and producible on request. This is what turns a claim of rigor into evidence of it.

7. Disclose how it was made. The AIast mark: which AI services were used, what job each did, and how the result was checked — compact enough to sit in a byline, honest enough to survive an audit.

8. Publish in the open. A permanent archive, a citable DOI, an open license. The work stands where anyone can check it — because being checkable is the whole point.

The mindset, in one line: the goal is not to look right — it is to be specific enough to be wrong, and to have done the checking before your critics do.