This Pendleton sat for months as 'wool shirt'. AI dated it to the 1980s from the tag in 10 seconds

The problem

Vintage sells for more and shows up in "vintage" searches, but telling a 1980s shirt from a 2010s one takes practice, and most of us don't call it when we could. A 1980s Pendleton listed as "wool shirt" sits.

The workflow

1. Shoot the care tag, brand tag, any union or RN label, and a close-up of the buttons or zipper. These are the same shots you need anyway.
2. Give an AI (we use Claude) every photo plus a written rulebook of dating clues: tag styles by era, when brands changed logos, what a union label means, zipper and button makers by decade, single-needle vs. chain stitching, "Made in USA" cutoffs for the brands you see most.
3. It answers in a fixed shape: decade, confidence, and the three facts the call rests on.
4. If it isn't sure, it asks a second AI (ChatGPT) the same question once. Two matching answers go in the draft; a disagreement comes to you.
5. The decade goes in the title and item specifics. The three clues go in a note you see when reviewing the draft, so you can check the reasoning in ten seconds.

The prompt, roughly

Here are the tag, label and hardware photos for one garment.
Using the dating rules below, give: decade, confidence (high/medium/low),
and the three strongest facts you used. If the evidence is thin, say so.
Never guess a decade without at least one tag or hardware fact.

What happened

Before the rulebook, we called vintage on a handful of items a month. After, it is called on every item that qualifies, with the reasons shown. Every correction we make becomes a test the rules have to pass before they change.