Full Glossary
AI Cataloging

Identify-First Protocol

An AI cataloging methodology where the system identifies what an item actually IS from visual evidence before considering brand names, labels, or text visible on the item. This prevents misidentification errors where AI assumes an item is genuine based solely on visible branding.

How It Works in Practice

Standard AI vision tools read a "Tiffany" label and describe the item as authentic Tiffany. Identify-first protocol analyzes the construction, materials, proportions, and manufacturing quality first — then cross-references visible branding as supporting (or contradicting) evidence. This is critical for auction cataloging where reproductions, fakes, and misattributed items are common. The protocol uses hedging language ('marked Tiffany,' 'appears to be,' 'in the style of') when visual evidence doesn't fully support a brand attribution.

Frequently Asked Questions

Why does identify-first matter for AI cataloging?
Without identify-first, AI reads a 'Louis Vuitton' label and describes the item as an authentic Louis Vuitton bag — even if it's a $15 knockoff. Identify-first examines stitching, materials, hardware, and construction quality before trusting the label. This protects the auctioneer from inadvertently misrepresenting items, which can create legal liability and damage reputation.
How does identify-first prevent misidentification?
The system analyzes visual evidence in order: (1) what is this object physically, (2) what material is it made from, (3) what era and style does it represent, (4) do visible marks or labels support or contradict the physical evidence. If a mark says 'Sterling' but the item shows signs of plating, the description uses hedging language: 'marked sterling, surface shows wear consistent with silver plate.'

Catalog Faster with AI

Gavelist generates professional lot descriptions from your photos in seconds — across every auction category, at any volume.