Technology
Purpose-built for watch authentication.
General-purpose AI can describe a watch. Loupe is built to authenticate one — with a reference corpus, a tells knowledge base, and a tiered analysis pipeline designed for exactly this problem.
The reference corpus
850,000+ verified reference images across Rolex, Patek Philippe, Audemars Piguet and Richard Mille. Every comparison starts from verified examples of the exact reference in question.
Multi-angle view analysis
Authentication happens per view: dial against dials, caseback against casebacks, clasp against clasps. That's why our flow guides your photos — comparable views are what make visual evidence meaningful.
The tells knowledge base
39,000+ documented tells — specific, checkable discrepancies observed on counterfeit examples, from font weight on a rehaut engraving to the finishing of a clasp interior. Each verdict draws on the tells documented for your reference and its family.
Tiered analysis, honest limits
Identification, visual comparison and knowledge checks run in tiers, and uncertain cases escalate rather than guess. Loupe does not measure physical dimensions from photos — no photo-only system credibly can — and reports state what could not be assessed rather than papering over it.