AI Cataloguing for Fine Art Auctions

What’s hard about fine-art cataloguing

Fine art demands precision. The wrong attribution, period, or medium turns into a buyer dispute or a lost sale. Condition reports get scrutinised — losses, repairs, relining, frame condition, signature legibility. And every cataloguing decision is an editorial choice that needs to read like the rest of your catalogue.

What Snaplot does for fine-art lots

• Photo-grounded titles. “After (artist), 19th century” — not “by (artist)” — when the AI isn’t certain. Less “in the style of” guesswork than your catalogues might fear.
• Period and medium identification. From paint application, support, frame style. The AI is conservative — it won’t claim a Constable from a phone photo.
• Signature detection. When a signature is photographed close-up, Snaplot reads it. Faint or illegible signatures are flagged as such.
• Condition reports from photo evidence. Tears, losses, foxing, craquelure, frame damage — only what’s visible. Defensible if a winning bidder queries.
• Comparables lookup. Snaplot searches recent sold prices for works by the same artist, period or genre to anchor the estimate range.
• Confidence scoring. Low confidence is your signal to bring in a specialist before publishing.

Categories Snaplot handles well

• Oil paintings: 18th–20th century British, Continental, modern
• Watercolours and gouaches
• Pastels and works on paper
• Prints: engravings, etchings, lithographs, screenprints (with plate/edition info from photos)
• Sculpture: bronze, marble, terracotta, plaster, mixed media
• Decorative arts and design: furniture, ceramics with art-historical context
• Studio pottery and glass
• Maps and topographical prints

Where Snaplot is conservative on purpose

The AI does not claim attribution to a major artist unless there’s strong photographic evidence (signed and dated, with stylistic markers). For ambiguous works it returns “Manner of”, “Circle of”, “After” or “Continental School” — which is what a human cataloguer would write. You can override or upgrade attributions where you have provenance evidence the AI can’t see.

Provenance and inscriptions

Photograph any verso labels, gallery stamps, exhibition stickers or inscriptions. Snaplot uses every photo in the lot and incorporates the provenance trail into the description automatically. Better photos of marks → better description.

Will the AI claim a Lowry I have is a Lowry?

No. The AI is trained to default to “in the manner of” or “school of” unless there’s a clearly photographed signature plus stylistic indicators. Misattribution is the worst-case error so the model is intentionally conservative.

Can Snaplot replace specialist cataloguers?

No. It accelerates the routine — frame, support, dimensions, period, condition — so specialists can focus their time on attribution, provenance and the lots that genuinely need expert judgement.

Does Snaplot handle dimensions?

If you photograph a tape measure or ruler in shot, Snaplot estimates dimensions from the photo. You can also enter dimensions manually — they’ll override anything the AI infers.

More from Snaplot

• How Snaplot works →
• Pricing →
• For antiques auctions →
• For industrial sales →
• For house clearances →
• For vehicle auctions →