AI Condition Report Writer for Auction Lots

Why condition reports are awkward

Condition reports are where buyer disputes start. Under-describe damage and a winning bidder feels misled. Over-describe and you knock the hammer price for nothing. Hand-writing them is slow and you’re tempted to skip the small stuff because there are 200 more lots to do.

How Snaplot’s condition writer works

• Photo evidence only. The AI describes wear, damage, repairs and marks it can see in your photos. It does not invent damage. It does not guess at hidden flaws.
• Conservative language. “Visible chip to the rim” rather than “as new” — the AI errs on disclosure.
• Structured output. Repeating fields (overall condition, damage, repairs, original parts, signs of restoration) so reports read consistently across the catalogue.
• Confidence flagging. Where photos are too blurred or angled poorly to assess, the AI says so — better a clear “condition cannot be assessed from photos” than a guess.

What gets caught

• Chips, hairlines, cracks (visible)
• Wear, surface scratches, paint loss
• Repairs, restoration (where visible — relining of paintings, replaced parts, soldered repairs to silver)
• Foxing, staining, water damage on works on paper
• Tarnishing, oxidation on metals
• Tyre wear, body damage on vehicles
• Wear to gilding, signs of cleaning

What needs a human

Hidden damage by definition isn’t in the photos — internal cracks in a clock, structural issues in furniture, hidden repairs to ceramics. The condition report should say “examined in person” if you’ve actually handled the item; you’ll override the AI’s photo-only flag.

Photo tips for accurate condition reports

• Bright, even light — natural light through a north-facing window is ideal.
• Shoot the item from multiple angles (3–6 photos minimum for furniture, ceramics, art).
• Take separate close-ups of any damage you’ve noticed — direct the AI’s attention.
• Photograph any maker marks separately. Better mark photos = better attribution context for the description.
• For paintings: photograph the back too — labels, stamps, frame condition.

Will the AI miss damage that’s clearly visible?

Sometimes — particularly small chips on busy backgrounds, or damage in the corner of a wide shot. The system is tuned to err on disclosure, but you’ll occasionally need to add a line. Re-runs based on a closer-up photo usually catch what was missed.

How does this protect us legally?

Snaplot doesn’t replace the auctioneer’s duty of accurate cataloguing. It produces a defensible photo-grounded draft — your responsibility is to verify and publish. The audit trail (original photos + AI output + your edits) is preserved so any post-sale query can be reconstructed.

More from Snaplot

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