The bottleneck in most auction operations is not the selling — it is the cataloging. Every lot needs to be photographed, measured, researched, described, and entered into a database before it can go to market. For a mid-size house running 800-lot sales, that process can consume weeks of staff time.
Two firms are now testing AI-assisted cataloging systems that promise to compress that timeline dramatically.
Kensington Auctions in Portland, Oregon, began a pilot program in January using a system built by LotVision, a startup based in Austin. The software ingests a set of photographs, identifies the object category, and generates a draft lot description including materials, approximate dimensions, period, and style.
“It’s not writing the final copy,” said Janet Okoro, Kensington’s director of operations. “It’s giving us a first draft that’s 80 percent there. Our catalogers edit rather than write from scratch, and that’s where the time savings come from.”
Okoro said the system performs well on furniture, decorative objects, and general antiques but struggles with fine distinctions — telling a period piece from a reproduction, for example, or distinguishing between similar pottery makers based on glaze alone.
“It will tell you it’s an American art pottery vase with a matte green glaze, circa early twentieth century,” she said. “It won’t tell you it’s Grueby versus Hampshire versus Teco. That’s still a human job.”
Rick Santelli, the founder of LotVision, acknowledged the limitation. “We’re not trying to replace expertise. We’re trying to eliminate the mechanical part of cataloging so the experts can focus on what they’re actually good at.”
LotVision charges a per-lot fee that Santelli declined to specify publicly, though he described it as “less than what you’d pay a part-time cataloger per lot.” Milestone Auctions in Willoughby, Ohio, is running a separate pilot with a competing system and reports similar results.
Neither house has committed to full adoption. Both say they are evaluating accuracy rates over six months before making a decision.