A son’s idle curiosity, a free chatbot, and a 60-year-old secondhand find collided into a six-figure auction result.
A painting Helene Plotkin bought for under $100 at a New York thrift store in 1966 sold at auction this month for roughly $254,000, after her son uploaded a photo to Google Gemini. The AI flagged it as a likely work by Scottish Colorist F.C.B. Cadell, and appraisers later confirmed it.
Picture this. An 88-year-old former art teacher in Florida has loved a colorful portrait of a woman in black for six decades. She always sensed something “regal” about it, but never once thought it was worth a fortune. Then her 60-year-old son pulls out his phone, snaps a picture, and types a question into a chatbot. Minutes later, the family is sitting on a hidden masterpiece.
What Happened
Barry Plotkin was visiting his mother’s home when a thought hit him: why not ask Gemini? He photographed the painting, uploaded it, and asked what the AI could tell him.
The response stunned him. Gemini pointed to the orange accents, Art Deco styling, and lilac backdrop as “unmistakably” Cadell, and tied the work to the Scottish Colorists, a group of four early-20th-century painters who blended Fauvist and French Impressionist influences into British art.
Then it went further. Gemini told the family this was a large-scale 1920s studio portrait of one of the artist’s muses, painted in his most famous Edinburgh studio. It even told Barry to check the back of the canvas, where he and his mother found an auction marking, a canvas stamp, and a processing date.
The painting later sold to a private buyer for £189,200 with fees, about $254,000.
The Backstory
Helene paid less than $100 for the painting in 1966 and hung it on her wall, where it stayed until last December. She has an art degree and always admired the brushwork, but the idea of authenticating it never felt worth the trouble.
Barry put it plainly: hauling the painting to an auction house, only to be turned away, would have been a hassle. That friction is the whole story. For decades, the cost and effort of getting an expert opinion outweighed the perceived odds of a payoff. So nobody bothered.
Here is the kicker. Months before the painting landed in suburban New York in 1966, Christie’s in London had sold it for £21, roughly $600 in today’s money. From £21 to £189,200 in sixty years.
The Plan
Gemini did not just identify the work. It handed the family a playbook. It recommended next steps for verification, including contacting an auction house and a professional appraiser, and named specific specialists, Nick Curnow and Alice Strang at Lyon & Turnbull.
The humans took over from there. Strang and Curnow confirmed most of the AI’s read through additional research and technical analysis, including inspection under three kinds of light. They titled the work “Interior: The Lady in Black,” mirroring Cadell’s own naming style.
“This is the stuff of auctioneers’ dreams,” Strang said.
The Business Model Angle
Here is the pattern worth stealing. AI did not replace the expert. It replaced the reason not to call the expert.
Authentication and appraisal have always sat behind a wall of cost, access, and intimidation. The trained human eye was the product, and it was expensive to reach. What Gemini did was collapse the top of the funnel. It turned a vague “I wonder” into a confident “go talk to Lyon & Turnbull, ask for these two people.”
Strang admitted the quiet part out loud: she could have attributed the painting on her own, but without AI, the family might never have walked through her door. Read that again. The AI did not do the appraiser’s job. It generated the appraiser’s lead.
For founders, the lesson is sharp. The biggest opportunity in AI right now is not automating the expert. It is automating the discovery layer that feeds high-touch, high-margin human services. Wherever valuable expertise is gated behind friction, hassle, intimidation, or uncertain payoff, there is a business in being the cheap, instant first opinion that converts curiosity into qualified demand. Think wealth advisory, legal triage, medical second opinions, property valuation, IP assessment. The money is in the handoff, not the replacement.
It is also a reminder of how a platform like Google compounds. Gemini is free, and it drove a real-world transaction that touched auction houses, appraisers, and buyers. That is the multisided platform logic that powers Google’s business model extending into yet another corner of the economy.
The Risk
Now the cold water. Gemini got something wrong. It identified the model as Bethia Hamilton Don Wauchope. The appraisers pinpointed her as May Easter, matching the turban she wears in Cadell’s “Pink and Gold.” Close, confident, and incorrect.
That gap matters. A chatbot that sounds certain while being wrong is exactly how overconfidence spreads. Scale this story and auction houses get flooded with false positives, people overpay for “AI-verified” fakes, and the trained human eye becomes more necessary, not less.
There is also survivorship bias baked into the headline. We hear about the $254,000 win. We do not hear about the millions of “your painting is worthless” verdicts, or the wrong attributions that never get corrected because no expert checks them. One viral success does not make AI a reliable appraiser. It makes it a useful first filter that still needs a human to close.
Quick Questions
Can AI really tell if a painting is valuable?
It can give you a fast, surprisingly informed first guess. In this case Gemini correctly flagged the artist and era. But it also misidentified the model, so treat it as a lead, not a verdict.
What AI did the son use?
Google Gemini. He uploaded a single phone photo and asked what it could tell him about the painting.
How much did the thrift store painting sell for?
It sold to a private buyer this month for £189,200 with fees, about $254,000. The original cost was under $100 in 1966.
Did the AI get everything right?
No. Gemini nailed the artist, style, and rough date, but named the wrong model. Human appraisers at Lyon & Turnbull corrected it and confirmed the rest.
The Bottom Line
The winning move was not the AI or the appraiser. It was the AI removing the friction that kept the appraiser out of reach for sixty years. If you are building, look for the expensive, intimidating, high-friction “first step” in any valuable service, and become the free, instant version of it. Just remember whose job it is to close: still a human’s.
Source: The New York Times
