Robert Levine recently closed on the sale of his four-bedroom three-bathroom home in Cooper City, FL, for $954,800. But instead of thanking a real estate agent, he thanked ChatGPT.
Levine says the chatbot helped him list the property, draw five offers after only 15 showings, and close for $100,000 over what other agents had told him his home was worth.
The story, first reported by Fortune, quickly became a flashpoint in the broader powder keg debate over whether AI tools like ChatGPT can replace the work of everyone from humble writers to highly specialized professions like lawyers—and now real estate professionals.
But according to Ines Hegedus-Garcia, managing partner at Avanti Way East Miami, where the buyer’s agent works, the viral takeaway may be too simple.
In her view, the sale shows that while ChatGPT may have helped Levine get to market, it didn’t help him fully capitalize on a multioffer scenario. And in a slow-moving market, that could have left the biggest advantage unrealized.
The pricing story is murkier than headlines suggest
Hegedus-Garcia is careful not to claim she knows the full seller-side picture.
“I don’t know if it’s true that he actually met with other agents,” she says of Levine’s claim that local agents wanted to list the home for far less. "Obviously, I don’t know the seller side of the story."
Still, after reviewing every nonwaterfront family-home sale in the area over the last year, she tells Realtor.com® that she came away skeptical of the idea that the sale proved ChatGPT had nailed the pricing.
Her review, which she shared with Realtor.com, covered 31 sales. Levine’s home sold for $358.41 per square foot, above the area's average of $339.20. But Hegedus-Garcia argues that the broader average may understate what this particular home could have fetched, because it was not an average property.
Levine's home had a larger lot, updates, and stronger overall condition than many homes in that wider sample. Just as important, she says, the listing may not have fully conveyed that.
“She said that the pictures didn’t do it justice,” Hegedus-Garcia says of her agent’s first impression of the property. "At first they thought it was more of a fixer-upper."
However, the buyer's agent found that "the property was much better than the photos."
That doesn’t prove the home was definitively underpriced. But it does suggest the house may have been stronger than its online presentation—and stronger than the viral narrative—initially made it seem.
Five offers came in—and that’s where the seller may have missed the most
Even if the pricing debate is set aside entirely, Hegedus-Garcia argues that the bigger issue was what ChatGPT couldn’t do once buyers started competing.
“Let’s assume I am completely wrong on the price,” she says. “You did not capitalize on that multiple offer scenario.”
In Hegedus-Garcia’s view, that was the moment to move aggressively.
“At that point, you contact everyone that submitted an offer, and you say, 'I’m looking for your best and highest by this certain date,'” she says. The goal is not just to see who will pay more, but to force buyers to sharpen the full deal: price, timing, contingencies, and other terms.
That critique carries extra weight in Cooper City’s current market. Realtor.com senior economist Joel Berner says the area is in a buyer’s market, with inventory up 5.1% and median listing prices down 14.6% year over year in February, even as time on market has fallen.
“Buyers have taken notice,” he says. In that context, Berner argues, getting immediate competition was a rare source of leverage.
“The seller had a nightmare scenario where he listed far too low and claimed to have generated competing bids right away,” Berner says. “Not taking advantage of that situation to claw back some lost price could have cost him six figures.”
Negotiation didn’t stop at contract signing
It's a stinging possibility, but there's another cost to selling without an agent: navigating the deal once it is under contract.
According to Hegedus-Garcia, Levine repeatedly had questions about addenda, disclosure requirements, inspection timelines, and other contract terms as the transaction moved forward—questions ChatGPT could answer in broad terms, but not always in ways that matched Florida practice.
“He would run it by ChatGPT, come back to her and say, 'How about this?'” she says. “And [the buyer’s agent] is like, no, that doesn’t apply in Florida.”
Some of the confusion, she says, involved forms required under state law, along with the kind of live judgment calls that do not translate neatly across markets.
While Levine worked with a real estate attorney, he also leaned heavily on the expertise of the buyer's agent.
Hegedus-Garcia says it's “not common” for a seller to rely that heavily on a buyer’s agent. While an agent may help a FSBO seller move through the process, she says, “If you really separate fiduciary duties, you want to make sure that you have someone representing your whole side of it.”
ChatGPT can help with tasks. It cannot yet replace judgment.
Hegedus-Garcia is careful not to frame this as an anti-AI story.
“I am very pro artificial intelligence and tools,” she says, describing herself as deeply tech-driven. But, she adds, “There's times where artificial intelligence makes mistakes and that's a problem.”
Her analogy is simple: “I use ChatGPT as my nutritionist, but I go to the doctor and get real blood tests. I'm not going to trust ChatGPT with my health, with something that's really important.”
The same logic applies to real estate, she argues: Consumers should use AI to ask smarter questions and move more efficiently, but they should be careful not to mistake that help for judgment.
Dan Weisman, director of innovation strategy at the National Association of Realtors®, puts it similarly: “AI tools like ChatGPT can be useful for research, but they are not a substitute for professional representation," he says.
"A home purchase is likely the largest financial transaction of a person's life, and it involves legal contracts, local market expertise, negotiation, and fiduciary obligation," he adds.
By Levine’s own account, ChatGPT appears to have helped with the steps: It helped him organize information, move through the process, and get the house sold. But the unresolved question is whether it helped him sell as well as he could have.
On that point, Hegedus-Garcia estimates that the cost of going without representation likely landed somewhere between $75,000 and $225,000—a number she got by subtracting Levine’s claimed commission savings (about 3%, or roughly $28,000) from what she sees as the home’s potential pricing gap based on comparable sales.
That isn't to say every seller needs a traditional agent or that FSBO can’t work. There is, Hegedus-Garcia suggests, a place for all of it.
But this sale shows the risk of confusing independence with expertise. A seller may be able to skip representation and still get to closing, but that doesn't mean a chatbot can leverage your most valuable asset in a stagnant market.
Realtor.com has contacted Levine for comment.
