Two homeowners used AI to sell their properties for over asking price within days, pocketing tens of thousands in saved commissions. One cleared $90,000 on a Hudson Valley sale. The other sold a Florida home for nearly $1 million - $100,000 above agent estimates. Their experiments show what happens when sellers replace traditional agents with tools like Google's Gemini and ChatGPT, and the results have sparked a direct challenge to the standard 3% commission model.
Stuart Thompson, a tech reporter for the New York Times, and Robert Levine, a CEO who teaches firms how to use AI, both bypassed real estate agents. Each used an AI assistant to handle listing language, staging advice, and negotiation strategy. They paid lawyers for paperwork instead. The savings were immediate and large.
$90,000 cleared and $36,000 in fees avoided
Thompson and his wife initially estimated their Hudson Valley home might fetch roughly $550,000 based on Zillow data and conversations with a few agents. He disliked the prospect of paying a 3% commission - up to $30,000 or more. He turned to Google's Gemini to write the listing, find a photographer, and guide him through posting to the MLS.
Within 24 hours, the couple had booked a weekend of showings. Three offers arrived, all above asking. Gemini advised on which offer to accept. It was not the highest bid, but it was a sure thing at over $600,000. Thompson paid a lawyer to handle the paperwork. Even after that cost, he and his wife cleared $90,000 on the sale. The buyers also agreed to cover their own agent's 2% commission, saving Thompson an additional $36,000 in fees.
Levine's experience in Cooper City, Florida, followed a similar arc. What began as casual ChatGPT prompts turned into a full consultation. The AI advised on staging and timing. The home drew 15 viewings and five offers, selling for nearly $1 million in five days. That was $100,000 more than agents had told him to expect. He saved $28,000 on commissions.
Where AI agents fall short
Both sellers acknowledged gaps. AI cannot show a home to buyers, move furniture, paint walls, or call buyers' agents. That labor falls entirely on the seller. One reader of Thompson's New York Times account questioned the time savings, pointing out the legwork of calling and staging. Thompson also said he missed the emotional support a human agent provides during a stressful sale.
Errors crept in. Gemini initially gave Thompson incorrect advice on paying the buyer's agent commission. Ines Hegedus-Garcia, managing partner of Avanti Way East Miami, the brokerage where Levine's buyer's agent works, told Realtor.com that Levine could have earned more with professional representation. She said his reliance on ChatGPT during negotiations exposed legal blind spots.
"He would run it by ChatGPT, come back to her and say, 'How about this?'" she said. "And [the buyer's agent] is like, no, that doesn't apply in Florida."
Still, Levine has no regrets. "I'd recommend it to everyone," he said. "ChatGPT is not coding. It is a conversation, and you're going to have to have that conversation with a real estate professional if you want to go that direction anyway."
Agents push back on the trend
Many real estate agents are not happy. Some question the true time savings. Others warn about legal risks, administrative mistakes, and leaving money on the table during negotiations. Thompson himself wrote that AI may transform real estate agents into something more like travel agents - once essential for navigating an opaque process, now a nice-to-have for busy people who want a more carefree experience.
For professionals watching this shift, the core tension is clear. AI handles listing language, pricing data, and staging checklists. It cannot replace the in-person judgment, negotiation nuance, and regulatory knowledge that a licensed agent brings. The question is how much those human skills are worth when sellers can pocket five figures by doing the legwork themselves.
Why this matters for real estate and construction professionals
The two case studies above are not outliers - they are early signals of a toolset that lets sellers keep more equity. For brokers, agents, and construction-adjacent professionals who earn fees tied to transaction value, AI-assisted DIY sales put direct pressure on the commission structure. Understanding how these tools handle listing strategy, pricing, and negotiation is no longer optional. Resources like the AI Learning Path for Real Estate Brokers and broader coverage of AI for Real Estate & Construction show what's already in play: property marketing automation, lead generation, and sales workflows that can reduce the seller's dependence on a full-service agent. The savings these two sellers captured are real, and the tools are only getting more capable.
Your membership also unlocks:
