How Realtors Are Really Using AI in the Age of "Real Estate Slop"

When Pasadena, California–based realtor Pilar Oliva was having trouble convincing a client to lower the price of a home they were selling, she asked Claude for help. 

The home in question was on a corner lot facing a somewhat busy street, and while Oliva had pulled a solid comp from a home down the street when they initially priced it for sale, the market had shifted and buyers weren’t biting. Meanwhile, the clients had found their new dream home in a neighborhood they liked better and were debating whether they should use a HomeLight bridge loan to purchase the property or lower the price on their home enough that they’d be able to quickly turn it around. The bridge loan required them to sell within 120 days and would put them on the hook for the full asking price, regardless of what they sold for, but dropping the price by almost $200,000—from $1.49 million to $1.299 million, the amount the market now dictated—felt like an insult.

Oliva decided to use Claude to build two financial models: One that would show what the homeowners would net or end up owing if they lowered the price and sold it themselves and one that showed what it could ultimately cost them if they opted for HomeLight. The client only wanted to spend $40,000 out of pocket either way, so Oliva asked the AI assistant to create manipulatable HTML models where she could fuss with various prices and offers. She toggled switches on and off, checked and double checked the math, and spent hours getting to a point where she felt she could trust the AI’s work. "It worked the hardest I ever worked," Oliva says, noting that all this went down while she was en route to a family wedding in France. "I was in my father-in-law’s car in the middle of the night, on my laptop, using the models to put together rebuttals and trying to convince them that they’d be making a wrong decision if they let HomeLight buy their property."

It worked, fortunately, and after a price reduction, the home ended up selling for $1.4 million, just $90,000 less than the original asking price. Oliva’s clients were happy, the new buyer was happy, and Oliva and her team had created a financial tool they could use again and again, should similar situations arise. "I’m a realtor, so I’m just using AI to fill in my gaps," Oliva says. "I’m not a writer and I don’t build financial models for a living, so I ask it to do what I can’t."

With 82 percent of U.S. real estate professionals now reportedly using AI in their everyday work, Oliva certainly isn’t alone. Ines Hegedus-Garcia, co-owner and managing partner of Avanti Way Realty in Miami, says her company is using AI "at every level, from workflows to analyzing data to helping agents come up with text that makes sense for their clients." Only some of the agents Dwell spoke to talked about struggling with the ethics of AI or the potential environmental impact; instead most seemed almost resigned to using it, if only to keep up with their competitors. "As a realtor, the choice is narrow," says Oliva. "The industry runs on dinosaur-era tech like the MLS and Zipforms, but we are simultaneously flooded with shiny new software. I can compete without AI but the time it saves me is invaluable. I hate the footprint it leaves, but the efficiency makes it impossible to look away."

"When an agent gets a listing, there’s [always been] a very specific process," Hegedus-Garcia says. "They get the listing signed. They have to order signs. They have to order marketing. They have to put the For Sale sign on the house. There’s a process that every agent follows. But if you can get AI involved, then you can create a system where, as soon as you get a listing, it automatically orders the sign or orders someone to put it on the property. That kind of automation just makes an agent more efficient." 

Realtors make their money by dealing with clients, Hegedus-Garcia says, and so the more busywork that can be freed up to let a realtor be there to hold a buyer’s hand during negotiations or show them additional properties, the better. And buyers do want that human touch, rather than feeling "like they’re being talked to by a robot," says Kati Spaniak, a realtor working in the Chicago suburbs. "Buyers want personalized attention, especially knowing that their agent is making thousands of dollars a transaction. I think there could be an issue soon where agents start realizing they’re using it too much and buyers start feeling less connected."

"You can say ‘Hey, ChatGPT, please don’t change the floorboards,’ and it’ll change the view out the window or put a cow in the corner."

Buyers frequently have difficulty imagining the potential a space could have; after all, that’s why real-life staging is a robust business of its own. Most of the real estate agents interviewed for this story said they’ve used AI design tools to help clients envision how their style might look in a home they were considering or to help identify architectural features in a listing, like a vintage tile’s potential maker or what you might call a certain kind of column. (Zillow, it’s worth noting, has already integrated on-site AI-driven listing enhancement features, including staging help.) Libby McKinney Tritschler, a realtor from Connecticut, described helping a client visualize what a deck might look like on a home they were considering, saying, "We went to the backyard, took a picture, and asked AI, ‘Can you show us some pictures of back porches and what kind of pricing we might be looking at?’ Three and a half minutes later, we had three different options with three different prices, all of which gave my buyers confidence in buying the house because they knew that, when they were ready, they could afford to make the upgrades."

On the flip side, sellers and realtors can go too far when it comes to altering listing photos, however, with more and more buyers bemoaning AI-staging of homes in listing photos as a deceptive practice in countless stories and online threads. (Online, it’s often called "housefishing" or decried as "AI real-estate slop.") New York City Mayor Zohran Mamdani has even moved to make listing sites and agents disclose when they’ve used digitally altered images, telling the press, "You shouldn’t have to worry about whether the apartment you are viewing online is real."

While there’s an argument to be made that virtual staging has been used for years in real estate listings—and that even physical staging is akin to making plain old cornflakes look a little better on a cereal box by throwing some strawberries on top for the photo—AI has certainly made tweaking listing photos a little easier, even if it does always add what licensed real estate salesperson Suzanne Goldklang calls "the same beige bedspread." Goldklang says it takes her about two hours to turn a single photo of a cluttered living room into something presentable using AI, though other realtors Dwell spoke to and elsewhere have claimed it can be done in minutes. "It’s because you’re having to do it through prompts," Goldklang says. "You can say ‘Hey, ChatGPT, please don’t change the floorboards,’ and it’ll change the view out the window or put a cow in the corner."

"Buying a home is emotional, and when buyers get to a home, one little thing can make them feel that it isn’t right," says Spaniak. Walking optimistically into an open house and realizing it’s been poorly or deceptively staged using AI is, as one agent said anonymously, "like the difference between watching porn and being with a real woman."

That doesn’t mean there aren’t companies out there trying to help realtors modify their photos in a more seamless way. AutoReel, for instance, lets realtors upload up to 20 photos of their listing, which it then blends into a video tour of the home. Alok Gupta, the company’s cofounder and CEO, says that over 75,000 real estate professionals around the world have used AutoReel’s service to date, making more than 200,000 listing videos.

"My staff was like, ‘it looks like AI,’ but I said, ‘I’m not a newscaster, so is it cheating?’" 

AutoReel can turn a photographer’s drone photos into a home flyover, restage a room in a more visually pleasing way, or even remove clutter from a shot, too. Users can digitally watermark everything they make to note it’s been created with AI if they’d like, but most real estate sites don’t require it. (There’s no national MLS standard for disclosure of AI-generated or enhanced imagery, though some states like California and Wisconsin passed new laws, and other states like Texas and New York have general advisories or MLS rules. Meanwhile, a Zillow blog post about the company’s AI-enhanced listing standards says it "supports clear disclosure when AI is used to materially alter a listing image" but also that "agents and brokers carry front-line responsibility for accuracy.") AutoReel accounts can be shared across real estate teams, meaning that a small group of agents could divvy up access to something like the company’s $90 per month "Growth" plan that includes 100 videos a year. "You can create a video in five or ten minutes for less than it costs to go to Chipotle," Gupta says. "With traditional video, the cost for one could be anywhere from $700 to $1,000." Some real estate photographers are also opting to add AutoReel to their sales packages, perhaps adding $90 to an overall cost but providing a video that can be used everywhere from Zillow to social media.

Some real estate agents are using AI not just to market their homes but also their local stomping grounds. Patrick Southern, a realtor with Serhant in New Jersey, says he’s used an AI service called Rayloom to create a virtual newscaster version of himself, which he’s dubbed "Pat AI." Southern uses his AI avatar to relay positive news stories about his region on his Instagram page, like "Hoboken waterfront restaurant named one of best in country," or "Loew’s theater coming back to life this fall." The talking heads look and sound like other videos Southern has posted of himself, but they’re relatively low cost to make ($699 for a dozen per month) and average between 100,000 and 200,000 views each. "When I got pitched on it [by Rayloom’s owner], I was like, I think it’s a good idea but I’m not sure," Southern says. "My staff was like, ‘it looks like AI,’ but I said, ‘I’m not a newscaster, so is it cheating?’ How is this any different than giving potential clients a really great email blast?"

Southern says it’s all in service of bettering the client experience, noting that he always adds a disclaimer in the caption when a video he posts uses his AI-generated faux-likeness. "AI isn’t the superhero," he says. "It’s really the utility belt. It’s all what you make of it."

Marah Eakin is currently reporting a follow-up piece on how AI is changing the homebuying process for prospective owners and sellers. Do you have an experience to share? Reach out.

Related Reading:

Does Going Viral on Zillow Gone Wild Help or Hurt a Home Sale?

Biohacking Has Come for Our Homes