The AI Anywhere Real Estate uses to boost leads and listings

Gartner describes AI as a ‘beacon of innovation’ that companies across all industries are leveraging to save money and increase productivity. But this doesn’t mean AI can do everything, cautions Rudy Wolfs, CTO at Anywhere Real Estate. “While no one can deny the potential and promise of AI, we must be aware of the immaturity of this technology today,” he says. “I spend a lot of my energy calibrating our team’s view of AI, and making sure they understand where it’s strong, where it’s still developing, and where they may be disappointed by what it can and can’t do.”

He does so because he doesn’t want users and business leaders to try AI, fail to get desired outcomes, and then lose faith in it. For Wolfs, it’s important to set the right expectations, use it in ways that best showcase its potential, and give users confidence it can deliver. This is what Wolfs and his team are doing with Anywhere Real Estate’s Listing Concierge, as well as their lead scoring efforts.

AI to enable better listings

According to Wolfs, Anywhere Real Estate’s Listing Concierge offering gives agents a highly intuitive toolset to promote listings. The goal is to make the real estate experience seamless for buyers, sellers, and agents. “We have thousands of agents, listing thousands of properties,” he says, “so anything they can do to speed up the process of putting together and posting a listing is a big win for the business.”

Listing Concierge composes the initial listing description based on photos taken of the property, and automatically tags images of specific rooms so buyers know what they’re looking at. “Depending on where you place the listing, you might have different character limits,” he says. “Listing Concierge will automatically adjust how much is written about the property so you still have a description that captures the essence of the property without having to sit and rewrite anything.”

This gen AI solution is also trained to ensure everything’s done in a responsible and ethical way. While this wasn’t a major issue before, Wolfs confesses, it’s just one more thing agents don’t have to worry about anymore.

Leading by example

Anywhere Real Estate has also deployed gen AI to power their leads engine, which analyzes the information shared by the client and then allocates the lead to the best agent. As the largest consumer real estate business in the country, they’re inundated with leads, says Wolfs. “In the past, we’d simply allocate them to agents using our judgment around which agent would be most suited to this particular lead, and who’s best positioned to close a deal in this particular market,” he says. The lead scoring engine allows them to deliver the right leads the right people. “This is a win for the consumer, because they get an agent who’s well positioned to meet their needs, and a win for our agents because they’re matched with the business opportunities that suit them best.” This also wasn’t a major issue before, but he says they weren’t really able to learn anything from their allocation hits and misses in the past. “By applying AI, we can better understand consumer comments and allocations,” he adds. “This allows us to identify patterns we may have missed.”

For example, using AI, they can analyze the types of words or phrases a customer uses to get a sense of their level of urgency: are they looking to sell their home in weeks, months, or years. In instances where the seller is looking to find a buyer quickly, this urgency can be easily conveyed to the agent, which will influence how they approach the lead.

When undertaking projects like this, it’s difficult to know which model or vendor is going to be the right one, Wolfs says. This is why he and his team built what he describes as a ‘harness’ that allows them to connect into a variety of AI models, and easily replace one with another so they aren’t locked into any particular model at any point. “This very thin technology layer was probably one of the best things we did architecturally because it allows us to experiment with a certain model, and then switch things out if another provider has something better suited to what we’re trying to do.”

This was critical across both projects. Beyond making sure they don’t lose faith in AI if it doesn’t work as they intended, building this AI fluency early on was also important so people trust the data and the results from the model. Here, he stresses the value of an iterative approach by running regular, small tests to identify where it succeeds, where it fails, and where there might be opportunities to use it elsewhere.