What Are Real Estate Marketplaces Doing with AI?

The two letters that found their way into a great many Linkedin bios last year also very much made it into the real estate marketplace industry discourse.

AI dominated the conversation at the Property Portal Watch conferences in 2023 and will no doubt be top of mind for attendees at this year's Bangkok event in March. Throughout 2023 we saw no fewer than 35 announcements from real estate marketplace companies trumpeting their use of AI.

The industry seems to be in the early days of working out what it can do with AI and, crucially, what it should bother doing. The uses we've heard about are broad but can be roughly categorized into three groups:

• Discovery - changing how users search for or are shown listings
• Copiloting - helping either the user or the agent perform a task
• Listing enhancement - improving the quality of the data the marketplace has for its properties

Discovery

It seems that changing the way their users find what they're looking for is the category of AI use that real estate marketplace companies are most keen on. Or at least it's the type they're willing to tell the rest of the industry about via press releases and social media posts.

Some marketplaces have used artificial intelligence to build natural language search applications. The idea here is that house hunters can input complicated long queries and the portal will match the unstructured query to the structured data it has for its listings.

While the U.S. digital brokerage site Doss was pioneering this form of search several years ago, other companies started getting in on the act in 2023 with Zillow and rentals specialist Zumper building property search plugins for the popular ChatGPT interface.

However, almost as soon as these later AI applications were announced they were disabled by ChatGPT's owner OpenAI due to concerns about contravening fair housing discrimination laws.

While there is scepticism in the industry as to whether spoken queries will ultimately catch on in real estate search, some such as UK-based portal OnTheMarket have started experimenting with the technology.

The other type of use case for AI in real estate discovery is what proponents like to refer to as 'intelligent matching'. This involves gathering or inferring information about user preferences and then matching that information to items in an index.

The likes of Southeast Asian portal operator PropertyGuru and Australian company REA Group have been using this method to show their users more personalised properties in their 'suggested' listings sections.

It's not just about helping house hunters find their dream home though. REA Group is also using user input to suggest the right agent for vendors and the European iBuyer-turned-agency Casavo is using AI-powered matching to help sellers select the perfect buyer from their database.

Ultimately the AI behind these intelligent matching applications is useless without rich user input. Portals will need to get good at retaining users, making sure they're logged in and striking the balance between imposing questions and inferring preferences from behaviour.

Copiloting

Another term that has been bandied about when talking about AI applications is 'copiloting' — helping someone to perform a task more efficiently.

One way to use AI to help copilot stakeholders is through chatbots. A few years ago most humans hated speaking to bots, but the advent of ChatGPT seems to be changing that and several real estate marketplace companies have built some form of AI-powered chat interface. For example...

• Brokerage firm eXp has built an application called 'Luna' that helps its agents with day-to-day tasks.
• Kiwi portal Realestate.co.nz has built a copilot that alerts agents when they have failed to upload any data related to a new listing.
• Bayut has made a chatbot that sits on its portal and helps users research property in the UAE.
• Singapore-based end-to-end marketplace Ohmyhome has built a bot named 'HomerGPT' that answers questions about specific properties.

What these applications have in common is that they connect one stakeholder in the property value chain to another or at least to information.

There might be another level that we see AI chat experiences aspire to in 2024 and beyond though.

As Simon Bray, President of Canadian real estate portal company REW, told attendees of Property Portal Watch Madrid, the ultimate use of AI-powered chat might bring in more than two stakeholders:

"A real true chat chat experience between three parties: the buyer the agent that's helping them buy and the platform itself that's informing them building their confidence giving them options and selection that could be a generative AI-powered experience."

AI copiloting is probably a lot more common among real estate marketplace companies than we know about. After all, if you're only using AI internally for everyday efficiency tasks, why bother telling the world about it?

We do know that international marketplace operator Prosus has released an internal AI tool that helps employees across their portfolio companies with things like the compilation of reports, translating documents and explaining and modifying codes. Surely most big marketplace operators have something similar either in production or in development.

Listing Enhancement

The most compelling use of AI on real estate marketplaces, at least as far as their users are concerned, is the improvement and enhancement of the property listings themselves.

Portals have been using machine learning for many years to do things like improve their results algorithms and scan listing descriptions for taggable features. In 2023 they started to branch out...

We covered marketplace operators like Iedalista, OnTheMarket and Rentsync who used AI to help agents generate better listings descriptions. We saw the likes of Zillow and Japanese portal company Lifull begin to use AI to extract data from floorplans and even generate their own.

Several portals including Anjuke in China have experimented with using virtual staging to allow potential buyers to view properties with different theoretical designs and decor. We'll see marketplace operators go further with generative AI in 2024.

AI on portals is in the experimental phase but won't be for long

The industry is still in an exciting experimental phase. Some are revealing the specifics of user-facing AI applications they've built and others are still just seeing what they can do.

"We've already identified over one hundred opportunities where AI can support our plans" - Tarah Lourens, Rightmove COO at a recent capital markets day.

AI isn't going to magically bring about new revenue streams or completely remove big steps from the home buying and selling process overnight. It will make the existing steps easier and whoever controls the tech to make those steps easier will take a bigger slice of the pie.

You'd have to think that portals are the stakeholders that are best placed to win the adoption battle and control those steps. According to pan-European portal operator AVIV Group, only 27% of agents are currently using AI. That figure seems low.

Portals, platforms and brokerages are putting AI-powered applications out there under the guise of helping agents save time. But whether it's intelligent matching showing consumers the best properties for their needs, chatbots answering their questions or portal backends writing listing descriptions, it seems clear that AI means less work for agents.

“We've managed to now work out how to book a holiday without going to an agency and having someone hold our hands. We take money out of our bank accounts without going and asking somebody to pass it through a window. So of course at some stage we're going to feel comfortable trusting a platform to enable those transactions to happen.” - Industry expert Malcolm Myers at Property Portal Watch Madrid 2023.