How artificial intelligence is changing real estate investment

AI startups raised $18.74 billion in venture funding between 2012 and 2017

Illustration by Daniel Nyari
Illustration by Daniel Nyari

Welcome to the world of AI.

Last March, the online asset manager AlphaFlow launched its first so-called automated investment fund for real estate loans.

Rather than tapping employees to search for investment opportunities, the company uses a computer program to sift through listings services and find mortgages to purchase based on property characteristics, borrower history, recent market data and other criteria.

“We’re not ready to take our hands off [our work],” said Ray Sturm, co-founder of the California-based company, explaining that humans still make the ultimate investment decision.

But, he noted, machine-driven algorithms are doing increasingly more. And a growing cohort of AI-powered companies — including AlphaFlow, Ten-X and Opendoor — is starting to transform real estate investment and brokerage much like passive investment funds revolutionized the stock market in the 1980s.

In its most basic form, AI has been around for decades. But the technology has gone through a recent growth spurt, most notably with machines now more capable of “learning,” meaning they can perform tasks without being programmed to do so. And between 2012 and 2017 alone, AI startups raised $18.74 billion in venture funding, according to research firm PitchBook.

Ten-X, which has raised $141.8 million in funding, per Crunchbase, and was acquired by private equity firm Thomas H. Lee Partners last year, created its own algorithm to match investors with properties, for instance. The system analyzes investment sales data to determine which properties certain investors tend to buy. It then picks buildings that match their criteria on its site and recommends them.

Ten-X’s Neelika Choudhury, vice president of product strategy and product marketing, said the algorithm creates better, faster and cheaper leads than humans could. But it requires regular maintenance to keep up with investor behavior.

“If we don’t constantly refine our training data, the model won’t work,” Choudhury said.

Meanwhile, California-based Opendoor, which was founded in 2014 and has raised a massive $320 million, buys single-family homes with the help of machine-learning algorithms that identify investment opportunities and determine the price (although sellers generally approach Opendoor first). It then puts those properties up for sale online.

Commercial and rental lease management firms are also turning to AI. Startups like Leverton and Counselytics sell software that sifts through lease contracts, extracts information and even makes decisions.

Sign Up for the undefined Newsletter

By signing up, you agree to TheRealDeal Terms of Use and acknowledge the data practices in our Privacy Policy.

“It saves an immense amount of time,” said Zach Aarons, an executive at the real estate investment firm Millennium Partners and co-founder of the New York-based tech accelerator and advisory firm MetaProp NYC.

But even Aarons seems to think lease abstraction may not be ready for prime time — at least in the complex New York market. Millennium considered using it, he said, but ultimately decided against it, partly because it was concerned about errors.

“Some of our leases are so complicated we’re not even sure humans would get it right the first time,” he said.

And while AI has yet to infiltrate the construction sector, sources say it’s only a matter of time.

Turner’s Barrett said AI could revolutionize construction by automating the design process, cost calculations and scheduling. While he declined to offer specifics, Barrett said Turner is researching machine-learning technology.

The applications for it in construction abound.

Ardalan Khosrowpour, a New York-based entrepreneur, recently launched OnSiteIQ, which creates 360-degree images of construction sites to track progress and potential hazards. The startup is working on machine learning software that will automatically analyze the images to spot potential safety hazards or other problems.

Khosrowpour said he’s now in talks with several major development firms, which he declined to name. “Machine learning will be the future,” he said. “I have no doubt about it.”

Check out the complete version of this cover story in the February 2018 issue