How AI Can Save Developers From Doomed Nine-Figure Investments

Steven Song, CEO of Diald AI, makes it clear that the massive advantage of using artificial intelligence to evaluate real estate deals is to help uncover key data that otherwise might have gone unseen.

“We’ve all had deals that made perfect sense on paper, where everybody agreed it would work — until it didn’t,” Song said during “Winning the Next Deal: How AI Is Reshaping Investment Strategies in CRE,” a virtual event hosted by Commercial Observer Partner Insights on June 12 and presented by Diald AI. 

“One of the most dangerous things in real estate today is that we’re all working from the same playbook: the same rent tables, the same data vendors, and the same tired assumptions,” said Song. “So the firm with the most aggressive pencil often wins because the data isn’t giving anyone a real edge.”

This leads to the horrifying possibility, said Song, that $100 million dollar decisions and higher are being made based on incomplete information.

“What if we’ve all become hostages to the same dubious data tables?” said Song. “What if we’re making multimillion-dollar decisions based on cleaned-up PDFs that nobody questions?” 

The experiences of his fellow panelists — Diald AI COO Eduardo Foss, Way Capital founder and CEO Malcolm Davies, and Max Collaborative co-founder & Managing Partner Kevin Ratner — bore out how projects in real estate often progress quite differently than expected. 

“A lot of times, the problem is that projects take so long,” said Ratner. “We did a project in Oakland in the early 2000s of close to 600 units. It took 36 months to build, and during that time the economy changed, so the lease-up went slower, and the rents weren’t where we thought they would be, which we didn’t have room in the deal for. That’s one of the big issues around real estate, that it just takes so long.”

Davies concurred, saying that the inability to obtain needed data in a timely manner is a perpetual problem for development projects.

“I can tell you a number of development deals throughout my 25-year career where we just couldn’t get the right data to show why somebody should do a deal or not,” said Davies. “Data that’s able to stop a project and save $100 million could be a really appropriate use of AI going forward.”

Asked how he believes AI might have assisted on his Oakland project, Ratner laid out several possibilities. 

“Maybe with more information we could have built more of a cushion, because we would have understood the market better,” said Ratner. “Maybe we could have added or subtracted some things from the building to either save cost or make it more competitive. AI can crunch the data and say, ‘It looks like this is going to be overbuilt,’ or, ‘The population is changing,’ or whatever. The better you’re able to understand the data, the better off you’ll be.”

Davies emphasizes the importance of AI in helping remove human bias from the decision-making around development projects. 

“Think about how many human decision-makers we use to actually get deals done,” said Davies. “That goes for everything from the company doing a feasibility report that takes two months, to the appraiser, to the investment committee. At the end of the day, we all have biases. I can’t tell you the number of deals we get done because someone says, ‘You know what? I like that area,’ or, ‘My friend said this was kind of cool.’ The data points we’re using are really irrational at the end of the day.”

For the event’s second half, Foss presented several case studies compiled by Diald AI that drive home how AI can make crucial contributions to decision-making regarding real estate investments.

Discussing a Chicago hotel that was ultimately foreclosed on, Foss outlined a Diald AI post-mortem analysis of the property showing how if the developers had used AI to analyze the surrounding area, they would have seen in advance why the location was suboptimal for their investment.

“The developers did not have the reality of what was happening on the ground,” said Foss, who then showed a detailed report, 100 percent compiled and written by AI, outlining the many problems in the area with rising crime and declining retail outlets.

“With our tools at Diald AI, we were able to track what happened in that location,” said Foss. “We were able to find instances of increasing crime less than half a mile from the location, including cases of violent crime and carjackings. We were able to find changes in the public sentiment through AI. AI is able to look at what’s happening in the neighborhood, get news about the political situation, the criminal situation, public works, etc., and present it to the operator.”

The Diald AI report also completed a comprehensive analysis of the local retail situation, uncovering issues with local malls and retail tenants. 

“All those problems could have been caught back then, and would have pointed to a bad investment,” said Foss. “Here at Diald AI, we look at all the qualitative aspects and we create a score. This site would have scored 57 (out of 100), which means that the site falls below the threshold of defaulted projects. Eighty percent of projects that fall below a Diald score of 60 fail.”

Foss showed a number of similar examples. For a Korean investor considering the purchase of a mixed-use San Francisco high-rise, Diald AI compiled a detailed 30-page report outlining the building’s strengths and weaknesses as an investment. While the report would have taken human analysts weeks to compile, Diald AI completed it in a matter of hours.

In the end, Daild AI uncovered significant issues around the area, including rising crime and declining transit accessibility, as well as questions about the stability of the building’s long-term tenants, and gave the property a score of 45.

Foss showed how Diald AI’s analyses are impeccably thorough, capturing data on a slew of essential factors and executing high-level analyses based on them. 

For the analysis of a 210-unit multifamily property in Washington, D.C., for example, in addition to information about the construction of the building and the area, Diald AI gave a detailed analysis of the exit valuation outlook, comparing the property’s cap rate and exit price per unit with other properties. 

“This report describes in detail how the cap rates should be adjusted for the reality of this unique building,” said Foss. 

The report also included a detailed conclusion, outlining the pros and cons of every aspect of the investment, and it was all compiled by Diald AI in around five hours.