The Robots Are Coming, and Other AI-Related Observations From the Experts

Isaac Asimov saw this coming.

Commercial Observer’s Power Tech & AI Forum and Reception at Marx Realty’s 10 Grand Central on Oct. 19 featured in-depth discussions about the latest in AI from real estate technology insiders. A little bit into it, Abhinav “Abe” Somani, founder of Cadastral and head of real estate for Legora, answered a question about what’s coming for AI in 2027 with an answer right out of “Blade Runner.”

“For real estate in particular, the big shock is going to be the humanoid robots that will start getting released next year,” Somani said casually. “That is really going to change everything we’ve done. From 2D text-based AI to 3D vision, voice and sensory details, this has more long-term implications for real estate than producing a legal bill.”

The event was emceed by Commercial Observer Deputy Editor Tom Acitelli, who greeted attendees — including honorees on CO’s annual power tech list — with a quick but powerful reminder of the import and impact of the current artificial intelligence revolution.

“In my tenure covering real estate, I’ve watched dozens of technologies arrive on the scene and promise to shake up the business,” said Acitelli. “Some have done just that. Others, like the Metaverse, never found their footing. But I don’t think any have brought the full scope of upheaval and disruption that AI has.”

Continuing on this theme, the evening’s first panel, titled Empowering Deal Acceleration: Inside The Tech-Enabled Transactions Shaping Today’s Market, included comments from Somani; Zach Aarons, co-founder and general partner at MetaProp; Nick Durst, principal and director of operations at the Durst Organization; Joseph Sipala, managing director at JLL; and moderator Cher Chang, founder and CEO of Literock Capital and partner and head of capital markets at JLJ Capital.

Chang kicked off the panel with a reminder of how quickly change is happening in the AI space.

“Three years ago, I did an AI panel for the first time. At the time, there was a lot of excitement, no applications and a lot of predictions,” said Chang. “Today, it’s a completely different ballgame.”

In response to Chang asking about how AI is accelerating their real estate transactions, Somani noted AI’s current base layer for commercial real estate application: the ability to parse through reams of documents.

“When we think about practical uses of AI, you can make it through transactions much faster by expediting reviews of documents and negotiations,” said Somani. “When we think about giving practical advice, we say, go for the lowest hanging fruit possible. Today, what we’re seeing is a tremendous amount of adoption of legal AI technologies, both to law firms and in-house counsel, to basically accelerate real estate transactions.”

Aarons noted that while there has been uncertainty concerning the accuracy of AI, advances in the technology are beginning to address this.

“No matter how cool AI was two years ago, it was worthless to [most people] because you questioned the accuracy of the output,” said Aarons. “So one area we’ve been heavily focused on, and invested a tremendous amount of our fund’s capital in, is the preconstruction process as it relates to AI.”

Aarons said blending computer vision algorithms with new generative technologies is allowing for agents that can take “millions of pages of construction documentation, spec sheets and more,” and identify errors, discrepancies and other potential problems. 

On that note, he said that one of MetaProp’s portfolio companies, Handoff, handles material takeoffs for residential projects with 97 percent accuracy.

“We backtested it,” said Aarons. “A team of humans doing the same work performs at only 94 percent accuracy, and Handoff can do the same task 20 to 50 times faster.”

Sipala noted that turnaround times have accelerated for many essential tasks at JLL due to this technology.

“Because of tools like Legora, Claude and others that we’ve built into our JLL platform, we can turn around portfolio analyses and single lease reviews in a quarter of the time,” said Sipala. “What took a week’s effort now takes a day. This breakthrough has become table stakes in our business — an expectation, not a differentiator.”

For owners, Durst noted that AI is also streamlining the process of acquiring development sites.

“We can load a data room into a program and be able to understand the ins and outs of a property in minutes,” said Durst.

Somani predicted that the next 12 to 24 months will see a shift among companies from using off-the-shelf AI tools to creating more specialized and custom applications.

“Companies are going to ask, ‘How can I give our AI tools some of our institutional knowledge and some of the things that make us special so that we can have an edge in the market?’ versus just having AIs go back and forth between each other,” said Somani.

Aarons expects a more specialized variety of applications as well.

“You’re going to see people in our industry getting more comfortable with using different models for different parts of their workflow,” said Aarons, who said this transition will be about cutting costs, and optimizing the most cost-effective tool for every task instead of just broadly adopting any particular tool.

The second panel, titled Working Smarter: AI’s Measurable Impact Inside Real Estate’s Leading Firms, featured Canay Deniz, CEO and co-founder of Ren Systems; Mark Domino, director of digital for the Durst Organization; Benjamin Levine, vice chair and principal at the Douglaston Companies; Mark Stutzman, global chief technology officer at Fisher Brothers and CTO at the company’s Area15 in Las Vegas; and moderator Don Laughlin, director of technology at the Real Estate Board of New York (REBNY).

The panel started with a consideration of whether AI is killing jobs as it increases efficiency. The panelists indicated that, to date, commercial real estate has mostly avoided the AI-led job attrition seen in other industries.

“AI is definitely not taking jobs,” said Stutzman. “The goal of AI at Fisher Brothers is to make staff more efficient, and to take away the bulk of the repetitive tasks.”

Domino, though, said that while AI may not be impacting current employees, it is likely to shrink future headcount.

“AI may not take your job, but it will probably impact whether or not we hire somebody to assist you or that budget goes to AI spend,” said Domino.

Stutzman said that at Fisher Brothers, AI has replaced around 60 percent of the software-as-a-service tools the company had been using, moving the firm into a “more of an agentic orchestration.” And, as quickly as AI technology is progressing, expectations for the technology are keeping pace.

“The conversation has shifted very quickly from how do we get people to adopt AI to how do we make AI produce outcome that we can measure,” said Deniz. 

Part of that, according to Deniz, is to implement these tools step by step as your team members learn their capabilities, rather than trying to impose top-down solutions that might not be optimal for certain tasks.

“Let the people producing the outcome bring you what works for them before you issue a mandate,” said Deniz. “Otherwise, you’re rewarding tokenizing rather than producing outcome.”

Domino agreed that a user-first approach could help companies develop effective strategies for AI.

“You can see what people are doing with AI, and who the power users are,” said Domino. “Finding your key users, who are either going to save you a lot of money with it or drive a lot of revenue with it, is a key thing to identify.”

Laughlin noted that REBNY has deployed a member services agent on its website that handles “tier 1 resolutions and tier 2 escalations” for its members.

“We’ve gone through the workflows to make sure it’s answering the questions and then routing them to the right person,” said Laughlin. “We’re measuring it, and we’re building on it.”

That said, caution in adoption and usage is still advised due to the incredible speed with which AI is evolving, and the protections that are necessary but not automatic.  

“Our CAO came to us a few months ago begging for help around governance for all of this,” said Levine. “A lot of these models and implementations are wonderful. But, until we’re really comfortable with our data security … the tools aren’t really there yet to get people as excited as they need to be. So it’s a balance. We’ve built seven or eight AI-related projects, our heads of finance want us to automate our investor reporting and our annual reports, but the problem is that governance, data safety and the infrastructure around it all has gotten complicated. 

“It’s all happening way faster than we expected.”

Larry Getlen can be reached at lgetlen@commercialobserver.com.