AI’s Expanding Influence in Investment Sales

Despite some risks, it's giving commercial real estate brokers more time to be human.

Just a few short years after artificial intellimoargence entered discussions about the future of real estate investing—and everything else—the impact of AI is appearing throughout the investment process.

“To a large extent, changes from AI are felt most at the employee and broker level,” said Hessam Nadji, president & CEO of Marcus & Millichap. “Scaling at the enterprise level is more difficult than applying AI on a smaller scale.”

Nadji and other industry insiders find AI most effective for property research and financial analysis, though they warn that using AI requires training and human judgment.

“The true benefit of AI comes from aggregating data with both enough depth and breadth to identify patterns and trends,” Nadji said. “But it’s a major undertaking to go back enough years to extract historical data.”

Human hand and robotic hand
Image by InfiniteFlow/Adobe Stock

In CRE, like any other business, brokers need to meet their customers where they are. Today, investors are using AI.

“People are experimenting with AI all the time now, but the danger is that AI can sound very confident even when the answers it provides are not always right,” said Doug Carpenter, director & head of data strategy and AI at CenterSquare Investment Management. “We’re coaching people internally to work better with AI because, if you want higher quality output, you need to give it more content and the right prompts to work with.”

CPE Capital Markets Newsletter

Executive Editor Therese Fitzgerald delivers the capital markets intel that moves the needle.

Carpenter said CenterSquare’s focus is on how to make more information and analysis accessible to their staff to turn it into a competitive edge.

About two-thirds (66 percent) of commercial real estate professionals use AI weekly or daily, according to a study by First American Data & Analytics and DealGround, but just 5 percent trust it enough to inform real deal decisions. More than half (53 percent) said they use AI for support only and exclude it from final decision-making.

For Julie Baird, president of First American Exchange Co., a 1031 Exchange qualified intermediary, AI is useful for summary documents and to research market conditions.

“AI arms consumers with more information, but sometimes it’s the wrong information,” Baird said. “In the 1031 Exchange world, there are a lot of nuances in terms of taxes, timing and related parties. Human judgment and critical thinking can’t be replaced.”

Best uses for AI

Hessam Nadji
It’s the interpretation of the data—not the tools—that sets firms arpart, said Hessam Nadji, president & CEO of Marcus & Millichap

While AI can sometimes produce slick marketing material, most CRE brokers use AI to quickly analyze data, synthesize information, uncover opportunities and automate documentation. Summarizing potential deals with key data, including market and submarket trends, can be faster with AI tools as long as they’re used by knowledgeable brokers.

“Most institutional acquisitions pros have a deal vetting AI tool that can be used to screen deals according to certain parameters, such as properties built in the last five years with high occupancy rates in certain markets where future supply will be low,” said Erik Edeen, principal & senior director of U.S. investment sales for Avison Young. “You set rules and, as you use the tool, it can get an idea of your preferences.”

Brokers then eliminate prospects or dig deeper to find out more, he said.

“AI is also good at noticing softer things,” Carpenter said. “For example, AI may flag that out of the 10 properties it found, three are near a major highway, so you may be interested in them based on past purchases.”

AI can be a great research tool, but communication with it is key, and so is access to data sources you trust, he said.

“AI can level the playing field on the number of deals that brokers can review,” Edeen said. “We’re seeing greater AI adoption across smaller investors because they’re more constrained by manpower issues. AI can make them more nimble.”

Many commercial real estate brokerages use a proprietary advanced enterprise version of common AI tools such as ChatGPT, Claude or Copilot, according to Edeen.

AI can speed the process of reviewing prospective buyers since it can be used to quickly analyze previous purchases, lender partners and their portfolio from various sources, according to Edeen. That research, which helps brokers and owners determine which offers to consider, could previously take an analyst a week. Still, Edeen doesn’t move forward without speaking directly to decisionmakers about their actual intent for a deal.

Communicating with customers who use AI

AI has made fraud schemes way more complex, so it can be harder to know if anyone in any part of the transaction is who they say they are, according to Julie Baird, President of First American Exchange Co.

Over the past year or so, Matt Lipson, senior vice president of Northmarq West Region & co-founder of its National Restaurant Group, started to notice a change in prospective buyers.

“Most of the buyers of the restaurant spaces we’re selling are doctors, lawyers or accountants—not restaurateurs or real estate experts,” Lipson explained. “All of a sudden buyers started asking to see the last three years of sales or for detailed franchise agreements. These were unsophisticated parties suddenly asking sophisticated questions, so, clearly, they were using AI.”

Unfortunately, AI sometimes fixates on negative information and investors using it can be misled, according to Lipson.

“We found some properties were unsaleable because of negative information potential buyers were finding through AI,” he said. “We started figuring out what AI didn’t like and fixed it, but sometimes AI responses to prompts would continue to fixate on things and bash certain companies.”

In response, Lipson doubled down with AI investments and uses it to focus on replacing tenants in distressed assets.

“We developed a back end AI tool to helps us model out which kind of tenants are right for each one of these distressed buildings and to model out the entire investment, which is very helpful for developers,” Lipson said. “We tell them which are the mostly likely top 10 tenants and why, plus what they can expect in rents and cap rates.”

Competing with AI Assistance

Companies with the best AI tools have a competitive advantage, Carpenter believes, primarily because speedy analysis means they can produce investment memos faster and decide to act more quickly.

“A good data platform can provide trend data and update it versus data from a single point in time,” he said.

AI sometimes fixates on negative information and investors using it can be misled, according to Matt Lipson, Senior Vice President of Northmarq West Region & Co-Founder of its National Restaurant Group.

While Lipson uses AI to model investment opportunities within 45 minutes, compared with a week or more in the past, he doesn’t see AI itself as a differentiator in deals.

“Real estate will always be a human relationship business,” he said. “There’s so much nuance in real estate, and AI can only produce textbook insights—not human ones.”

Still, using data to identify potential deals “definitely gives us a leg up and supercharges the investment process,” Lipton said. “Now we can bring in 50 opportunities a month instead of two.”

Human interpretation and incorporation of AI-generated insights into the investment process is what sets brokers and companies apart—not the AI tool itself, Edeen said.

“Everyone is running their own AI, so when you walk into a presentation, you need to do a human reality check,” Nadji said. “Sellers may have used AI to generate a valuation that’s not what we see, so our agents need to do more work and exchange notes on data sources.”

AI Risks

The biggest concern investors share about AI is the danger of hallucinations or wrong information. While most AI users said the technology is improving and hallucinations are less common with the most recent iterations, systems are also in place to catch errors.

“Understanding where the data is coming from that AI is using is critical,” Baird said. “AI does a pretty good job of citing sources but having judgment and the experience to know what the data sources are and whether they’re reliable is important.”

The best use of AI is to take the extra hours it gives you from data analysis and automation to connect directly with people, said Eric Edeen, Principal & Senior Director of U.S. investment sales for Avison Young.

Another way to combat bad data that could impact a transaction is to work with an AI platform that makes it easy to request those sources and to check the math on the numbers provided, Carpenter said.

“If you outsource your own knowledge and put all your trust in AI, you have a lot to lose,” Edeen said. “Even as hallucinations have been reduced, you still need human discretion and relationships.”

Without human interaction, it can be difficult to trust buyers and sellers.

“There’s a phantom AI company that automatically generated a number of offer letters well below the actual value of the properties,” Edeen said. “Their model seems to be to generate 10,000 offers in case one of them sticks. When I reached out to talk to them, there was never anybody there. No deals should be done without human interaction.”

AI has made fraud schemes way more complex, so it can be harder to know if anyone in any part of the transaction is who they say they are, according to Baird. AI almost creates an opportunity for people to unknowingly reveal confidential information, not realizing that there may not be security around it, she added.

“Unfortunately, AI is good at making information look really good even when there’s no understanding of the context,” Baird said. “It can be biased to give you the answer it thinks you want. So, the closer you are to making an investment, the more you need to turn to knowledgeable people to verify what AI provides.”

AI’s potential

So far, Marcus and Millichap has seen cost increases rather than cost reductions from their AI investments, according to Nadji.

AI can sound very confident even when the answers it provides are not always right, so training personell how to use AI effectively is required, said Doug Carpenter, director & head of data strategy and AI at CenterSquare Investment Management.

“Eventually, this will speed up production levels, especially for repeatable tasks,” he said. “But rather than eliminate jobs, I think this will just create reassignments so everyone can take advantage of the higher and better use of their human skills.”

While AI can support marketing efforts, Edeen said, it can be a mistake to have AI bots reach out to customers.

“The best use of AI is to take the extra hours it gives you from data analysis and automation to connect directly with people,” Edden said. “What I hope to see next is for AI to help everyone manage communication overload, to assist people in determining which emails and calls need to be made, but not communication for them.”

Real estate is relationship driven and local knowledge is needed, so you don’t want to automate that away, according to Baird.

“Ideally, AI can help you make better informed decisions with less friction along the way,” she said. “We’re getting closer, but we’re not there yet.”