REAL ESTATE NEWS

Predictive AI is rapidly reshaping how retailers make real estate decisions, prompting store closures, relocations, and resizing well before traditional indicators like foot traffic or comparable sales reveal trouble. Across the industry, technology is becoming less of an experiment and more of an integral guide for investors, operators, and site strategists.

According to Bryn Feller, Managing Director at Northmarq, the sector stands at a pivotal moment. She tells GlobeSt.com that commercial real estate has reached "a fascinating inflection point" as artificial intelligence begins to supplement the intuition that long defined the business.

"For decades, this business has been built on personal experience, pattern recognition, and imperfect information," Feller told GlobeSt.com. "What AI is beginning to do is introduce an entirely new layer of analytical clarity into a sector that has historically operated with more intuition than instrumentation."

Feller described the current landscape as split. Much of commercial real estate still uses AI for administrative or efficiency tasks—summarizing leases, organizing research, and automating once time-consuming jobs. "Those uses are helpful but not transformative," she said.

The real momentum, Feller added, lies with the institutional and forward-leaning segment of the market. A small but growing group of investors, operators, and asset managers are now using AI not just for productivity but as "a genuine decision framework." "That is where technology stops being a novelty and starts becoming infrastructure," she said.

Feller pointed to platforms such as CFS (cfsnnn.com) and its retail-focused deployment, Renewal.ai, which apply machine learning and predictive modeling to assess tenant durability and renewal probability—an area that has long lacked precision. "Commercial real estate has never had a data problem. It has always had a signal problem," Feller tells investors. "We are drowning in data. Rent rolls, traffic counts, demographics, sales figures, competitive sets. What AI is beginning to do is turn those disconnected data points into a coherent signal."

Traditional Metrics, Amplified

Despite AI's growing role, Feller emphasized that classic real estate fundamentals are not going anywhere. Rent levels, sales productivity, traffic patterns, and demographics remain vital, but AI now contextualizes them at scale.

Instead of merely asking, "How strong are the sales at this store?", AI allows investors to ask more layered questions—such as why certain tenants renew in one corridor but not another, or what early signals might precede a store closing or relocation. "In that sense, AI is not replacing the traditional toolkit of commercial real estate," Feller said. "It is amplifying it."

While AI certainly improves efficiency, Feller reminded that time savings are just part of the story. "Workflows that once required days or weeks of data assembly can now occur in seconds," she said. "But the real advantage is not efficiency, it is informational asymmetry."

That head start can translate into a material advantage. "If an asset manager can identify tenant weakness before it becomes obvious to the brokerage community or anticipate renewal risk before a lease expiration becomes a headline issue, that informational head start can materially protect yield and asset value," she said.

"A Flashlight, Not a Crystal Ball"

Still, Feller warned against treating AI as an all-knowing oracle. "AI is not a crystal ball. It is an incredibly sophisticated flashlight," she said. "It can illuminate patterns that were previously invisible, but it still requires judgment, experience, and intellectual humility to interpret correctly."

She also pointed to the "black box" problem, where models produce authoritative-looking results without transparent assumptions. Responsible firms mitigate that risk through rigorous back-testing and ongoing recalibration. "At firms pushing the frontier, including platforms like CFS and Renewal.ai, there is a growing emphasis on blind back testing and transparency in predictive frameworks," Feller said. "That discipline will be critical if the industry is going to trust these tools at scale."

Ultimately, she said, pattern recognition remains at the core of retail real estate. "The best investors historically succeeded because they developed an almost instinctive feel for tenant behavior, location dynamics, and consumer patterns," she said. "Artificial intelligence is simply accelerating that process. Or as I like to frame it: The future of commercial real estate will not belong to AI. It will belong to investors who know how to think with AI. Those who learn to integrate these tools thoughtfully into their decision-making will have a meaningful advantage. Those who ignore them entirely may find themselves competing against people who simply see the chessboard more clearly. That clarity can be extraordinarily valuable."

CRE Platforms Move Past ChatGPT

At Crexi, VP of product strategy Adam Siegel said AI-powered underwriting tools are still in early development but advancing quickly. "Right now, adoption is happening in several ways," Siegel told GlobeSt.com. "Some investors are putting property information into tools like ChatGPT to see what insights they generate, while startups are building solutions to address specific challenges such as lease abstraction."

AI is already making property evaluation faster and deeper, Siegel said. "It will take some time for the technology to mature, but given how quickly AI, what I often describe as 'abundant intelligence,' is advancing, that timeline may be shorter than expected."

Use cases are emerging across the board. Siegel noted that developers and expanding retailers are leveraging AI to identify sites that meet zoning or demographic criteria—tasks that once took hours of manual research but can now be completed "in just a few clicks."

Still, he said it's too soon to say whether AI is redefining the metrics investors use to underwrite assets. For now, the technology is expanding horizons, revealing markets investors might not have previously considered.

He offered a simple example: an investor comfortable with Tucson's multifamily market might discover similar demand profiles in more affordable markets like Logan, Utah, or Glassboro, New Jersey.

Siegel said some asset types are easier for AI to analyze. Single-tenant net-lease deals involve a single lease, and multifamily properties rely on standardized 12-month agreements, whereas complex retail centers and offices present more interpretive challenges. "Over time, as AI improves its ability to read and interpret the wide range of documents involved in real estate transactions, the technology will become more reliable," he said. "There is a lot to look forward to, but the industry is not fully there yet."

JLL: From Retrospective to Predictive

At JLL, AI has shifted site selection from backward-looking analysis to forward-looking probability modeling, said Sarah Wexler, Managing Director of Consumer Goods and Services. She told GlobeSt.com that brokers once relied on basic demographics and past performance. Now, AI weighs dozens of additional factors—from evolving business mixes to infrastructure plans and socioeconomic trends.

"This doesn't replace traditional metrics like average household income or traffic counts, but rather, it contextualizes them within predictive frameworks that anticipate three- to five-year market conditions," Wexler said.

She added that transparency remains a critical issue, as many AI tools still operate as "black boxes" with opaque logic. "Data quality directly impacts output reliability; data integrity is amplified at scale," she said.

Over-reliance on historical data also poses risks, Wexler said. "AI models trained on pre-pandemic data initially failed to predict behavioral shifts in 2020-2021," she explained. "The tech optimizes for pattern recognition but struggles with unprecedented scenarios."

Legal implications are also top of mind. "Human oversight remains essential—AI should augment, not replace, experienced judgment in final underwriting and investment decisions," Wexler said.

Realty Income: A Long-Term Head Start

For Realty Income, AI's promise is not theoretical—it's already fully embedded. President and CEO Sumit Roy told GlobeSt.com that the firm's predictive analytics platform has been "fully implemented and embedded across core investment and portfolio workflows" for seven years. "This is certainly not in an experimental capacity," he said.

The system combines proprietary financial and leasing data from more than 15,500 properties with millions of external data points housed in the company's data warehouse. Roy said that the foundation has supported more than $50 billion in transaction volume, reinforcing disciplined underwriting and capital allocation.

Realty Income integrates predictive modeling across acquisitions, asset management, leasing, development, and disposition—making data-driven insight a standard input into every decision. The platform relies on supervised and unsupervised machine-learning models rather than generative AI, which Roy said ensures outputs are "structured, explainable, and statistically validated."

Avoiding generative AI also mitigates the risk of unreliable conclusions, he added. A team of 10 data scientists continuously validates and monitors the models. "The insights generated by our team complement traditional fundamental analysis and underwriting, ensuring that quantitative insights and human expertise work together in unison," Roy said.

Source: GlobeSt/ALM