Facilitated by Lisa Nickerson, CEO of Infinityy, roundtable discussions at CREDA’s AI in Commercial Real Estate Summit brought the conversation to the attendees. Nickerson started with a familiar challenge: new technology only pays off when people actually use it. She compared today’s AI rollout to the early days of CRM adoption, when success depended less on the software than on culture, training, and how organizations encouraged and rewarded their people.
With that in mind, she asked each table to discuss where AI is meaningfully influencing investment, development, and operations; what’s working and what isn’t; where the hype doesn’t match reality; and what lessons might help others in the industry. She also asked each table to name one AI development from the past six months that genuinely surprised them. The themes they coalesced around include:
Data is the low-hanging fruit. The clearest value so far is in finding data, pulling it together from separate parts of the business, and getting useful output from it. Operations stood out as the easiest place to start. Lease abstraction was named as a quick win that saves real money. One group noted that AI doesn’t just speed up analysis but looks at data differently than people do, which makes it a creative tool as well as an efficient one. The obstacles are practical ones: inconsistent property identifiers across teams make linking datasets hard, and old, poorly scanned leases are still difficult for models to read.
Legacy software is losing ground. Many firms can now quickly build customized tools in-house, often matching or beating long-standing real estate software. The tedious process of moving underwriting work from Argus into Excel came up as a workflow that seems likely to change. Off-the-shelf “do everything” underwriting tools haven’t delivered yet. Firms are getting more out of tools and skills they shape around their own processes.
Adoption, not technology, is the real hurdle. People in data-heavy roles tend to be eager to adopt, while operations and off-site staff are harder to bring along. Approaches that are working include in-person training that demonstrates what AI can do, and recurring informal conversations where colleagues compare what’s working and what isn’t. One thing that isn’t working: AI committees made up of the busiest people in the company, who have no time to follow through. The advice was to commit real time and resources and accept some early failures.
Make it matter to employees. Promises of higher revenue or lower costs don’t motivate staff on their own. Adoption gains traction when people can see that AI makes their own work easier and more meaningful, and when organizations reward those who lean in. Several speakers warned of a growing divide between firms that adopt and firms that don’t.
Security and governance remain the biggest hang-ups. Attendees worried about letting AI tools access company files, particularly around liability, and about employees putting company information into personal accounts. A recurring conclusion was that firms need an enterprise-level solution, because people will use these tools anyway, and defining the firm’s acceptable level of risk early, since proprietary data and knowledge are among a company’s most valuable assets.
Hype vs. reality. Some expected wins haven’t arrived yet. Site feasibility tools took more effort to configure than they saved. Excel models built with AI were often accurate but overly complex and hard for a person to follow or review. Several people compared AI output to work from a junior team member: useful, but it needs checking. “AI slop” also came up, meaning messages that are obviously machine-written and get ignored, which raises questions about keeping a consistent voice as these tools roll out across a company.
Trusting the output. One example involved AI reviewing a large set of due diligence documents and ranking risks from high to low, only for colleagues to insist someone still had to read every page. The group pointed out that experienced reviewers have always skimmed for known risk factors, and AI can direct that attention more efficiently. Keeping knowledge in the company. When individuals build their own custom tools, those tools can walk out the door when the person leaves. Groups saw this as another culture and governance issue: firms need to encourage sharing, not private pockets of expertise.
The post AI’s Biggest Hurdle in Commercial Real Estate Isn’t the Technology appeared first on Market Share.



