

A veteran multifamily operator says shifting institutional strategies are creating new opportunities for specialized property managers, particularly in the workforce housing sector. Ron Kuta...




The commercial real estate industry has seen a surge of AI-powered tools that claim to transform how professionals handle documents and data. Yet, according to Oded Noy, Chief Technology Officer at Crexi, most of these solutions overlook a key aspect of user behavior that is hindering widespread adoption.
“Chat is awesome, but cannot be where you start,” Noy explained, describing a finding that emerged during Crexi’s development of its AI document extraction platform, Vault. This insight challenges the common belief that conversational interfaces are the best entry point for AI in real estate.
Noy’s team initially built Vault around a chat-based interface, following the industry’s trend toward ChatGPT-style interactions. The outcome was underwhelming. “We started by, you can ask it any question you want, and it was a trade form, and we gave it to people to use, and they looked at the screen and said, ‘What do I do now?’”
The issue was not with the technology, but with user expectations. Unlike general-purpose AI tools, where users expect to have open-ended conversations, real estate professionals interacting with specific documents have different needs and mental models.
“When you go to a ChatGPT, you know that you’re talking to an open ended interface, and you come with that mindset,” Noy said. “When you go to a particular document you really need to be [thinking differently].”
Even after Crexi added prompt buttons with suggestions like “show me what’s on your rent roll” or “show me where the comps are,” users would only interact briefly before moving on. “People started using them, and they click a couple of buttons and say, ‘Well, that was nice,’ and then they moved on doing whatever.”
The turning point came when Noy’s team recognized that AI must blend into existing workflows rather than introduce new ways of interacting. “Our job as a product and engineering, and particularly on AI [team] is that it completely disappears, meaning I’m used to dealing with, if I deal with retail, and I have a bunch of offer memorandums, because I constantly get them.”
Rather than requiring users to learn new methods of interacting with technology, effective AI integration means understanding the context in which professionals already operate. “If I go to Crexi, there’s a listing on, you know, 52 Myers Street, and not minimal for some retail space, if I had an offer memorandum from a month ago, it should appear right next to it completely in context.”
This type of integration depends on combining multiple data sources, public data, licensed information, proprietary platform data, and user documents, into a unified experience. “That is what’s hard to do, and it took us a while,” Noy acknowledged.
This realization has broader implications for AI adoption throughout commercial real estate. Many startups are still building chat-first interfaces, assuming that conversational AI is the future of professional tools. Noy’s experience suggests otherwise: the most effective AI implementations will be those that users barely notice as AI.
“People don’t know how to ask the right questions, or they don’t know what they don’t know,” he explained. Instead of expecting users to become better at prompting AI systems, successful products must anticipate user needs within the workflows they already use.
This lesson goes beyond document extraction and applies to how AI will change commercial real estate work overall. According to Noy, “much of how professionals are going to spend their time is going to be different, because AI is going to do some of the grunt work that they don’t want to do, up to 50% of the time is going to be done differently.”
However, this shift will only succeed if AI tools integrate seamlessly into established professional routines, rather than requiring users to adapt to entirely new interaction models. The chat interface, despite its popularity in consumer AI applications, may ultimately not be the right path for enterprise real estate technology.
Every month we conduct hundreds of interviews with
active market practitioners - thousands to date.
Explore similar articles from Our Team of Experts.


A veteran multifamily operator says shifting institutional strategies are creating new opportunities for specialized property managers, particularly in the workforce housing sector. Ron Kuta...


AI-powered search is rapidly changing how buyers discover real estate agents and properties. This shift is creating new winners and losers, but many brokers haven’t noticed yet. Rummy Dhan...


The commercial real estate industry stands at a technological tipping point where network effects are dramatically amplifying the value of digital platforms, according to Steve LaMotte Jr., ...


When COVID-19 struck, Johnny Yeh watched as the coworking empire he’d helped build suddenly faced an existential threat. “Immediately, we saw a hit on freelancers using single de...


Brian Durham, managing broker and team owner at WeGo Real Estate., has spent his career on the other side of that gap — first processing over 177 transactions in 15 months for a hedge fund...


Real estate listings have become sharper and more responsive, with better photos, clearer descriptions, and faster replies. But the impact of AI in real estate goes far beyond surface improv...
