Most conversations about AI readiness in commercial real estate sound the same: Which model is best? Have you tried this tool? Did you see the new chatbot? At the recent Yardi Executive Summit in New York, Justin Segal, president of Boxer Property and a Yardi client of 28 years, argued that those conversations miss the point.<\/p>\n\n\n\n
\u201cIf you\u2019re a senior leader and you\u2019re fetishizing use cases, you\u2019re missing a big opportunity,\u201d Segal told the room during a fireside chat with Shivani Kumar, who leads Yardi\u2019s Applied AI team. His thesis: The value of AI in your business is set long before any chatbot or agent goes live. It\u2019s set by the data pipelines, the security model and the architecture decisions that make AI consumable across the enterprise. If you skip that work, every use case rebuilds the foundation from scratch.<\/p>\n\n\n\n
Segal reached for an analogy any New York developer would recognize. \u201cIf you\u2019re a real estate developer in New York, you\u2019re looking at a site and you\u2019re thinking about air rights and financing and traffic counts. But you\u2019re not thinking about the early 1800s when they put a grid system in Manhattan so that all the buildings are rectilinear and everybody can find where they\u2019re going. And you don\u2019t think about the water that in the mid-1800s was piped in from Croton,\u201d Segal said.<\/p>\n\n\n\n
Developers don\u2019t think about water and electricity because someone built that infrastructure in advance. AI deserves the same treatment inside a company. Segal continued, \u201cWe\u2019re in a phase now in our companies where we have an opportunity to put in those foundational infrastructure elements that people should not even have to think about. If we don\u2019t do that, we end up where the city has evolved with no central planning. Everybody has their own water filtration system, and there\u2019s solar panels on everything, but not for the good reasons we might want them.\u201d<\/p>\n\n\n\n
Boxer activates AI just over 100,000 times a month. About 90% of those activations are not initiated by a human and about 60% never produce output on a screen. \u201cWhen most people are talking about AI today, they\u2019re talking about a person deciding to use it and a person seeing the results. There\u2019s a tremendous amount of opportunity and power in automated AI activation and non-screen, downstream result handling,\u201d Segal said.<\/p>\n\n\n\n
Once data is clean, indexed and accessible, AI runs on schedule, on trigger or in response to other agents without anyone clicking a button. Boxer maintains a service that vectorizes and indexes about 10 million records, distilled from roughly 1 billion underlying records. Any new use case at the company starts with that pipeline already in place. \u201cIf I started with the use cases and didn\u2019t go down to the foundational level and build a pipeline of data that is ready for AI to be consumed, then each use case would either have to replicate that or be limited to whatever information they could get,\u201d Segal stated.<\/p>\n\n\n\n
Building this kind of foundation takes people most CRE companies don\u2019t have on staff, including data engineers and AI solution architects. Segal was direct about the gap and said if you\u2019re looking for a data engineer or an AI solution architect who knows commercial real estate, good luck.<\/p>\n\n\n\n
Leaders don\u2019t need to become technologists, he argued. They need to understand the business well enough to know which problems matter, then bring in the right specialists to build the underlying systems. He compared it to plumbing and electricity at a conference venue, which is of course different work from cooking the food or running the AV.<\/p>\n\n\n\n
End users, meanwhile, need adaptability. \u201cMore likely is that the users will be pressuring the leaders to go faster because they don\u2019t want to do things the old way,\u201d Segal said.<\/p>\n\n\n\n
Asked where AI lands in three years, Segal pointed to a term he expects will define the next phase: \u201cheadless enterprise context.\u201d This includes less time spent finding the right tool and less time pointing AI at the right data. The interface gets dynamic, and the underlying systems handle routing and context behind the scenes.<\/p>\n\n\n\n
That direction tracks with where Yardi is heading. The accounting backbone of an ERP system already provides the clean data, user authentication and security model AI applications need. Connecting to that as a service rather than rebuilding it for every project is what makes the use cases work. It\u2019s also why Kumar\u2019s Applied AI team exists. Rather than waiting for clients to surface AI requirements, the team works alongside operators to identify where AI fits inside existing workflows and where the foundations need shoring up first. <\/p>\n\n\n\n
\u201cIt\u2019s an exciting three to five years ahead of us,\u201d Kumar said as the session closed. For the leaders in the room, the more useful question may be the one Segal kept circling back to. Before asking what AI can do for your business, ask what your business has done to prepare for it.<\/p>\n\n\n\n