In the rush to adopt AI for facilities management, the initial conversations about deploying AI tools almost always focus on speed of deployment. However, that’s not a sound approach. Instead, managers should first ask whether the operational data behind the scenes can support what the tool promises to do, because, without the proper data foundation, any AI initiative will likely stall out within a year and fail to produce tangible results.
Facilities organizations have no shortage of data. In fact, from maintenance logs and asset records to lease files and energy meters, most organizations generate far more data than they realize. But volume alone is not enough. AI can only draw reliable conclusions from data that is clean, connected and consistently structured across systems. For example– when asset names differ from platform to platform, and when work orders live in scattered notes, AI will simply automate disorder at scale.
Where AI initiatives typically go wrong
The first mistake is to choose a tool and then work backward to determine whether the data the tool requires exists. Instead, before choosing any AI tool, facilities leaders need a clear picture of what data they have and where this information lives.
The second mistake is deploying narrowly defined AI tools, instead of incorporating AI into the basic operating system for facilities management. AI delivers the most value when it is embedded across the full range of facilities operations, drawing from the same governed data set on which maintenance, real estate and finance teams already rely. Beginning with targeted use cases isn’t a bad way to begin, but facilities should always have an eye towards creating a larger, AI-enabled system to really see AI move the needle.
The third mistake is to skip over data governance. Someone has to own data quality on an ongoing basis, and without a named owner and a defined process for keeping records current, even a well-integrated system can degrade rapidly as new equipment, renewed leases, and space changes go undocumented.
What good data governance looks like
AI does not require perfect data across the entire enterprise, at least, not in the beginning. So long as there’s a clearly defined starting use case, IT just needs to provide access to the minimum data required to support it and a workflow owner with the authority to act on what AI surfaces.
Let’s look at a concrete example in maintenance. During their work, experienced technicians build an instinct for which equipment is trending toward failure. However, that knowledge is unscalable, and it leaves the organization the day the technician does. In comparison, AI can track asset records, service history and maintenance activity, so long as they all live in one connected system and detect failure patterns as they emerge. AI transforms signals that would otherwise go unnoticed into warnings that enable a team to act before a breakdown occurs.
Space and lease data benefit from AI adoption just as much. Badge counts and meeting room bookings tend to overstate real utilization, and the people who see the space day to day often know the true numbers but lack the data to make their case to the people controlling portfolio decisions. When utilization data sits alongside lease expiration dates and headcount projections, a facilities leader can walk into a portfolio review and show, with evidence, that a floor has run at low occupancy for two straight quarters while its lease renews within the year.
The real question before any tool selection
Every one of these outcomes depends on the same underlying condition: a unified operational platform where asset records, maintenance history, space data, lease obligations and energy consumption all speak a common language. That foundation is what separates organizations getting real value from AI from the ones still explaining why their pilot never delivered.
Before the next AI demo lands on your desk, bring a different question to leadership. Not which tool to buy, but whether the data foundation exists to make any tool worth buying. It is a harder conversation than a software pitch, and it is the one that determines the outcome.
Kimberly Castle is President of Planon North America, where she leads regional strategy, operations, and customer engagement. She has more than 30 years of experience in corporate real estate and workplace technology, having held senior roles at CBRE, MetLife, JLL and Suncor.
