Key Takeaways:
As vice president of facilities management at Miami Dade College in South Florida, Leobardo Bobadilla’s efforts to use AI to capture and analyze the college’s facilities data also have led to several important lessons applicable to managers looking to implement the same process in their organizations. One key lesson involves training.
“We really have to prepare our people, our staff for this change that’s coming,” Bobadilla says. “Training is one of the top, if not the top item I would put on the list with what’s coming with AI. You have staff who are at different points on the spectrum as far as what they know about this and how they feel about it. What does it truly mean for what they do day in and day out?”
A successful AI-PdM process also requires that managers create a robust path for data to make its way through the various platforms that departments and technicians rely on when planning and prioritizing maintenance activities.
“We have to start figuring out how to integrate different parts of this together,” Bobadilla says. “Our HVAC experience has taught us that (AI) can be very helpful when it comes to predictive maintenance. But now we have to figure out how to make the connection between that information we're getting back from AI and our work order system.
“If we can do that, it eliminates a manual step, and automatically, AI will be able to more or less issue work orders to let our technicians know that there’s something that may be going on here and that someone should check it out. That will further streamline our process and determine how quickly we can respond to issues.”
Among the hurdles facing managers and their teams who are beginning the AI-PdM process are a host of misconceptions and myths surrounding the technology and its impact, says Paul Morgan, chief operating officer, real estate management services and head of workplace management with JLL. Among them are these:
AI will replace technicians. “In reality, AI augments diagnosis and prioritization. It still takes a skilled technician to execute the repair and validate the root cause.”
We just need to buy the software, and it works out of the box. “AI is only as good as the data feeding it. Without clean, consistent data and some tuning period, predictions will be unreliable or noisy.”
PdM means we no longer need preventive maintenance. “PdM complements preventive maintenance. It doesn’t eliminate the need for baseline maintenance hygiene, and blending both approaches is usually more effective than an all-or-nothing switch.”
AI predictions are always accurate. “Early models will produce false positives and false negatives. Managers need to treat AI as a maturing capability, not an infallible oracle.”
This is just the automation of alerts we already had. “True AI-driven PdM is fundamentally different from simple threshold-based alerting. It is pattern recognition across multiple variables and historical failure modes, not just a temperature exceeding a set point.”
While managers might still hesitate to commit time and energy to tapping into AI to support PdM, there is little doubt that applications of AI in facilities will continue to expand.
“AI is moving facilities management from reactive and calendar-based approaches toward continuous, condition-based operations,” Morgan says. “The near-term future looks like sensor data, BMS and IoT feeds, and historical work order data being fused into models that predict equipment failure before it happens — shifting PdM from run to fail or fixed-interval maintenance to true failure-mode prediction.
“I expect AI to move beyond whole-building to portfolio-level optimization: dynamically balancing energy use, occupant comfort, equipment longevity and technician capacity in real time. The biggest shift will be AI acting as a decision-support layer that recommends prioritized work orders, occupancy, flags anomalies humans would miss and continuously retrains itself as buildings and equipment age and occupancy patterns emerge, turning facilities management from a cost center into a source of operational and financial predictability.”
Bobadilla says that while his department and technicians have made the commitment to AI and PdM, he also knows that challenges lie ahead.
“We are all still learning — everyone, including us,” he says. “Every day, there’s something new I hear about AI, someone who has thought of another great, amazing application to incorporate AI into. A big part of it is to continue to engage in this to help the team and the staff with change. If there’s one thing that is important for leaders to do, it’s to grow and develop your team.
“Change can be scary to some people. We’re going to have people at both ends of the spectrum — people who are charging ahead and leading the way, and on the other end, people who want nothing to do with it and would rather be left alone and do things the way they’ve always done them.”
Dan Hounsell is senior editor for the facilities market. He has more than 30 years of experience writing about facilities maintenance, engineering and management.
AI, Predictive Maintenance and the Path to Success
How AI Predictive Maintenance Succeeds: Training, Data Integration and Trust
