Artificial intelligence tools are finding their way into many aspects of facilities management, from security to HVAC maintenance—but many facilities professionals aren’t quite sure about them. At the “AI Strategies for Facility Leaders: Separating Fact from Fiction” session at IFMA World Workplace 2025, a word cloud compiled by attendees contained keywords like “assistant,” “future,” and “innovative”—but other FMs contributed words like “scary” and “jobless.”
The actual implementation of AI tools can indeed be scary—but these tools can also serve as a true assistant to free up facilities managers for more complex tasks that really need a human’s input. The key is to have a good strategy that guides your investment decisions and ensures you don’t end up with a tool that doesn’t serve you well. The panelists at this session, moderated by Bob Clarke, SVP, Client Experience & Operations Support at ABM Industries, shared these tips for building your roadmap to AI implementation.
1. Align on Goals Early
It’s critical that you first figure out what you’re looking to do with AI, because a broad term like “artificial intelligence” can mean any number of things, said Kayla Oliver, head of product, partnerships, and innovation for ABM. “Are you looking to go after energy savings and sustainability? Are you looking to go after cost savings or the customer experience?” Oliver asked. “It really matters, defining and aligning on a goal. If you don’t do that up front and talk about actual use case implementation or buy-in, it’s going to be challenging. Knowing what you’re going after matters.”
Think about “the art of the possible” with AI and spend time ideating, Oliver added. AI chatbots can help with this. Determine use cases where you could pilot the technology, then pick one to start with.
2. Prepare People for Change
“Not every use case needs a process change, but by and large, if you want to have big impact, you need to change your process a little bit,” Oliver said. “You’re not going to get outsized impact if AI is just overlaying on top of it.”
3. Spend Time on Data Cleaning
Right now, you probably have siloed data across multiple systems, including your BMS and CMMS if you have them. This fragmented data complicates AI deployment and impacts your efficiency, said Reshmi Chakraborty, VP of tech and data for ABM. “I’ve seen data where the same HVAC asset has three different names,” she said. “If the foundation is not right, whatever you’re getting as an output will not be real, so the first step is data quality.”
Go through your streams of data and figure out where you can remove duplicates, fill in missing information, or standardize naming conventions, Chakraborty recommended. This cleanup needs to be an ongoing process, not a one-time event. AI operates by looking for patterns in data and predicting what should come next, so it’s important to give it good data in which it can recognize those patterns.
4. Reskill and Upskill Your Existing Team
As the word cloud at this session demonstrated, not everyone is on board with AI tools yet. Rather than leaving those folks behind, bring them along with you. Make sure your teams understand the value of your new tool and how it works.
“Data is wonderful if you know where to use it and how to apply it, but sometimes we assume folks know how to leverage data,” said Sean Luster, Vice President, Real Estate and Workplace Services at Salesforce. “We need to take a step back, evaluate ourselves, and make sure we’re leveraging data appropriately first.”
There are plenty of free courses online to walk people through the basics of AI, Oliver added. Lowering the barrier to participation makes your AI solution more practical and actionable.
“Don’t leave anyone behind,” Luster said. “Make sure everybody has access to knowledge and information. Have a beginner’s mindset. We found that once we embarked on the journey, not everyone was coming with us, and we had to go back and grab some folks and say, ‘We need you with us.’ We had to figure out different ways to educate folks and make sure they were along for the ride.”
5. Vet Vendors Critically
Vendors of AI tools need to be evaluated against what you’re hoping to achieve and whether their products do what the vendors say they will do, Chakraborty said. “Don’t just go with the first vendor who comes to you and says, ‘It’s plug and play. Everything works fine,’” she added. “Don’t just see the demos, because that won’t give you anything. Ask questions like, ‘Show me your case studies. Show me the numbers where you’ve done this use case for another client.’ If you want to scale up, ask how feasible it is to scale that solution within your tech ecosystem.”
6. Track Value
Find ways to determine your AI tool’s value to your organization, whether it’s the number of work orders you’re able to process, average time spent on an issue, or another metric, and consistently track that value. “That helps you when you want to make the case to scale and go after additional use cases,” Oliver said.
“Focus on the process, because if you’re planning to do AI, you need to fix the process first,” Chakraborty added. “If you do not fix that, whatever software tool you put on top of it is not going to work. Build that foundation so you can launch the AI whenever you need to.”
