As we all know, AI these days comes up in almost every executive meeting, LinkedIn post and vendor pitch, often presented as the answer to everything. In property and facilities management, that usually means predictive maintenance, energy cost reduction, tenant engagement and much else.
Some of that enthusiasm is justified, of course. But as many applications as AI undoubtedly has, we should remind ourselves that it is not a magic ingredient that will somehow make everything better. What matters in the property sector, for example, is not whether a tool is ‘AI-powered’, but whether it improves the way we manage buildings, assets and people, and whether it operates reliably within the reality of dispersed estates, legacy infrastructure, compliance obligations and tight budgets.
Integrating AI into property management workflows systems today can add a lot of value, but it’s also a long way from being a mature technology.
Across the industry, organizations often let the promise of shiny new technology lead them, instead of being outcome-focused. But “we need to put AI in our business” is the wrong place to start.
The right starting point is the set of questions you would ask even if AI didn’t exist, such as:
Can we reduce the time our teams spend reading long helpdesk trails?
Can we get a clearer view of the payments we are processing across our portfolio without manually trawling through thousands of documents?
Can we increase the utilization of our underused spaces and improve the effectiveness of hybrid working?
If you can’t map a clear route to answering questions like these, adding AI to a project and pushing engineering teams to deploy a chatbot likely won’t deliver meaningful change.
Much of the real value AI delivers today sits in unglamorous, text and document-heavy back office work that absorbs time and attention. Typical examples include:
Take a recurring HVAC fault in a shopping center. Over its lifecycle, it can generate a long, fragmented trail: tenant emails, helpdesk notes, engineer scheduling, on-site comments, quotes, photos and follow-on works. By the time it reaches a contract manager or client, it’s difficult to see the whole story: what’s happened, what’s been tried, what’s outstanding and who needs to do what next.
Large language models (LLMs) can be really useful in surfacing key insights from these long chains. They are already good at condensing lengthy histories into clear summaries, extracting key dates, decisions and risks, and suggesting sensible next steps based on context.
It’s not about asking AI to run the job; it’s about using it to structure the information load so that the accountable human can reach a better decision, faster.
Property management generates a heavy stream of financial paperwork: supplier invoices, purchase orders, credit notes, service charge reconciliations and more. Tracking where money is going across a portfolio remains a largely manual process.
AI-powered document intelligence tools such as Microsoft Document Intelligence are already helping here by:
Automatically extracting key fields from invoices and other documents
Standardizing supplier and cost category information
Flagging duplicates, anomalies or items that don’t match expected patterns
Feeding cleaner data into finance and BI systems without endless rekeying
This gives finance teams structured, reliable data and frees their time for work that requires judgement: challenging costs, identifying trends and advising clients.
Many organizations still lack a clear picture of how their space is used in a hybrid world. Meeting rooms and desks can be both overbooked and underutilized depending on the day, and unused space carries a cost that organizations can no longer justify. There is strong potential for AI-integrated systems to:
Analyze booking patterns, occupancy data and no-show rates
Identify consistently underused areas and peak pressure periods
Suggest changes to layouts, neighborhoods or booking rules
Inform decisions about consolidation, repurposing or investment
AI does not set workplace strategy. It provides the insight that enables workplace and property teams to make better-informed decisions about how much space they need and how to configure it.
Looking further ahead, computer vision, AI that understands images and video has great potential. Imagine walking through a plant room with a camera and having assets recognized and tagged. Or identifying missing labels, damaged equipment or housekeeping issues from images; or automatically comparing current site photos with last year’s to spot changes or deterioration.
The building blocks for this already exist, though they are not yet mature enough for every use case. The essential point is to be precise about what current computer-vision advances can and cannot do. AI-enabled systems can support inspections, but they cannot yet automate them end-to-end. Image annotations can assist a surveyor or engineer but not replace their judgement.
The value comes when vision models are combined with robust processes, domain expertise and the data held in a company’s systems.
We’re doing lots of work on AI, and one principle underpins all of it: AI suggests; humans decide.
We keep a human in the loop, especially where decisions affect people, carry safety or compliance implications or involve significant spend or contractual commitments. In those cases, AI-enabled systems present the right information to the right person, surface sensible options or highlight anomalies and reduce the effort needed to apply sound judgement.
Before applying AI in property operations, we all need to focus on three direct questions:
Where are we repeatedly reading or rewriting the same kind of information? That’s a clear candidate for summarization or drafting support.
Where do we lose the big picture because the data is large or fragmented? That’s where AI can help join dots and highlight patterns that are hard to see manually.
If you can answer these in plain language, you’re already ahead of most AI strategies. The technology will continue to evolve at pace. The organizations that benefit most will stay close to operational detail, identify pain points before proposing solutions, stay realistic about what AI can and can’t do, and keep people in charge of decisions.
Editor's Note: Author, William Wildridge is the Director of AI and Innovation at Bellrock, where he leads the development of practical, AI-driven solutions across the group’s products and services. He previously founded WiggleDesk, a B2B SaaS platform for workplace operations, acquired by Bellrock in 2025. Prior to this, William spent nearly a decade at Google working on large-scale data, machine learning and AI projects. He has led award-winning AI initiatives used in public-sector and enterprise environments, and serves on BSI committees shaping AI and synthetic data standards. His work focuses on translating advanced AI into practical, high-impact operational systems.