The Building Is Talking: How Agentic AI Is Reshaping Facilities Management

Global facilities and operations management has never been a solo endeavor. The work happens in teams across disciplines, time zones, cultures, and building types and the most effective operational strategies are always the product of that collective intelligence.

Over two decades of leading facilities and corporate real estate programs, our teams have managed more than a million square feet of headquarters space, executed brownfield renovations and manage day to day operations across 14 countries, opened 45 bank branches in 17 months, and guided operational programs above $15M across pharmaceutical and high regulated environment, financial campuses, commercial offices, and retail portfolios.

That work has been grounded in the US and Latin America, built through direct operational leadership, and informed by close collaboration with EMEA markets where governance, labor structures, and infrastructure maturity add layers of complexity that a single-region perspective simply cannot account for. The operational strategies that emerged from that experience were not designed in boardrooms. They were stress-tested on the floor, refined through failure, and validated through the kind of multi-stakeholder, multi-country accountability that leaves little room for theory.

That is the perspective this article is written from. Not a vendor briefing. Not a research summary. The voice of the field.

And the field is changing faster than most organizations are prepared for.

What "Agentic AI" Actually Means in a Building

There is a lot of noise right now about AI in facilities and operations management. HVAC manufacturers are embedding chatbots directly into building interfaces. Work order platforms are auto-routing service requests based on equipment history. Predictive maintenance vendors are promising to detect failures before they happen, without a technician in the room allowing remote locations and regional portfolios to thrive.

All of this is real. But what makes it *agentic* meaning an AI system that can perceive conditions, reason through options, and take action is the leap from *flagging* a problem to *acting* on it. And that is precisely where the conversation gets interesting for organizations serious about operational strategy.

Across the portfolios our teams have managed, we have deployed layered detection systems that combined vibration and acoustic monitoring on rotating equipment, AI-powered autonomous drones conducting asset condition assessments supporting labor accident prevention in hazardous location, an energy and controls anomaly detection at the building system level, and IoT-enabled building automation platforms across HVAC, air quality, and lighting. The outcome: predictive alerts that prevented unplanned failures, condition monitoring programs that avoided between $250,000 and $500,000 in avoidable repair costs per site, and a shift from reactive firefighting toward a planned-maintenance posture that consistently held above 95% PM completion across regulated and commercial environments alike.

Each of those systems generated signals. The question was always the same: who decides what happens next?

The answer was always a human. And that should remain true with much sharper human judgment supported by much better AI assistance.

What the Data Says and What It Confirms

Industry data published in late 2025 puts real numbers on what practitioners already know from experience. In a survey of over 1,000 business leaders and facility managers, 65% of business leaders and 67% of FMs reported already using AI in facility operations. Predictive maintenance is the leading use case, cited by nearly half of current adopters. And the number one barrier to scaling AI across portfolios is not budget, not cybersecurity, not even lack of expertise.

It is data integration.

That finding will not surprise anyone who has tried to reconcile asset records, work order histories, building automation data, and predictive maintenance alerts all maintained in separate systems, all requiring manual reconciliation in time for a board report or a capital planning cycle. The fragmentation is real, the reconciliation cost is significant, and any AI layer built on top of siloed infrastructure will inherit the errors and gaps of each one.

The technology is not the bottleneck. The data architecture underneath it is.

This is the conversation has repeatedly with clients: you cannot buy your way out of a data-quality problem. You have to design your way out of it. For investors evaluating platforms or portfolio companies in this space, that distinction matters enormously the value of an AI product in FM is only as durable as the integration strategy surrounding it.

The Human in the Loop Is Not a Weakness, It Is the Strategy

One of the most persistent myths in vendor briefings and conference panels is the idea that the end state of AI in building management is full autonomy a system that detects, decides, and dispatches without human involvement.

That is not a strategy. That is a liability.

The most expensive failures our teams have encountered over two decades were not caused by a lack of sensors or data. They were caused by a lack of judgment someone who trusted a number on a screen and did not walk the floor, or an algorithm that had not been trained on the specific failure mode it encountered.

The BAS analyst, the mechanical engineer, the calibration technician these are not roles that AI replaces. They are the roles that make AI worth deploying. A seasoned HVAC technician can put a hand on a unit and know something is wrong before any sensor threshold has been breached. That same technician is generating the ground-truth observations that make predictive models more accurate over time. Reduce that workforce too aggressively in pursuit of automation savings, and organizations do not just create a coverage gap they starve the system of the feedback loop it depends on to stay reliable.

The organizations getting this right are not asking "how do we remove people from the loop?" They are asking "how do we give our people better information, faster, so they can cover more ground without losing judgment?" That is the right question. And it is a fundamentally different product, investment, and implementation thesis than the labor-replacement narrative some vendors still promote.

For investors, this distinction carries direct implications for market sizing: agentic AI in FM is a workforce augmentation story, not a headcount reduction story. The TAM looks different, the buyer conversation looks different, and the competitive moat is different.

Where It Gets Regional and Why One Playbook Never Works

One of the most valuable lessons from managing operations across 14 countries is that a strategy that works in a Brasil campus does not automatically transfer to a education facility in Chicago or a commercial headquarters in Western Europe. This is especially true for AI and building technology adoption.

• In the United States, the driver is capacity, not cost-cutting. The FM workforce is aging faster than it can be replaced, and AI-assisted tools systems that digitize institutional knowledge, triage dashboards that let a regional engineer cover a wider portfolio, drone-assisted condition assessments for remote or understaffed sites are filling that gap. The challenge is that the US also carries the heaviest legacy infrastructure debt. Integrating cloud-native AI with aging brownfield building systems is not a software problem; it is an architecture problem, and it takes deliberate strategy, not just procurement.

• In Europe, the entry point is regulatory and environmental. AI is being adopted, but filtered through decarbonization mandates, energy security obligations, and data sovereignty law. Directives governing cybersecurity and sustainability reporting mean that any platform with US-centric data residency will face real legal friction in European deployments, regardless of product capability. This is not a bureaucratic inconvenience it is a fundamental market-shaping condition that operational leaders and investors need to account for before any global rollout conversation begins.

• In Latin America, and this tends to surprise people unfamiliar with the region, AI adoption enthusiasm is the highest of any region in recent global surveys 69% of respondents ranked AI as their top impact technology. The reason is structural: LATAM has less legacy on-premise infrastructure to unwind, which means organizations are moving directly to cloud-native predictive maintenance, connected-worker tools, and IoT-enabled controls without the decade-long transition costs that slowed adoption elsewhere. Our teams saw this dynamic firsthand in brownfield mobilizations across the region sites with less inherited infrastructure adopted new tools faster precisely because there was less to dismantle. The challenge is unevenness: data connectivity, infrastructure quality, and governance maturity vary significantly across markets and site by site. The leapfrog is real, but not uniform.

The synthesis: a global AI platform strategy must account for three fundamentally different starting conditions. What accelerates adoption in one region actively creates friction in another a lesson that has direct implications for any operator, investor, or technology vendor approaching this space with a single global deployment model.

What This Means If You Are a Leader Facing a Decision Right Now

Whether leading as a COO, a Corporate Real Estate executive, a Head of Facilities, or a private equity sponsor evaluating a portfolio company's operational maturity the question is no longer "should we adopt AI in building management?" That question has been answered by the market. The question is "how do we adopt it in a way that actually works, at our scale, across our portfolio?"

The most common mistake we see is treating this as a technology procurement decision when it is actually an operational design problem.

Before evaluating a vendor, organizations need honest answers to a different set of questions:

• What is the actual reactive-to-planned maintenance ratio today, site by site? That number alone will tell more about operational maturity than any platform demo.

• What does the data infrastructure look like across asset management, work order, and building automation platforms? Integrated, or manually reconciled in spreadsheets between reporting cycles?

• What is the real cost-of-failure profile for critical equipment across the portfolio? Not the vendor's modeled savings the actual dollar exposure from deferred maintenance, accelerated capital replacement, and unplanned downtime?

• Does a deliberate maintenance strategy exist for every site, or have some locations defaulted to run-to-failure because the alternative was never properly funded or planned?

These are not technology questions. They are operational strategy questions. And they are exactly the kind of questions where an experienced outside perspective one that has been on the floor, not just in the briefing room changes the quality of the answer.

The Case for Getting Ahead of the Challenge

FOM Solutions was built around a pattern seen repeatedly across global organizations: facilities and operational complexity arriving faster than the internal architecture to manage it. Mergers and acquisitions that double a portfolio overnight. Divestitures that leave behind fragmented systems and no continuity of operational knowledge. Rapid expansion programs where speed-to-open conflicts with long-term operational sustainability. Technology transformations where the platform is procured before the data strategy exists.

In each of these situations, the gap is rarely a shortage of technology options. It is a shortage of experienced operational design the kind that translates strategy into implementable frameworks, aligns service scope and KPIs across diverse building types and regions, and builds the governance structure that keeps a portfolio performing as it scales.

The Fractional COO model exists precisely for this moment. Not every organization needs or can justify a permanent executive to lead an operational excellence program, guide a post-acquisition FM integration, or design the infrastructure for an AI-enabled maintenance strategy. But every organization in this situation needs someone with the experience to do it well, the independence to say what others will not, and the discipline to build something that holds after the engagement ends.

The best time to build that operational foundation is before the challenge arrives. The second-best time is right now.

If your organization is navigating growth, a transaction, or a technology transformation in facilities and building operations or if you are an investor trying to understand the real operational maturity of an asset in this space the conversation starts with the right questions, not the right software.

Find out more about Maria Gonzalez-Burgos