State of AI in FM (Artificial Intelligence and Facilities Management

Enterprise AI has quietly left the lab. Your buildings did not get the memo.

Open AI just released its 2025 “state of enterprise AI” report. Buried in the charts is a simple truth: AI is no longer the side project of a keen intern, it is becoming the nervous system of how work gets done.

ChatGPT Enterprise usage is up multiples, structured workflows are up even more, and the average organisation is consuming hundreds of times more "reasoning" than a year ago. That is not people asking for holiday recipes. That is systems work.

What struck me reading it, as someone building Cleverly to bring AI and automation to real estate and facilities, is how clearly the data exposes a gap our industry still has. Enterprise AI is sprinting ahead. Most buildings are still jogging behind, carrying clipboards.

AI is no longer a toy, it is a workflow

One of the headline findings is that usage of structured workflows like Projects and Custom GPTs is up 19x year-to-date. That is the moment AI stops being "ask me a question" and starts being "run this process whenever X happens."

In software companies, that looks like AI-assisted code reviews, automated security checks, personalised marketing. In facilities, it should look like:

• Triaging incoming work requests, classifying them by trade, priority and location, and routing it to the 'right' supplier.
• Generating the quote, sanity-checking it against historical data, and flagging the ones that look suspiciously optimistic.
• Building the compliance pack, chasing missing certificates, and politely nagging whoever is blocking the fire door sign-off (or the fire door for that matter!).

The technology clearly exists. The OpenAI data show enterprises turning AI into repeatable, integrated workflows. The question for our sector is not "does this work" but "why are we still accepting email inboxes as a workflow engine."

Workers are getting 40–60 minutes a day back. Imagine that in FM.

The report says 75% of workers using AI feel it has improved the speed or quality of their work, and typical users are saving 40 to 60 minutes per day. Heavy users save more than 10 hours a week.

Translate that into a hard FM context:

• A maintenance coordinator who spends less time copying notes from PDFs into a CAFM and more time fixing the scheduling conflicts that actually matter.
• A property manager who no longer hand-assembles quarterly reports, because the system can summarise the last 90 days of jobs, spend and SLA breaches on demand.
• A head of operations who can ask, "Which of our suppliers consistently over-run on call-outs for out-of-hours jobs?" and get a chart instead of a promise.

The numbers in the report are sector-agnostic, but the impact in facilities is unusually asymmetric. We run on thin margins, fragmented data and a heroic amount of human glue. Every reclaimed hour in an FM team tends to go straight back into resident happiness, uptime or safety.

The frontier vs the median... and the FM reality

OpenAI talks about "frontier" workers and firms. The top 5% of users send 6x more messages than the median. Frontier firms send 2x more messages per seat and integrate AI more deeply across teams.

You can feel that split in our world already. I meet FM leaders who are using AI to:

• Design and roll out new PPM regimes, simulated against years of historical failures.
• Auto-score incoming quotes and suggest target negotiation points.
• Build on-the-fly dashboards from natural language across sites, assets and suppliers.

Then I meet others who are still trying to get a clean asset list out of three spreadsheets, two contractors and a filing cabinet.

The gap is not intelligence. It is implementation. Frontier FM operators behave as if AI is a colleague that sits inside their existing systems. Median operators treat AI like a search engine they occasionally ask for a template.

The constraint is no longer the model, it is organisational courage

OpenAI says it releases a new feature or capability roughly every three days. The bottleneck is no longer what the models can do, it is whether organisations can adapt processes, governance and culture quickly enough to take advantage.

In real estate and facilities, the usual suspects show up:

• Data is scattered across CAFMs, email, SharePoint, contractor systems and someone's personal Dropbox.
• Processes live in people's heads, or at best in a PDF called "FINAL_v7_really_final".
• There is a reasonable fear among engineers and coordinators that "AI" translates to "headcount reduction", rather than "less drudgery, more judgment."

If we do nothing, the outcome is predictable. The "frontier" operators who lean in will offer faster response times, better reporting, cleaner audits and more competitive pricing. Everyone else will be competing with a 10–20 percent operational handicap they can neither see nor quantify.

A simple playbook for real estate and FM leaders

So what do you do if you run an FM business, property portfolio or operations team and do not have a resident AI lab in the basement? A few pragmatic steps I see working with Cleverly customers:

1. Start with one ugly workflow, not a grand strategy Pick a process that everyone hates but that happens all the time: reactive job triage, quoting, compliance packs, voids management. Instrument just that.
2. Put AI where the work already is Do not ask your teams to open a new tool for "AI." Put AI inside the systems they already live in.
3. Create your own "frontier" users You do not need your whole organisation to become AI-native overnight. You need a few champions who are unreasonably curious.

2025: the year your CAFM grows a brain

The OpenAI report is largely about big enterprises across sectors. The pattern is clear though: AI is moving from experimentation to infrastructure.

In real estate and facilities, we have a slightly unfair advantage. Our data may be messy, but our workflows are incredibly rich and repeatable. That is exactly where AI and automation are most powerful.

If your competitors are quietly turning AI into the invisible operating system of their business, and you are still manually forwarding jobs from one shared inbox to another, you are not standing still. You are moving backwards.

The good news is that the technology curve is finally working in our favour. You no longer need a data science team to get started, you need intent, a few good processes and systems that actually talk to each other.

That is what we are trying to build with Cleverly: a platform that lets facilities and real estate teams plug into this new wave of AI, without needing to become AI companies themselves.

The state of enterprise AI is clear. The remaining question is: what is the state of AI in your buildings?

#workingcleverly