Artificial intelligence has become almost unavoidable in workplace strategy conversations.

Every technology provider has an AI roadmap. Every workplace platform is adding a copilot. Every facilities conference has an AI session. And almost every executive presentation contains some variation of “AI will transform the future of work.”

Perhaps.

But workplace operations are not transformed by impressive demonstrations.

They are transformed when a facilities team can reduce downtime, close service gaps faster, improve compliance, eliminate repetitive administration, or make better decisions with the same resources.

That is a much higher—and more useful—standard.

After separating the genuinely useful applications from the AI theatre, a pattern is becoming clear:

The AI applications creating the most practical value today are not trying to run the workplace. They are solving small, repetitive, data-heavy problems extremely well.

Five use cases stand out:

1. Predictive maintenance
2. Smart scheduling and resource optimisation
3. AI-assisted vendor and SLA monitoring
4. Natural-language search across policies, SOPs and operational knowledge
5. Automated first-draft reporting and management insights

None of these replaces an experienced facilities professional.

That is precisely why they work.

1. Predictive Maintenance: Moving from “Fix It” to “Intervene Before Failure”

Traditional facilities management is still heavily reactive.

Something breaks.

Someone raises a ticket.

A technician investigates.

A vendor is called.

Parts are ordered.

The workplace absorbs the disruption.

Preventive maintenance improved this model by introducing scheduled inspections and servicing. But preventive maintenance has its own weakness:

A calendar does not know the actual condition of an asset.

AI-enabled predictive maintenance attempts to close that gap.

By analysing equipment data—including operating patterns, alarms, temperature, vibration, energy consumption, runtime, fault history and maintenance records—AI models can identify patterns associated with impending failure.

The objective isn't simply to predict failure.

The real operational objective is:

Intervene at the economically optimal moment.

For example, an HVAC system showing abnormal performance may not have failed yet. But if its operating signature increasingly resembles previous failure events, the facilities team can investigate before the failure becomes a workplace incident.

Where it creates value

• Reduced unplanned downtime
• Better maintenance prioritisation
• Fewer emergency call-outs
• Improved asset availability
• Better spare-parts planning
• Longer useful asset life
• Reduced disruption to employees

The practitioner reality

This is where many AI presentations become unrealistic.

AI cannot manufacture good asset data.

If your BMS data is incomplete, sensors are unreliable, asset registers are inaccurate and maintenance history is inconsistent, an AI model will not magically solve the problem.

Garbage in still produces garbage out—just faster and with a sophisticated dashboard.

The practical starting point

Don't begin with every asset in the building.

Start with a small number of critical assets where failure has a measurable operational or financial consequence.

For example:

Critical asset → reliable data → known failure pattern → defined intervention → measurable outcome.

That is a much stronger AI business case than “AI-powered predictive maintenance across the entire portfolio.”

2. Smart Scheduling: AI for the Constant Puzzle of People, Space and Resources

Facilities teams schedule everything.

Meeting rooms.

Cleaning.

Security coverage.

Engineering resources.

Transport.

Preventive maintenance.

Moves and relocations.

Events.

Visitor services.

Workspace capacity.

And almost all of these schedules interact with one another.

That makes workplace operations an optimisation problem.

AI can help identify patterns in demand and recommend how resources should be allocated.

For example, workplace utilisation data may show that certain floors are heavily occupied on particular days while other areas remain significantly underused.

Instead of simply reporting utilisation, an AI system can begin answering:

“Given expected demand, available space and operational constraints, what is the most efficient configuration?”

The same principle applies to service scheduling.

Cleaning resources can be aligned with occupancy patterns rather than fixed assumptions.

Maintenance activities can be scheduled around workplace demand.

Meeting-room resources can be adjusted according to actual usage.

Transport demand can be forecast against employee attendance patterns.

The value isn't “automation.”

It is better utilisation of finite resources.

That distinction matters.

A facilities manager does not necessarily need an AI system to make every scheduling decision automatically.

A better model is:

AI recommends. Operations validates. The system executes.

Where it creates value

• Improved space utilisation
• Reduced unnecessary service coverage
• Better workforce allocation
• Fewer scheduling conflicts
• Reduced overtime
• Better employee experience
• Improved responsiveness during demand peaks

The practitioner reality

Workplace scheduling is rarely a pure mathematical problem.

There are contractual obligations, safety requirements, employee preferences, union or labour considerations, business-critical events and exceptions that an algorithm may not understand.

So the winning model is not:

Human vs AI.

It is:

Human judgement + AI optimisation.

3. AI-Assisted Vendor & SLA Monitoring: Turning Contracts into Operational Intelligence

This may be one of the most commercially interesting applications of AI in facilities management.

Most organisations have enormous amounts of vendor information.

Contracts.

SLAs.

KPIs.

Scope documents.

Service reports.

Invoices.

Helpdesk tickets.

Inspection reports.

Escalations.

Email correspondence.

Compliance documents.

Performance reviews.

Yet the information often exists in separate systems and formats.

The result?

Facilities teams spend significant time manually checking whether vendors actually delivered what the contract requires.

AI can help connect these dots.

Imagine a system capable of comparing:

Contract requirement → SLA → ticket → service completion → user feedback → inspection result → invoice.

Suddenly, vendor management becomes much more evidence-driven.

Consider a simple example

A cleaning contract specifies:

• defined cleaning frequencies
• response times
• quality standards
• staffing levels
• consumable requirements
• escalation procedures

AI can assist in monitoring operational records against those requirements and flagging potential exceptions.

Not:

“Vendor is failing.”

But:

“These 17 service records appear inconsistent with the contractual SLA and require human validation.”

That difference is extremely important.

Where it creates value

• Faster SLA reviews
• Early identification of recurring service failures
• Better contract governance
• Reduced manual reconciliation
• Stronger vendor negotiations
• Improved invoice validation
• Evidence-based performance discussions

The practitioner reality

AI should flag exceptions—not become the judge and jury.

A system may identify a potential SLA breach, but context matters.

Was the building closed?

Was the service intentionally rescheduled?

Was access unavailable?

Was there an emergency?

Was an agreed deviation approved?

The facilities professional still owns the decision.

AI simply reduces the amount of searching required to reach it.

4. Natural-Language Search: Making the Organisation's Operational Knowledge Usable

This is perhaps the least glamorous AI application—and potentially one of the most useful.

Facilities organisations accumulate enormous amounts of institutional knowledge.

SOPs.

Policies.

Emergency procedures.

Fire plans.

Business continuity documents.

Security protocols.

Lease documents.

Vendor contracts.

Risk assessments.

Maintenance manuals.

Escalation matrices.

Compliance requirements.

Induction materials.

The problem is rarely the absence of information.

The problem is finding the right information at the right moment.

Traditional document repositories force people to know:

• where the document is stored
• what it is called
• which version is current
• which section contains the answer

Natural-language AI changes the interaction.

Instead of searching for a document, an employee can ask:

“What is the escalation process for an after-hours security incident?”

Or:

“What is the SLA for an emergency HVAC call?”

Or:

“Who needs to be informed if the fire alarm system is unavailable?”

A properly governed AI knowledge system can retrieve the relevant information and provide an answer based on approved organisational documents.

This has a powerful operational implication.

It converts organisational knowledge from:

documents people have to find

into:

knowledge people can access.

Where it creates value

• Faster employee support
• Reduced repetitive queries
• Better SOP accessibility
• Faster incident response
• Improved onboarding
• Reduced dependency on individual knowledge holders
• Better consistency of information

The practitioner reality

This use case has one non-negotiable requirement:

Document governance.

If the AI is searching five versions of the same SOP, nobody knows which one is authoritative.

Therefore:

AI knowledge management begins with information management.

Version control, ownership, permissions, document lifecycle and source hierarchy matter as much as the AI itself.

And for safety-critical or compliance-sensitive information, the system should make the source visible rather than presenting an unexplained answer.

5. Automated First-Draft Reporting: Eliminating the Administrative Tax on Facilities Leaders

This is where many facilities professionals can see immediate productivity gains.

Facilities generates reports constantly.

Monthly operations reports.

SLA dashboards.

Energy reports.

Incident summaries.

Helpdesk analysis.

Vendor performance reports.

Occupancy updates.

Management reviews.

Audit summaries.

Executive presentations.

Much of the work involves collecting information from multiple sources, identifying changes and converting data into understandable narrative.

AI is particularly good at producing a first draft.

It can help identify:

• significant deviations
• recurring incidents
• SLA deterioration
• unusual consumption patterns
• service trends
• major operational events
• emerging risks
• areas requiring management attention

It can then convert structured information into an initial management narrative.

For example:

Data:
Helpdesk response time increased 18%.

AI-assisted first draft:
“Average response time increased during the reporting period, primarily coinciding with higher service demand and increased engineering-related tickets. The trend should be monitored, with particular focus on recurring HVAC-related requests.”

The facilities leader then checks the facts, adds context, challenges the interpretation and makes the final judgement.

That is the important distinction:

AI writes the first draft.

The practitioner owns the final story.

Where it creates value

• Less time spent compiling reports
• Faster management reporting
• Consistent reporting structures
• Faster identification of anomalies
• Better executive communication
• More time for analysis and decision-making

And this may be one of the most underrated benefits of AI.

The goal isn't necessarily to reduce headcount.

The goal is to move skilled people away from administrative production and towards operational judgement.

What These Five Use Cases Have in Common

At first glance, predictive maintenance, scheduling, vendor monitoring, knowledge search and reporting appear unrelated.

They aren't.

They share the same underlying characteristics.

1. The task is narrow

AI performs better when the problem is clearly defined.

2. The data already exists

The organisation doesn't need to invent an entirely new data universe.

3. The output is measurable

Downtime.

Response time.

Cost.

Utilisation.

SLA compliance.

Hours saved.

4. Human judgement remains important

AI supports the decision rather than pretending to be the decision-maker.

5. The workflow already exists

AI is inserted into an existing operational process rather than creating an entirely artificial one.

This is an important lesson for workplace leaders.

The strongest AI use cases are often boring.

And boring is good.

Because boring problems are usually easier to measure.

Where AI Is Still Overhyped

There is another side to the conversation.

Not every workplace problem needs AI.

A poorly designed process does not become a good process because an AI chatbot has been added to it.

Likewise:

AI cannot compensate for weak governance.

If your service catalogue is unclear, SLAs are poorly written, asset data is unreliable, vendors are inadequately managed and responsibilities are ambiguous, introducing AI may simply automate the confusion.

I would be particularly cautious about claims such as:

• “Fully autonomous facilities management”
• “AI replaces the facilities manager”
• “Zero-touch workplace operations”
• “AI predicts everything”
• “One platform eliminates all FM complexity”

Real workplaces contain too many variables, exceptions and human consequences for such claims to be taken literally.

The more credible model is:

AI handles the repetitive cognitive workload. Humans handle accountability, judgement and exceptions.

The Real AI Opportunity in Workplace Operations

The biggest opportunity may not be replacing facilities jobs.

It may be eliminating the administrative tax attached to facilities jobs.

Think about the amount of time experienced professionals spend:

• searching for information
• reconciling spreadsheets
• chasing vendors
• checking tickets
• preparing reports
• compiling evidence
• comparing versions
• scheduling resources
• responding to repetitive questions

These activities consume professional capacity without necessarily requiring professional judgement.

AI can increasingly absorb portions of that workload.

That creates a different proposition:

From

People managing information

To

People managing decisions.

And that is a much more compelling vision for the future of workplace operations.

A Practical AI Adoption Framework for Facilities Leaders

Before approving an AI project, ask five questions.

01 — Is the problem repetitive?

If it happens once a year, AI may not be the answer.

02 — Is the process already reasonably stable?

Don't automate a process that nobody understands.

03 — Is there usable data?

If not, fix the data foundation first.

04 — Can the outcome be measured?

Define the baseline before deploying the technology.

05 — Where does human accountability remain?

This should be explicit—not assumed.

A simple scoring model can help:

Business impact × Frequency × Data availability × Measurability ÷ Implementation complexity

The highest-scoring opportunities should generally be investigated first.

What I Would Pilot First

If I were starting an AI programme in workplace operations today, I would not begin with an ambitious “AI workplace” transformation.

I would begin with three relatively contained pilots:

Pilot 1 — Operational Knowledge Assistant

Connect approved SOPs, policies, service procedures and operational documents.

Measure: time-to-answer and reduction in repetitive queries.

Pilot 2 — Vendor Performance Intelligence

Analyse tickets, SLA records, service reports and contract requirements.

Measure: exceptions identified, review time and recurring failures.

Pilot 3 — Management Reporting Copilot

Convert validated operational data into first-draft monthly reporting.

Measure: reporting preparation hours saved and quality/consistency of management insight.

Once these demonstrate value, move into more complex areas such as predictive maintenance, workplace optimisation and demand forecasting.

The Bottom Line

The AI revolution in workplace operations probably won't arrive as one giant system that suddenly runs the building.

It is more likely to arrive one workflow at a time.

One maintenance decision.

One scheduling optimisation.

One vendor exception.

One policy question.

One management report.

And eventually, thousands of small decisions become faster, better informed and less administratively expensive.

That is where I believe the practical opportunity lies.

Don't ask, “Where can we put AI?”

Ask:

“Which repetitive operational decision is consuming valuable human capacity—and can AI make that decision faster, better or more consistently?”

That is a much better starting point.

Because the future of workplace operations isn't necessarily AI replacing the workplace professional.

It is the workplace professional operating with significantly more intelligence, context and capacity.

And that future is already beginning—not in the headlines, but in the workflows.