Loading...
CBRE's AI is learning your building. JLL's Falcon covers 1.1B sq ft. When the FM contract ends, the model stays with them. Your data portability clause probably doesn't exist.
Why AI is now standard in FM — and what governance gaps arrived with it
Your FM contract renewal is approaching. The renewal conversation will be framed as service continuation. Three unresolved questions are embedded in it that the standard renewal process will not surface.
Who owns the operational intelligence built on your building data over the contract term? If you have not established data portability rights, the model trained on your data stays with your FM provider at contract end. This is a negotiating question. The renewal is the moment to resolve it.
Does your FM data processing agreement satisfy GDPR Article 28 for AI-processed occupancy data? The DPA your legal team approved when the FM contract was executed was almost certainly not written against AI processing requirements. The Article 28 gap is real and requires a formal amendment.
What is your FM provider's liability for an AI maintenance recommendation that precedes an adverse outcome? The limitation of liability clause in your FM agreement was written for service delivery failures. It does not clearly allocate accountability for AI recommendation accuracy.
Lowest friction. Data ownership, DPA compliance, and AI liability gaps persist. Renegotiation at contract end is more expensive and adversarial than renegotiation at renewal.
Highest-leverage moment. FM providers at CBRE/JLL scale will negotiate for significant accounts. Requires raising the question before the contract is signed.
Maximum control. Platforms: Willow, Mapped, SpaceIQ. Significant internal investment. Right for large, stable portfolios with internal real estate technology capability.
Appropriate if current exposure is unclear. Produces the negotiating position for the renewal conversation. Delays renewal decision by 60—90 days.
Five years of building-specific calibration does not transfer under standard contract terms. Successor FM provider starts cold. Switching cost includes operational accuracy gap during transition, not just procurement friction.
Occupancy sensor, badge access, and desk utilization data are employee behavioral data under GDPR. FM DPAs predating AI processing of this data do not satisfy Article 28 requirements for sub-processor disclosure, processing purpose specification, or return-or-deletion provisions.
Standard FM liability caps cover service delivery failures. Accountability for AI-generated maintenance recommendations that precede adverse outcomes is not clearly allocated. The enterprise approved the deferral; the FM provider generated the recommendation; neither framework assigns liability cleanly.
Buildings that shifted from 95% Monday-Friday occupancy to 60% hybrid occupancy between 2020 and 2022 carry AI models trained on distributions that no longer reflect actual use. FM providers without formal retraining schedules tied to occupancy changes deliver optimization against historical patterns, not current ones.
Landlords whose FM AI processes occupancy data from multi-tenant buildings are collecting employee behavioral data belonging to their tenants' employees. Current lease templates almost universally do not address this. GDPR exposure for landlords in EU jurisdictions who have not established a data processing chain with tenants is real and unaddressed.
The AI operational case is real — 15—30% energy savings in mature deployments. The data governance case — portability at contract end, GDPR Article 28 compliance, AI recommendation accountability — has not been built at the same pace. The renewal is the moment to close that gap.