AI-Ready CRE Begins With Data Governance

Corporate real estate has no shortage of AI ambition. JLL's 2025 global real-estate technology research reported that 92% of surveyed CRE organisations were piloting AI or planning pilots.[9] Yet JLL's 2026 occupancy benchmark reported that 70% had not begun AI implementation in occupancy planning, despite 73% reporting data-governance programmes.[1] CBRE found that 55% cited data quality or lack of expertise as AI challenges.[2]

Analysis

These findings measure different populations and activities, but together they expose the gap between intent and operational readiness.

A governance programme can exist on paper while location identifiers, capacity records, sensor feeds, HR populations and booking taxonomies remain inconsistent.

Osmos Global analysis: the first AI investment should be a decision-grade data product. For a use case such as demand forecasting, the team must define the decision, prediction horizon, permitted data, master identifiers, freshness requirements, error tolerance and human override. Without that design, a model may produce precise-looking output that cannot be trusted.

Data governance also has an ethical and employment dimension. Workplace signals can become personal data when linked to identifiable employees. Purpose limitation, transparency, retention, access controls and jurisdiction-specific consultation should be established before scale.

A pragmatic roadmap starts with one reversible use case: room no-show prediction, service-load forecasting or energy scheduling.

Measure baseline performance, test the model against simple rules, monitor drift and record avoided cost or service improvement. AI value in CRE will not come from the number of pilots. It will come from the number of governed decisions that improve.

Decision lens

Recommended action. Fund a governed decision-grade dataset and one measurable use case before scaling AI pilots.

---

Evidence basis: JLL technology survey 2025; JLL occupancy benchmark 2026; CBRE global workplace 2026. Public-page limitations apply where full reports are restricted. Rights: Original Osmos Global analysis; no third-party visual or protected table reproduced. Confirm licence status before publication.