From Smart Buildings to Decision-Ready Operations

Central proposition

A building becomes smart only when it can turn trustworthy signals into governed action and measurable value—not when it merely contains connected technology.

Research context

Buildings increasingly contain networked controls, submeters, access systems, occupancy sensors, work-order platforms, booking tools and cloud analytics. Each can improve visibility. Together, they can also create new dependencies across operational technology, IT, service providers and occupants.

The investment case is often framed through energy, predictive maintenance, space optimisation, experience and automation. The International Energy Agency estimated that existing AI-led building interventions could save around 300 TWh of electricity globally if scaled, while identifying fragmented ownership, limited digitalisation and inadequate incentives as barriers.[5] This is a modelled global potential, not a guaranteed saving for an individual building.

Facilities teams also report growing AI use. Johnson Controls' 2026 public summary stated that 67% of surveyed facilities teams were using AI and 61% planned to expand it; 22% identified energy optimisation as AI's greatest long-term potential and 20% cited data quality and integration as the top barrier.[6] Detailed sampling information remains in the downloadable report, so the figures are retained with a methodology caveat.

The operational challenge is therefore two-sided: organisations must create a trustworthy data and control foundation while avoiding architecture that is too complex to maintain, secure or recover.

Evidence review

Adoption is high, but evidence strength varies ASHB's public summary reports widespread smart-system use and substantial investment.[1] Because the full report is restricted, Osmos Global does not infer the exact representativeness of the percentages. The findings are used as a North American adoption signal.

JLL's CRE technology study reported 92% piloting or planning AI.[2] “Piloting or planning” is an intent measure and must not be presented as scaled deployment or realised value. Johnson Controls' facilities result uses a different definition of “using AI,” reinforcing the need to retain each survey's terminology.

Data governance activity does not equal usable data JLL's occupancy benchmark reported that 73% had data-governance programmes while 70% had not begun AI implementation in occupancy planning.[3] The coexistence of these findings suggests that formal governance activity can precede operational readiness.

IFMA's digital-thread guidance identifies fragmented lifecycle information as a source of process and cost inefficiency and calls for reliable information about assets, spaces, systems and people.[7] It is industry guidance, not a measured performance study, but it clarifies the foundation required for scalable decisions.

AI opportunity is material but context dependent The IEA's 300 TWh estimate demonstrates global technical potential from scaled building interventions.[5] Local outcomes depend on climate, operating schedules, controls, asset condition, tariffs and incentives. An organisation should not apply the global estimate as a portfolio saving percentage.

CBRE's finding that 55% identified data quality or expertise challenges and Johnson Controls' 20% data-integration barrier signal both point to implementation constraints.[4][6] The denominators and survey designs differ, so the percentages are not directly comparable.

Findings

1. Connected-system prevalence is not digital maturity

Technology presence does not prove interoperability, workflow adoption, resilience or outcome value.

2. AI intent is ahead of scaled operational use

Planning and pilots are widespread, while occupancy-planning implementation and data readiness remain constrained.

3. Governance programmes need operational tests

Policies and councils matter, but readiness depends on usable identifiers, quality rules, access and accountable data products.

4. Interoperability is a lifecycle and procurement requirement

Open interfaces, export rights, identity, support periods and recovery should be defined before purchase.

5. Cybersecurity and recoverability are operational outcomes

Cloud and connected controls change access, update, incident and fallback responsibilities.

6. AI value must be measured against a baseline

Forecast accuracy, avoided failure, energy reduction and service improvement require pre-agreed methods.

7. Human capability remains part of the control system

Operators need the authority, competence and trust to interpret, challenge and override automated recommendations.

Osmos Global analysis

The seven-layer Decision-Ready Building Framework Layer Core question Minimum control Proof of readiness Purpose Which decision or outcome matters?

Use-case owner, baseline and boundary Decision and success measure approved Foundation Are assets, spaces and signals trustworthy?

Common IDs, quality rules, metadata and history Critical data passes acceptance tests Integration Can systems exchange usable context?

Interfaces, semantics, time sync and export rights End-to-end flow works under change Intelligence How is insight produced and challenged?

Model scope, error tolerance, validation and override Performance beats a simple baseline Action Who responds and closes the loop?

Workflow, authority, escalation and evidence Action is completed and recorded Assurance Is the system secure, recoverable and lawful?

Access, updates, privacy, retention and fallback Recovery and control tests pass Value Did the outcome improve? Benefit method, counterfactual and review cadence Verified operational/business result The framework reverses technology-led planning. Instead of asking what the platform can do, it asks which decision needs to improve and what foundation is necessary to make that improvement trustworthy.

Osmos Global infers that many building programmes stall between intelligence and action. A dashboard identifies an anomaly, but nobody owns the response; a model predicts demand, but space and service decisions remain unchanged.

The action layer should therefore be designed at the same time as the analytic layer.

Assurance must include degraded operation. Leaders should know how critical services continue when connectivity, a cloud service, an integration or a model becomes unavailable. Manual fallback is credible only when staff retain access, competence and current procedures.

Value should be verified conservatively. Energy must be normalised for weather and operating conditions. Maintenance benefits should distinguish avoided events from postponed work. Workplace benefits should remain separate from attendance and utilisation. A use case that cannot define its outcome should not scale.

Recommendations 1. Select use cases through operational value and risk, not vendor feature lists. 2. Create common asset, space and service identifiers before portfolio integration. 3. Contract for interfaces, data export, cybersecurity duties, support periods and recovery. 4. Validate AI against a simple rule or existing process before claiming improvement. 5. Design the response workflow, escalation and closure evidence alongside analytics. 6. Test degraded operation and recovery for critical cloud and connected controls. 7. Train operators to interpret, challenge and override automated recommendations. 8. Scale only after the outcome and benefit method have been independently reviewed.

Risks, limitations and unresolved questions

• Vendor-sponsored surveys may use broad definitions of AI and smart-building adoption. • Restricted reports limit review of sampling, weighting and question wording. • Modelled global energy potential cannot be applied directly to an individual portfolio. • Integration can increase cyber exposure and vendor dependency. • Algorithmic performance may drift as occupancy, equipment and operating conditions change. • Evidence linking building AI to long-term enterprise outcomes remains uneven.

Executive readiness checklist Can leadership confirm… Yes/No The use case names a decision, owner and baseline?

Critical identifiers and data-quality rules are accepted?

Interfaces and export rights are contractually protected?

Model accuracy and error tolerance are defined?

The response workflow and escalation are tested?

Cybersecurity, privacy and fallback controls are approved?

Operators can challenge or override automation?

Value is verified before portfolio scale?

References

[1] Klopotowska, M., ASHB, & Harbor Research. (2025, 9 December). 2025 Smart Building Trends & Technology Adoption. https://www.ashb.com/new-research-on-smart-building-trends-technology-adoption-2/ [2] JLL Research. (2025, 27 October). Reality check: The true pace and payoffs of AI adoption in corporate real estate. https://www.jll.com/en-us/insights/global-real-estate-cre-technology-survey [3] Xie, W., & Holmes, M. (2026, 19 May). Global Occupancy Planning Benchmark Report 2026. JLL. https://www.jll.com/en-us/insights/occupancy-benchmark-report [4] CBRE Workplace & Occupancy Research. (2026, 20 January). 2026 Global Workplace & Occupancy Insights. https://www.cbre.com/insights/reports/2026-global-workplace-and-occupancy-insights [5] International Energy Agency. (2025, 10 April). Energy and AI. https://www.iea.org/reports/energy-and-ai [6] Johnson Controls. (2026, 23 July). Top 3 insights from the 2026 AI & Digitalization in Facilities Management Report, FM Edition. https://www.johnsoncontrols.com/building-insights/feature-story/top-3-insights-2026-ai-survey-facilities-managers [7] Ritter, T., IFMA IT Community, Autodesk, & Clarke, S. (2025, 3 June). Advancing Facilities Management with a Digital Thread. https://knowledgelibrary.ifma.org/advancing-facilities-management-with-a-digital-thread/ [8] Buildings Breakthrough, WorldGBC, & GlobalABC. (2026). Near-Zero Emission and Resilient Buildings. https://worldgbc.org/article/landmark-report-launched-to-make-near-zero-emission-and-resilient-buildings-the-global-norm-by-2030/ [9] Trane. (2026, 11 August). The Future Ready Building Report. https://knowledgelibrary.ifma.org/the-future-ready-building-report-8-business-drivers-advancing-building-technology/ [10] International Energy Agency. (2025, 20 November). Energy Efficiency 2025. https://www.iea.org/reports/energy-efficiency-2025 Editorial note This publication presents original Osmos Global analysis based on publicly available and cited research. Source findings and Osmos Global interpretations are distinguished throughout. Third-party trademarks and source materials remain the property of their respective owners. This publication is provided for research and professional-information purposes and does not constitute legal, financial, investment or technical advice.