Singapore MEI Regime 2026: How Building Owners Can Build an AI-Ready Energy Monitoring and Improvement Plan

A practical workflow for turning building energy data into better decisions, audit evidence and measurable improvement actions.

For owners and facility managers of large, energy-intensive buildings, energy management is becoming both a compliance consideration and an operational cost-control priority. Singapore’s Building and Construction Authority (BCA) Mandatory Energy Improvement (MEI) regime applies to energy-intensive buildings with a gross floor area of at least 5,000 square metres. BCA states that the regime is intended to improve energy performance and reduce operational costs.

The practical challenge is not simply installing more sensors or producing another dashboard. Building owners need a repeatable process that connects reliable data, engineering review, energy audits, improvement actions and evidence records. Artificial intelligence can support this process, but it should assist—not replace—qualified professionals, site verification and regulatory requirements.

What an AI-ready MEI workflow should achieve

An AI-ready energy programme should help your team answer five basic questions:

• How much energy is the building using, and when?
• Which systems, zones or operating conditions are driving consumption?
• Is the data complete, accurate and suitable for comparison?
• What operational or equipment issues require investigation?
• Can the building demonstrate what was measured, decided, implemented and reviewed?

These questions provide a useful structure for preparing information for an energy audit and managing improvement work after the audit. They also help prevent a common mistake: treating AI as a standalone product instead of embedding it into facilities-management workflows.

1. Confirm the building scope and assign ownership

Start by confirming whether the building falls within the MEI regime and what current BCA requirements, submission processes and timelines apply. The BCA MEI webpage should be treated as the primary reference for the latest regime information. Owners should also clarify responsibilities among the building owner, managing agent, facility manager, energy consultant, maintenance contractors and internal finance or sustainability teams.

Create a simple responsibility matrix covering data collection, meter access, equipment records, audit coordination, improvement-plan approvals and evidence retention. Without clear ownership, energy information can be spread across utility bills, building-management systems, spreadsheets, contractor reports and email threads.

2. Establish a trustworthy energy baseline

A baseline is only useful when the underlying data is consistent. Begin by gathering available whole-building energy data, utility records, operating hours, occupancy or usage patterns, major equipment schedules and relevant weather or production information where appropriate.

Before using analytics, check for:

• Missing or duplicated readings
• Changes in meter configuration or tariff records
• Inconsistent time zones, intervals or units
• Unexpected resets or sensor replacements
• Periods when equipment was offline or the building was unusually occupied

Document the source, time period, unit, frequency and responsible owner for each dataset. A digital data dictionary can make this information easier to maintain. It should identify what each meter measures, where it is located, how often it reports and how its readings relate to the building’s energy account.

3. Use sub-metering to improve visibility

Whole-building consumption can show that performance has changed, but it may not explain why. Sub-metering can provide better visibility into major loads or operational areas, subject to site conditions, technical feasibility and the requirements of the relevant assessment.

Potential priorities may include central plant systems, air-conditioning and mechanical ventilation, lighting, lifts, tenant areas, warehouse processes, kitchens or other significant loads. The right configuration depends on the building and its operating profile. More meters are not automatically better if the data is unreliable or no one is responsible for reviewing it.

For each meter, record its location, measurement boundary, communication method, calibration or maintenance information where available, and the person responsible for resolving data issues. This creates a stronger foundation for both analysis and audit discussions.

4. Connect equipment performance with energy data

Energy readings become more useful when they are linked to equipment and operating context. Build or update an equipment register covering major assets, system relationships, control points, maintenance status and known operating constraints.

For example, an increase in electricity use may be related to longer operating hours, control settings, fouled equipment, abnormal cycling, changes in occupancy or a faulty sensor. AI-assisted analytics can highlight patterns and prioritise investigation, but the cause still needs to be checked by a competent facilities or engineering team.

Useful data connections can include:

• Energy meters and submeters
• Building-management or automation systems
• Equipment runtime and alarm histories
• Temperature, humidity and indoor environmental readings where relevant
• Maintenance work orders and inspection records
• Occupancy, production or operating schedules

5. Introduce anomaly detection carefully

Rule-based alerts are often a practical starting point. Examples include a meter that stops reporting, energy use outside scheduled hours, a sudden change from an established pattern, or simultaneous heating and cooling indications where the system data supports that analysis.

Machine-learning models can later help identify less obvious patterns, but they require enough clean historical data and proper validation. Set alert thresholds that reflect the building’s operating reality. Too many false alarms will cause teams to ignore the system; too few alerts may hide important issues.

Every alert should have a workflow: who receives it, how quickly it is reviewed, what evidence is checked, what action is taken and how the outcome is recorded. This turns analytics into operational improvement rather than another unattended dashboard.

6. Prepare for the energy audit with an evidence trail

Digital records can reduce the time needed to assemble information for an energy audit. Maintain a controlled evidence library containing meter data, data-quality checks, equipment registers, operating schedules, maintenance records, previous studies, identified issues, decisions and completed actions.

Use version control and timestamps where possible. Keep a clear distinction between raw readings, processed datasets, analyst observations, engineering conclusions and approved actions. This makes it easier to explain how a recommendation was formed and whether an improvement was implemented.

AI can assist with document classification, missing-record checks, summary generation and search. However, generated summaries should be reviewed by an accountable person before they are used for formal reporting or technical decisions.

7. Convert findings into a governed improvement plan

An improvement plan should be specific enough to manage. For each proposed action, record the issue, proposed measure, affected system, expected operational benefit, dependencies, owner, target date, verification method and status.

Actions may include operational adjustments, control improvements, maintenance interventions, equipment upgrades or changes to scheduling. Prioritise measures using criteria that matter to the organisation, such as energy impact, implementation complexity, operational risk, tenant disruption, capital requirements and maintenance needs.

Do not treat an AI recommendation as an automatic instruction. Facilities and engineering teams should validate the operating context, safety implications, comfort requirements, system interactions and any applicable technical or regulatory considerations before implementation.

8. Build a review cycle instead of a one-off project

MEI preparation should not end when an audit report is filed. Establish a monthly or quarterly review cycle for data quality, energy trends, open anomalies, completed actions and unresolved engineering issues. Use the review to update the baseline when operating conditions materially change and to confirm whether improvement measures are delivering the intended result.

BCA’s resources on AI for the Built Environment and Smart Facilities Management point towards practical uses such as workflow automation, system optimisation, data analysis and predictive maintenance. For many organisations, the most valuable first step is not a complex AI model. It is a reliable digital workflow that makes information visible, assigns responsibility and preserves the decision trail.

Practical starting checklist

• Confirm the building’s potential MEI applicability and current BCA requirements.
• Assign owners for energy data, audit coordination and improvement actions.
• Catalogue utility accounts, meters, submeters, systems and operating records.
• Run basic data-quality checks before building analytics.
• Link major energy loads to equipment and maintenance information.
• Start with actionable alerts and a defined response process.
• Create a controlled evidence library for audit and management review.
• Track every improvement action through approval, implementation and verification.

For Singapore building owners, the strongest MEI preparation is a combination of sound engineering, disciplined facilities management and well-governed digital tools. AI can help identify patterns, automate repetitive work and improve visibility, but its value depends on the quality of the data and the people who act on it.

ISS supports organisations exploring engineering, facility-management and AI automation requirements. Contact ISS to discuss how a practical energy-data and workflow approach could fit your building operations.