Why Your AI Building Platform Needs IPMVP Verification — Not Just AI Claims

The Procurement Question Nobody Can Answer Yet

VTS launched Asset Intelligence on April 1, 2026. The feature delivers expert-in-the-loop verification for lease abstraction across more than 600,000 leases — trained AI extracts terms, human experts confirm ambiguous clauses before they enter the database. The category marker: verified output. Not AI-generated text. Verified output.

That distinction is now the defining quality signal in enterprise CRE AI procurement. And it raises an uncomfortable question for the building operations side of the house: who is providing expert-in-the-loop verification for HVAC fault detection, energy baseline calculations, and predictive maintenance recommendations?

The honest answer, for every vendor currently in the market except AISB, is: nobody.

Ask the Agent: Query your building's IPMVP baseline and verification readiness →

What IPMVP Verification Actually Means

IPMVP — the International Performance Measurement and Verification Protocol — was developed by the Efficiency Valuation Organization (EVO) to provide a standardized methodology for quantifying the savings resulting from energy conservation measures. It is the M&V standard used in ESCO contracts, green bonds, PACE financing, and ENERGY STAR certification. It is not a marketing framework. It is an auditable, contractually enforceable methodology.

IPMVP defines four options for measuring performance:

• Option A — Retrofit Isolation (Key Parameter Measurement): Measures a subset of parameters with engineering estimates for the rest. Appropriate for simple, well-understood retrofits (e.g., lighting). Lowest cost, lowest verification confidence.
• Option B — Retrofit Isolation (All Parameter Measurement): Measures all parameters affecting energy use. Appropriate for complex retrofits where all performance variables matter. Higher cost, higher confidence.
• Option C — Whole Facility: Measures whole-facility energy use before and after, adjusting for independent variables (occupancy, weather, production). Used for multi-measure retrofits where interactive effects are significant. Gold standard for AI-HVAC optimization claims.
• Option D — Calibrated Simulation: Uses a calibrated building energy simulation to determine savings. Required when baseline data is unavailable or incomplete.

When an AI building platform tells you it reduced your energy costs by 22%, the IPMVP question is: which option did you use, what was your baseline period, how did you normalize for weather and occupancy, and what independent variables did you control for? Without answers to those questions, you have a software-generated estimate, not a verified savings figure. Those are different things — especially when your sustainability report, PACE lender, or green lease covenant requires the latter.

The Verification Gap Across Current Platforms

A systematic review of the five most-deployed AI building intelligence platforms reveals a uniform gap: none explicitly market IPMVP M&V methodology for building operations. The table below documents what each platform provides, based on publicly available documentation and feature announcements as of April 2026:

Why the Savings Gap Matters Now

Three converging pressures are making IPMVP verification a procurement requirement rather than a nice-to-have.

Green lease covenants are tightening. As Class A office tenants negotiate green lease clauses with landlords, both parties need auditable performance data. A landlord claiming "our AI system reduced energy use by 18%" cannot enforce a green lease clause without documented baseline, methodology, and normalized comparison. IPMVP Option C — whole-facility measurement with regression-based normalization — is the contractual anchor. Without it, neither party can prove compliance or performance.

LL97 and similar compliance frameworks require verified baselines. New York's Local Law 97 fines buildings that exceed carbon intensity limits. Demonstrating that an AI building system reduced emissions below a threshold requires a verifiable baseline from the period before the measure was implemented, not a software-generated dashboard number. The same applies to Toronto's Building Emissions Performance Standard, UK MEES requirements, and the EU's CSRD reporting framework.

PACE lenders and ESG bond issuers require third-party M&V. Property Assessed Clean Energy financing often requires M&V reports aligned to IPMVP. Green bond proceeds allocated to building AI deployments face the same scrutiny as physical retrofits. If the AI system cannot produce IPMVP-compliant savings documentation, the financing structure is at risk.

What AISB's IPMVP-Aligned Verification Layer Includes

AISB's verification methodology is built around three components that map directly to IPMVP protocol requirements:

Baseline establishment: Pre-intervention energy consumption is documented using actual meter data, normalized for heating degree days, cooling degree days, and occupancy patterns. The baseline period is explicitly defined and archived with provenance anchors — not retroactively adjusted.

Boundary definition and parameter selection: For each building and each energy conservation measure, AISB defines which systems are in scope, which parameters are measured directly, and which are estimated with disclosed uncertainty ranges. This maps to IPMVP's distinction between Option A (partial measurement) and Option B (full measurement) — and AISB documents which option applies to each measure.

Continuous M&V reporting: Savings calculations are updated monthly using the IPMVP reporting period framework. Weather normalization uses TMY3 data or site-specific historical weather. Occupancy normalization uses badge data or building calendar integrations where available. All adjustments are disclosed in the verification report.

The result is a savings figure that can be presented to a lender, regulator, or auditor with a documented methodology — not a marketing percentage.

The Expert-in-the-Loop Structure

VTS's expert-in-the-loop model for lease abstraction is well-designed: AI extracts terms, human experts review flagged ambiguities, and the verified output enters the database. The expert catches what the model misses — ambiguous renewal clauses, non-standard CAM definitions, jurisdiction-specific carve-outs.

AISB applies the same logic to building operations intelligence. The agent stack identifies faults, calculates savings opportunities, and flags anomalies. An IPMVP-trained reviewer examines outliers — unexpected consumption spikes, sensor data gaps, apparent savings that don't reconcile with meter data. The verified figure enters the building intelligence record.

This is what distinguishes AISB from platforms that report AI-generated savings figures directly to operator dashboards: the number on the dashboard has passed a human verification gate before it appears.

The Procurement Implication

The next wave of AI building platform procurement is going to ask a simple question: can you produce IPMVP-compliant M&V documentation for your savings claims? Enterprise FM directors evaluating vendors for green lease compliance, LL97 reporting, or ESG financing will require this documentation. Mid-market operators who have been burned by AI pilots that showed 20% energy reductions in dashboards while their utility bills stayed flat will demand it.

IPMVP verification is not a technical differentiator. It is a trust differentiator. And trust, in the current AI building market, is the scarce commodity.

Ready to establish your IPMVP baseline?
Ask the AISB agent for an IPMVP Option C baseline framework for your building →

Related reading: Why no platform connects lease CAM terms, LL97 compliance, and building sensor data · Your building doesn't need more data — it needs better inference