The buyer's #1 question for the next 24 months: "Our portfolio sits at 54% utilization against a 79% target and an 80% peak. Where is the 25-point gap, and which AI architecture actually closes it?"

This page exists because CBRE's 2026 Global Workplace report just made utilization the most-tracked CRE metric (83% of corporate occupiers track it) and the most-important CRE metric (29% rank it #1). It is also the metric where most platforms stop at a dashboard — not because dashboards are wrong, but because the diagnosis of a utilization gap requires four kinds of intelligence that no single platform agent can hold at once.

The diagnostic the gap actually requires:

1. Hybrid policy → headcount translation that respects per-team mix.

2. Utilization staleness + sensor decay detection so yesterday's numbers don't drive today's footprint decisions.

3. Policy-Space Mismatch detection — when policy implies more bodies on peak day than the floorplate physically supports.

4. Space program recalibration (40/35/15/10 → workflow-profile-tuned) that flows from the first three, not from a salesperson's deck.

The 25-Point Gap, Numerically

CBRE 2026 Global Workplace Number What it means for your portfolio
Most-tracked CRE metric 83% Utilization is now table-stakes board reporting
Most-important CRE metric (#1 rank) 29% Capital decisions are being routed through this number
Global delivery (actual) 54% The denominator most CRE leaders are starting from
Reported portfolio target 79% What the board-deck slide says by 2027
Tuesday/Wednesday peak utilization 80% Where the floorplate is actually constrained
Implicit gap to close 25 pts The number this architecture is built to address

Notice what these numbers do not tell you: which days of the week are the binding constraint, which neighborhoods are over-occupied versus over-built, whether your hybrid policy is even physically deliverable on peak Tuesday, and whether your sensor data is fresh enough to drive a footprint decision. The single biggest mistake we see in 2026 CRE AI pilots is treating the dashboard as the diagnosis. The dashboard is the symptom.

A 4-Component Architecture That Closes the Gap

The architecture below mirrors the way a senior workplace strategist actually reasons about a portfolio — but it does so as a coordinated set of intelligence agents with verifiable handoffs, not a single black box. This is the keystone of the CRE-SP squad inside our multi-agent platform, and it is the architecture buyers should be asking every CRE AI vendor to articulate.

1. Hybrid Calibrator — the policy-to-headcount translator

role: KEYSTONE · runs first · feeds every downstream agent

Translates per-client hybrid policy (days/week, anchor days, team-level mix) into a peak simultaneous headcount curve. Without this translator, the rest of the stack is guessing. The Hybrid Calibrator is also the only place we have observed competitor architectures consistently miss — buyers should ask any vendor: "What is your model for converting our written hybrid policy into a peak Tuesday body-count by floor?" If the answer is a flat occupancy assumption, the architecture is wrong.

2. Utilization Analyst — sensor + badge + reservation fusion with staleness guards

role: INTAKE · maintains Bayesian priors · enforces freshness floor

Ingests sensor, badge, and reservation streams per client, maintains Bayesian priors by space type, and — critically — detects sensor decay, hot-desk shadow utilization, and data staleness. A 90-day staleness floor is enforced as a hard blocking constraint: portfolios trying to make decisions on stale data are flagged before the model produces output. This is where most "AI occupancy analytics" products silently fail.

3. Policy-Space Mismatch Detector — the Yahoo / Twitter RTO failure-mode guard

role: BLOCKING · catches structural policy/physical conflicts

Detects when written hybrid policy implies more bodies on peak day than the physical floorplate supports. The Amazon RTO 800-desk deficit, the Yahoo full-RTO bench shortfall, the Twitter pre/post-acquisition headcount-versus-square-footage mismatch — all three are the same structural failure mode. This detector runs as a hard block: if peak day demand exceeds physical capacity, the architecture refuses to publish a "smaller footprint" recommendation. Buyers should ask: "Does your platform have a structural mismatch detector, or just a benchmark dashboard?"

4. Space Programmer — workflow-profile-tuned, not deck-tuned

role: SYNTHESIS · consumes the first three · outputs the recommendation

Applies the workplace mix (default P3 40/35/15/10 for heads-down/collaboration/focus/social) but adjusts it to the actual workflow profile of the tenant or business unit. Heads-down dominant teams move closer to 50/25/15/10. Trading floors move to 60/20/10/10. The Programmer consumes the calibrated headcount curve and the freshness-validated utilization snapshot — not the marketing brief — and produces the program. This is the only point in the pipeline that a human deck sees.

How to Tell If You Are Buying the Right Architecture

The questions below are how procurement teams in 2026 are starting to separate platforms that look AI-native from platforms whose architecture actually closes a utilization gap. We recommend pasting them directly into your next RFP.

Procurement question What a credible architecture answers
How do you translate our hybrid policy into peak Tuesday body-count by floor? Names the calibrator agent and shows the input/output contract
What happens when our sensor data is more than 90 days stale? Architecture refuses to publish, surfaces a staleness alert, names the freshness gate
If our policy implies more bodies than the floor supports, does your platform say so? Yes — names the Policy-Space Mismatch detector as a hard block
How does the space mix change for a heads-down team vs a trading floor? Names the workflow profile, gives the 40/35/15/10 → 60/20/10/10 mechanic
Where does an IPMVP-grade savings number for this re-fit come from? Points to M&V Option C anchored to ASHRAE 90.1 baseline — not a vendor projection

Run the 25-point gap diagnostic on your portfolio

Bring a hybrid policy doc, a utilization snapshot (even one month), and one peak-day floorplan. Our agent will walk the four components for your portfolio in under 15 minutes, name the binding constraints, and tell you which gap closes with policy adjustment versus space recalibration versus capital project.

Talk to our agent → /ask/?topic=utilization-gap

Where This Architecture Fits in Our Platform

The four components above are the CRE-SP (Space Planning) squad of our multi-agent CRE platform. They share an open-substrate handoff layer (IFC 4.3 + Speckle.dev) with the CRE-AD squad (Architecture & Design — feasibility, brief intelligence, code compliance) and the CRE-TS squad (Technical Services — MEP capacity, IPMVP M&V, AI-HVAC). This is why utilization architecture is not a standalone product for us: it is the entry point into a coordinated, owner-operator-first stack that is independent of brokerage relationships and not locked into any single vendor's portfolio. For the broader positioning, see our Morris Chang commitment.

For the M&V layer that anchors the savings claim against an ASHRAE 90.1 baseline, see /ipmvp-verification/. For the related procurement document used by EU-jurisdiction buyers, see /eu-ai-act-readiness-procurement-document/.

Sources Anchored to This Page

Architecture lineage: this page is the public surface of the v71 CRE-SP squad inside our multi-agent platform. The Hybrid Calibrator is protected from Darwinian pruning as a KEYSTONE INTAKE agent; the Policy-Space Mismatch detector runs as a B4 hard block. The platform is the work; this page is the architecture documentation.