Skip to content

Predictive maintenance

Predictive maintenance without new hardware.

Fault signals from the telemetry your building already produces. Ranked tickets, verified fixes, audit trail.

Book a demo

Signal catalog

Signal sources, detected patterns and automatic actions
Signal sourceDetected patternAutomatic action
Sensibo (temp / rH / setpoint)Setpoint not reached for more than 30 min under a stable outdoor loadDiagnosis candidate, dry-run recommendation, ticket in Inventory
Shelly (P, kWh)Consumption drifts more than 15% against baseline in the same operating modeAnomaly with a physical explanation, candidate for maintenance
Home Assistant historyHot-water meter shows a systematic -3% against ShellyCalibration notice, correction factor 1.031 active, noted in the EvidenceRecord
Asset battery≤ 15% warning · ≤ 10% critical · 0% offlineStaged alerts with duplicate suppression and notification audit
Maintenance planGrundfos ALPHA1 25-60 / Vaillant aroSTOR VWL BM 270/5: 12-month cycleAutomatic service-plan ticket 30 / 14 / 7 days before due date
CAD ⇔ as-built modelMissing geometry or deviating zonesMISSING_GEOMETRY_AREA reported, no substitute value, engineering review
Temperature watchdogRoom below 24 °C with rH above the target bandBlocks deeper cooling; recommends dry mode / ventilation
Emergency stop endpointCritical anomaly with a safety implicationEmergency route triggers immediate HVAC shutdown outside the dry-run gate

From anomaly to verified fix

  1. 1

    Signal

    Sensibo, Shelly, Home Assistant, CAD, battery state and the time-series baseline are read continuously. Nothing new is installed.

  2. 2

    Diagnose

    The deviation gets a physics-anchored explanation instead of a bare alarm code: what the building should have done, what it did, and the most probable cause.

  3. 3

    Ticket

    A service-plan item is opened in the Inventory module, ranked by expected impact and linked to the asset, room and system it belongs to.

  4. 4

    Verified fix

    After the repair the same baseline is measured again. The fix rate is logged, so a recurring fault cannot quietly stay open.

Service plan

Every asset carries its own maintenance cycle. A Grundfos ALPHA1 25-60 N 180 runs on a 12-month cycle, a Vaillant aroSTOR VWL BM 270/5 likewise. BuildingIQ opens the service ticket automatically 30, 14 and 7 days before the due date, attributes it to the room and system, and keeps it open until the fix has been verified against the baseline.

Battery health

Battery alert tiers
LevelAction
≤ 15%Warning — service plan item created, no escalation
≤ 10%Critical — escalated notification to the operations contact
0%Offline — asset marked as unreadable, data gap recorded

Repeated alerts for the same asset are suppressed, and every notification that did go out is recorded in the notification audit.

Confidence and explanation

A fault reaches the dashboard only with an anomaly score calibrated against the reference model for your building — not a black-box classifier. Each entry carries the expected behaviour, the observed behaviour, the deviation, and the reasoning behind the diagnosis. Human review is the default; automation is an explicit choice per site.

OPS benchmark and continuous savings

Every ticket carries its resolution time, technician and outcome — so BuildingIQ benchmarks your OPS team across your buildings and against comparable sites, and turns the findings into a savings backlog that compounds month over month.

OPS efficiency metrics and how they are used
Benchmark metricWhat it showsHow it drives savings
Mean time to resolveHow long a fault stays open, per building and technicianHighlights where hours are lost and which sites need support
First-time-fix rateShare of tickets closed on the first visit, without a repeatLow rates point to missing spare parts or unclear diagnoses
Repeat-fault rateFaults that reappear after being marked fixedEach repeat fault is a verified savings candidate with a root cause
Backlog per technicianOpen, ranked tickets per team member against comparable sitesBalances workload before small faults grow into costly ones

The savings loop: each verified fix and ranked finding enters the savings backlog with an expected impact. You approve the measures, they are implemented on your schedule, and each one is verified against the same baseline — so the savings in the monthly OPEX report are proven, not promised.

FAQ

Do we have to retrofit hardware for this?

No. Battery state, runtime, power drift, setpoint tracking and humidity control cover most fault classes. Add-on vibration or acoustic sensors are supported but optional.

How do you avoid false positives?

Anomalies are scored against a calibrated reference model for your envelope, not a generic threshold. Below the confidence bar, nothing reaches the dashboard.

Can it act on its own?

Only if you turn that on per site. Dry-run recommendations are the default, and every automated action stays inside the physics guardrails.

What data quality do we need?

Minute-level or five-minute telemetry from the relevant systems plus a CAD or floor-plan source for the Core model. Gaps are recorded, not interpolated away.

Does it integrate with our CMMS or ERP?

Tickets can be exported or pushed by API into an existing CMMS. The audit trail stays in BuildingIQ so the evidence chain remains intact.

The AI orchestration for hotels, professional buildings and commercial premises.

30 minutes, live on one of your own sites. No slides.

Book a demo