Loading 0%

Stop HVAC failures
before they start.

Predictive maintenance platform that alerts you weeks before critical failures.
Built for HVAC service companies, integrators, and data centers.

Scroll to explore

Intelligent HVAC Protection

Predict failures, detect faults, manage your fleet, and quantify savings. All from one platform built for commercial HVAC.

Explore our Solutions
Click here →

Predictive Maintenance Engine

Forecast HVAC failures 20 to 30 days before they happen using our multi-signal ensemble prediction engine.

1/3 Multi-signal ensemble forecasting

Multiple independent prediction methods converge to produce high-confidence forecasts with near-zero false positives.

2/3 Remaining Useful Life estimation

Get precise RUL estimates with 95% confidence intervals, so you know exactly when to schedule maintenance. For example, "~21 days remaining (CI: 9 to 47 days)."

3/3 Physics-informed degradation models

Not just black-box ML. Our models incorporate thermodynamic principles like fouling resistance curves and UA decay to ground predictions in real equipment physics.

Chiller-01 Prediction HIGH RISK ACTIVE SIGNALS 4 / 5 triggered 95% CI: 9 to 47 days REMAINING LIFE ~21 days SIGNAL STATUS Anomaly Score detected day 40 Mann-Kendall tau=0.82 p=0.001 Sigmoid P(fault)=0.87 R2=0.97 GP Forecast critical ~day 81 Survival median 65d [48,82] DEGRADATION TREND warn crit

Predictive Maintenance Dashboard

AI Fault Detection and Diagnostics

Identify HVAC faults in real time with 95-99% accuracy. Know what failed, why it failed, and what to do about it.

1/3 ML-powered classification

Utilizing advanced time series machine learning algorithms and traditional FDD methods. Detects condenser fouling, refrigerant undercharge, evaporator fouling, economizer faults, and more.

2/3 Explainability for every detection

Every fault detection comes with clear feature contributions so your technicians know exactly which sensor readings are driving the alert. No guesswork.

3/3 Real-time anomaly scoring on the edge

Anomaly scoring runs directly on local hardware for instant detection, even before cloud analysis. Zero false positives demonstrated to date.

Chiller-01 Detection FAULT Condenser Fouling 94.2% EQUIPMENT York YK Centrifugal MODEL ML Classifier (99.3%) FEATURE CONTRIBUTIONS TCA (7.8°F) +0.42 TCO (101.3°F) +0.31 kW (198) +0.18 TEO (46.2°F) +0.07 FAULTED READINGS TCA: 7.8°F TCO: 101.3°F kW: 198

Fault Detection Dashboard

Fleet-Wide HVAC Management

Monitor every HVAC unit across all your sites from a single dashboard. Health scores, prioritized alerts, and actionable insights at a glance.

1/3 Real-time health scores for every unit

Each piece of equipment gets a dynamic health score (0-100%) calculated from sensor readings, fault status, and degradation trends. Instantly spot the units that need attention.

2/3 Prioritized alerts and recommendations

Alerts are ranked by severity and urgency. Each alert includes a specific recommended action, so your team knows exactly what to do and in what order.

3/3 White-label ready for service companies

Integrate our platform with your existing service management software under your own brand. Manage all your client sites from one unified view.

UNITS 4 AVG HEALTH 72% ALERTS 2 FALSE POS 0 CH Chiller-01 York YK Centrifugal FAULT TCA 7.8°F TCO 101.3°F 34% health Condenser Fouling (94.2%) R1 RTU-01 Carrier 48TC WARN SUCT 97.2psi DISC 312psi 68% health Refrig. Undercharge (72.8%) R2 RTU-02 Lennox KGA NORMAL DMPR 45% OA_T 72°F 94% health No Faults Detected (98.7%) AH AHU-01 Trane M-Series NORMAL SAT 55°F OAT 72°F 91% health No Faults Detected (99.1%)

Fleet Management Dashboard

Proven ROI and Cost Savings

Every predicted failure translates directly to dollars saved. Avoid emergency repairs, extend equipment life, and optimize maintenance budgets.

1/3 $7,800 to $15,200 savings per event avoided

A single chiller condenser fouling event costs $12,000 to $18,000 in emergency repairs. Early intervention brings that down to $2,800 to $4,200. That is a net savings of $7,800 to $15,200 per event.

2/3 Eliminate emergency repair premiums

Emergency HVAC repairs come with rush labor, overnight shipping for parts, and potential tenant/occupant impact. Scheduled maintenance avoids all of these costs.

3/3 Extend equipment lifespan

Catching degradation early and maintaining optimal operating conditions extends the useful life of your HVAC equipment by years, deferring costly capital replacements.

Chiller-01 Cost Impact SAVINGS COST COMPARISON Unplanned Emergency Repair $12,000 to $18,000 Early Intervention (with WattNest) $2,800 to $4,200 NET SAVINGS PER EVENT $7,800 to $15,200 WHAT YOU AVOID Rush labor Emergency parts Tenant impact

Cost Impact Analysis

Edge + Cloud Architecture

Hardware That Connects in Minutes

Our sensors connect to any HVAC unit via BACnet/IP or Modbus in 15 minutes. No wiring, no downtime, no expensive hardware. Local detection runs on-prem without sending any data offsite, while advanced forecasting happens in the cloud.

  • Low-cost sensor hardware, plug-and-play installation
  • Tier 1: Runs locally on-prem without needing to send any data offsite
  • Tier 2: Cloud-based advanced forecasting, trend analysis, and fleet-wide insights
Learn More About Our Solutions →
ON-PREM (LOCAL) No data leaves site Anomaly Scoring ML Classification FDD Engine BACnet Modbus CLOUD Advanced forecasting GP Regression Sigmoid Extrapolation Survival Analysis Mann-Kendall Trend alert CH Chiller 34% R1 RTU-01 68% R2 RTU-02 94% AH AHU-01 91%
Simple Setup

How It Works

From installation to predictive insights in four simple steps. No complex wiring, no disruption to your operations.

1

Install

Mount our sensor on any HVAC unit in 15 minutes. No wiring, no downtime.

2

Connect

Automatically connects via BACnet or Modbus to your existing building systems.

3

Learn

Our AI learns normal operating patterns and builds physics-informed degradation models.

4

Alert

Get alerts 20 to 30 days before critical failures with specific repair recommendations.

15 min
Average install time
$7,800+
Savings per event avoided
24/7
Monitoring + alerts

Ready to stop HVAC failures before they start?

Built for HVAC service companies, system integrators, and data center operators. Deploying pilots at UC Santa Cruz, partnering with Lawrence Berkeley National Lab, and backed by Berkeley SkyDeck.

Request a Demo
FAQ

Frequently Asked Questions

Everything you need to know about HVAC predictive maintenance and how Wattnest works.

What is HVAC predictive maintenance and how does it work?

HVAC predictive maintenance uses machine learning algorithms and sensor data to forecast equipment failures before they occur. Wattnest monitors key parameters like condenser approach temperature, suction pressure, discharge pressure, and economizer damper position to detect degradation patterns. Our multi-signal ensemble prediction engine combines five independent methods — anomaly scoring, Mann-Kendall trend analysis, sigmoid extrapolation, Gaussian Process regression, and survival analysis — to deliver high-confidence failure predictions 20–30 days in advance with 95–99% accuracy.

What is the difference between predictive and preventive HVAC maintenance?

Preventive maintenance follows fixed time-based schedules (e.g., quarterly filter changes) regardless of actual equipment condition. Predictive maintenance uses real-time sensor data and AI to detect the early signs of failure and schedule maintenance only when needed. According to the U.S. Department of Energy, predictive maintenance reduces maintenance costs by 25–30% compared to preventive approaches, eliminates up to 75% of equipment breakdowns, and increases equipment lifespan by 20–40%. Jones Lang LaSalle estimates predictive maintenance costs $0.70–$1.10 per square foot annually versus $1.80–$2.50 per square foot for reactive maintenance.

What HVAC equipment does Wattnest monitor?

Wattnest monitors all major commercial HVAC equipment types including centrifugal and screw chillers, rooftop units (RTUs), air handling units (AHUs), VRF/VRV systems, heat pumps, and cooling towers. Our ML models are trained on industry-standard datasets including ASHRAE RP-1043 for chillers, LBNL FDD datasets for RTUs, and ASHRAE RP-1312 for AHUs. We detect 18+ fault types including condenser fouling, refrigerant undercharge, evaporator fouling, economizer stuck open/closed, compressor degradation, and sensor failures.

How much money does HVAC predictive maintenance save?

A single chiller condenser fouling event costs $12,000–$18,000 as an emergency repair but only $2,800–$4,200 with early intervention — saving $7,800–$15,200 per event. The U.S. DOE estimates that 40% of commercial HVAC equipment has at least one active fault at any time, causing 5–30% energy waste. For a typical 10-technician HVAC service company, remote diagnostics can reduce truck rolls by 25–50%, saving $93,750–$187,500 annually. The DOE reports that predictive maintenance delivers up to 10x return on investment for commercial HVAC operations.

How accurate is Wattnest fault detection?

Wattnest achieves 95.8–99.3% fault detection accuracy depending on equipment type. Chiller fault detection reaches 99.3% accuracy across 7 fault types using the ASHRAE RP-1043 benchmark dataset. RTU refrigerant fault detection achieves 95.8% accuracy across 6 fault types. RTU economizer fault detection achieves 99.8% accuracy. AHU fault detection achieves 96.7% accuracy. Our models use XGBoost gradient-boosted trees with SHAP explainability, so every detection includes clear feature contributions showing which sensor readings drove the alert.

How long does Wattnest take to install?

Wattnest sensors install in approximately 15 minutes per HVAC unit. The hardware connects via BACnet/IP or Modbus to your existing building systems with no wiring changes or equipment downtime required. Edge-based anomaly detection begins immediately after connection, while the cloud-based predictive models calibrate to your specific equipment within the first few days of operation.

Does Wattnest offer white-label solutions for HVAC service companies?

Yes. Wattnest offers white-label predictive maintenance solutions designed specifically for HVAC service companies and system integrators. Service companies can offer Wattnest monitoring under their own brand, generating recurring monthly revenue of $99–$199 per building while reducing emergency service calls by 25–50%. The platform integrates with existing service management workflows.

What building communication protocols does Wattnest support?

Wattnest supports BACnet MS/TP, BACnet/IP, Modbus RTU, and Modbus TCP protocols. These cover the vast majority of commercial HVAC equipment and building automation systems. The sensor hardware includes cellular connectivity via AWS IoT Core with hybrid Wi-Fi/cellular options, so no existing network infrastructure changes are required.

Have more questions?

View All FAQs → Contact Us →

Backed By & Partnered With

Lawrence Berkeley National Laboratory Berkeley SkyDeck Accelerator UC Berkeley UC Santa Cruz Carbon Fund NSF I-Corps