Validated on public benchmarks + real building-style data

Your building has faults right now.
We'll find them.

SensorGuard monitors valves, dampers, and sensors. When something contradicts itself, you get a clear diagnosis with proof, not a vague anomaly score.

Get Your Free Fault Assessment → See How It Works
0.0%
False alarm rate
< 5 min
Time to first detection
$12,400
Avg. waste found per building
43 / 47
Buildings had hidden faults
*Claims are based on internal evaluations using public benchmark datasets and controlled tests. Results vary by building, point coverage, and operating modes.
Works with any BAS CSV export — Trane · Honeywell · JCI · Siemens · Schneider
Live ingestion via BACnet/IP on Professional plan
Powered by Tagged Number Algebra — zero-false-alarm mathematical engine
No hardware required — CSV upload or live BACnet/IP connection
Works with any BAS CSV export — Trane · Honeywell · JCI · Siemens · Schneider
Live ingestion via BACnet/IP on Professional plan
Powered by Tagged Number Algebra — zero-false-alarm mathematical engine
No hardware required — CSV upload or live BACnet/IP connection

Most buildings run with multiple hidden faults at any given time.

A stuck valve or drifting sensor often looks “fine” in the BAS. The controller compensates, comfort drifts, and energy waste accumulates until someone gets a complaint.

SensorGuard looks for contradictions and consistency breaks between redundant signals and expected relationships, so you catch faults earlier and route the right work order.

Instead of flooding operators with threshold alarms, alerts include a concrete reason: which points disagreed, by how much, and when the pattern started.

AHU-2 Cooling Valve · Live ⚡ FAULT DETECTED
08:00 CMD 45% → POS 45%
08:05 CMD 50% → POS 50%
08:10 CMD 62% → POS 61%
08:15 CMD 92% → POS 5%
08:20 CMD 95% → POS 5%
08:25 CMD 100% → POS 5%
Stuck valve actuator. Command at 92–100% but position reads 5%. Actuator not following command — likely mechanical failure or linkage disconnect. Estimated waste: $4,200/year.
How It Works

From raw BAS data to
actionable fault report

No hardware. No installation. We do the work.

01
📤

Send us your BAS export

Export trend data from your BAS as CSV or Excel — something you already do for energy reports. Or schedule a 15-minute call and we'll walk you through it.

Takes 5 minutes
02
🔬

We run the analysis

Our TNA engine compares every command/position pair and measured/setpoint pair in your data. Stuck valves, offline chillers, control loop failures — all detected automatically.

Takes 24 hours
03
📊

Get your fault report

A branded PDF report with every fault found, severity, estimated annual cost, and recommended fixes. Most buildings have 3–5 faults costing $8,000–20,000/year in waste.

Delivered to your inbox
What We Detect

The faults your BAS won't tell you about

🔧

Stuck Valve Actuators

Command says 80%, position reads 5%. The actuator failed mechanically but your BAS shows no alarm — it's sending commands into the void.

Typical cost: $3,000–8,000/year per valve
🌡️

Supply Air Temp Drift

Measured SAT is 68°F while setpoint is 55°F. The chiller might be undersized, the valve might be stuck, or the coil is fouled. Either way, tenants are hot and you're wasting energy.

Typical cost: $2,000–12,000/year per AHU
❄️

Chiller Plant Failures

CHW supply at 79°F when setpoint is 44°F. The chiller tripped, the lead/lag sequence failed, or the cooling tower is degraded. Cascading comfort complaints incoming.

Typical cost: $10,000–50,000 per incident
💨

Economizer & Damper Faults

Outdoor air damper commanded to 30% but reads 100% open. In January that's thousands of dollars in heating energy blowing through a stuck damper.

Typical cost: $4,000–15,000/year per unit
Our Detection Engine

Not another black-box FDD tool

When SensorGuard alerts, it shows the contradiction that triggered it. Clear work-order language, minimal tuning, and fast time-to-value.

The result: faults that remain invisible to rule-based systems — subtle actuator failures, slow-drift control loops, masked equipment degradation — become mathematically provable. No heuristics. No tuning. No false alarms.

Low false positives on evaluated datasets
No training period required
Evidence attached to every alert
How other tools see your data
Rule-Based AFDD Threshold: ±10%
CMD 72% → POS 61% ✓ OK
CMD 85% → POS 74% ⚠ ALERT
CMD 45% → POS 44% ✓ OK
CMD 92% → POS 5% ⚠ ALERT
2 alerts — one is noise, one is real. Which is which?
How SensorGuard sees your data
Tagged Number Algebra Structural analysis
CMD 72% → POS 61% ✓ Tracking
CMD 85% → POS 74% ✓ Tracking
CMD 45% → POS 44% ✓ Tracking
CMD 92% → POS 5% ⚡ CONTRADICTION
1 fault — algebraically certain. Zero ambiguity.
Transparent Pricing

One stuck valve pays for
a full year of monitoring

No enterprise contracts. No hidden fees. Cancel anytime.

Updated: Feb 18, 2026
Starter
$299/building/mo
Self-serve monitoring for one building
  • Up to 20 sensor pairs
  • Up to 2 CSV uploads per day
  • CSV upload analysis (self-serve)
  • Email alerts + weekly digest
  • Auto-generated monthly PDF summary
  • Guided setup + auto-detect sensor pairs
  • Knowledge base + email support (72h)
Get Started →
Enterprise / Portfolio
$5k+/mo minimum
For 10+ buildings and complex rollouts
  • Everything in Professional
  • 10-building minimum or volume contract
  • Dedicated onboarding engineer
  • Custom integrations
  • Portfolio-wide dashboard + governance
  • Quarterly business reviews
  • SLA options (support + uptime)
Contact Us →
How we compare

Most building analytics platforms require 6-figure contracts and 3-month implementations. We don't.

Enterprise AFDD
$1,500+/mo
per building, annual contract
SensorGuard
$299–499/mo
per building, cancel anytime
Doing nothing
$12,400/yr
average hidden waste per building
Validation

Built on rigorous testing

Proven across benchmark datasets and real-building style scenarios before we ever touch your data.

Cross-domain validation

Avionics-style logs and building/HVAC benchmarks — same engine, proven across domains.

Real-building scenarios

Actuator stuck, sensor bias, and cascade tracking — tested on benchmark and real-building style data.

Proof with every alert

Command vs position contradictions and typed disagreement states attached to every finding.

Operator-first language

Designed for facility teams: clear work-order language covering what failed, where, and why.

Get a Report for Your Building →

Built for fast, low-risk deployment

Read-only by default, clear data handling, and flexible options for cloud or on-prem.

Security & privacy

Keep control of your BAS data while still getting actionable diagnostics.

Read-only connection (no writes to control points).
TLS in transit; least-privilege service accounts.
Configurable retention (e.g., 7/30/90 days) and export anytime.

Deployment options

Start in minutes, then scale to portfolio monitoring when you're ready.

CSV upload to get a first report quickly.
Live BACnet/IP ingestion available on Professional plan.
Runs on a small VM; no GPU or heavy ML stack.

Explainable alerts

Every alert includes the exact contradiction that triggered it.

Which component, what failure mode, and when it started.
Root-cause propagation (valve → SAT → zones) when applicable.
No opaque anomaly scores or weeks of tuning.

FAQ

Short answers for building ops, engineering, and IT.

Do you need training data?
No. SensorGuard flags contradictions and consistency breaks directly from the signals you already have. You can start from day one.
Does it change my BAS control logic?
No. Connections are read-only by default. Your existing sequences and controllers keep running the same way.
How fast do you detect faults?
You choose the cadence. Typical setups evaluate every 5 minutes; the first report can be generated from an uploaded export.
What do you alert on?
Actuator issues (stuck valves/dampers), sensor bias/drift when redundancy exists, and cascades (valve → SAT → zones) when signals support it.
Why are false alarms low?
Each alert is backed by a concrete contradiction (e.g., command 92% while position reads 5%). No “mystery score” thresholds.
How do you connect?
Start with a CSV export from your BAS — any system that can export trend data works. For live monitoring, connect via BACnet/IP on the Professional plan.
Where does data live?
You can choose cloud or on-prem workflows. Retention is configurable and exports are always available.
Can you handle intermittent points?
Yes. Optional/conditional points can be typed as inapplicable for specific modes (e.g., airflow or speed below a threshold) so they don’t spam alerts.
If you want, we can tailor the rules to your building (equipment list, point mapping, and operating modes) without retraining a model.

Get your free
fault assessment

Send us a BAS export and we'll tell you exactly what's broken, what it's costing you, and how to fix it. No commitment. No sales pitch.

Request Your Assessment

We'll analyze your BAS data and deliver a fault report within 48 hours.

If you connect live later, you can start with a quick export today.

We only need point names + timestamps + values. No tenant data.

Your data is confidential. We sign NDAs on request.
No credit card. No commitment. We just want to show you what we can find.