An AI that tells you when your machines will fail and what to do about it.

Diagnose the fault. Identify the root cause. Forecast the time to failure. Recommend the next action. All from data you already collect.

Validated in production with U.S. energy supermajors.

Talk to Novity See how it works

Reciprocating compressor on an offshore platform with TruPrognostics AI diagnostic overlay showing the fault (leak in throw 4 suction valves), the time to failure (12 to 20 days), and the recommended action (non-intrusive inspection of throw 4 with ultrasound, vibration, and PV analysis).

What's Different

Diagnoses your engineers can act on.

TruPrognostics AI delivers a fault diagnosis, multiple likely root cause hypotheses, a recommended next step, and a remaining useful life forecast where the fault mode and data support it. Every output is documented per fault mode and data tier so engineers know what to expect before they trust it.

1

Remaining Useful Life

Most tools detect. Novity truly predicts.

TruPrognostics gives you a probabilistic time-to-failure window for every critical machine. Your team schedules around it instead of scrambling.

2

Hybrid AI Architecture

Physics, machine learning, and contextual AI combined.

Physics encodes how machines fail. Machine learning detects deviations from normal. Contextual AI identifies root causes and translates them into specific actions.

See the full architecture →

3

Sourced Recommendations

Every action cites the manual it came from.

The reasoning space is constrained to your OEM manuals, customer SOPs, and prior work history. The system cannot hallucinate.

Results

Validated in production.

92–93%

Detects over 90% of faults across validated fault modes — with a low false alarm rate. Performance validated in production against customer-confirmed concurrent events. Multiple operators, multiple sites.

two engineers looking at a Novity TruPrognostics dashboard

Compressor predictive maintenance dashboard showing loss of efficiency detected 40+ days before OEM alarm

Proven In Production

40+ days of lead time on a critical compressor at a U.S. energy supermajor.

TruPrognostics AI flagged a developing fault on an Ariel reciprocating compressor 40+ days before the customer's existing monitoring would have caught it. The fault was confirmed at the next inspection. The maintenance team scheduled the work. No emergency callout, no production loss.

Talk to Novity about your assets

Start With The Data You Have

Three tiers per machine, defined by data. Day-one value at every level.

Start at Base with data already in your historian. Add sensors per machine to unlock Plus or Premium. No new software, no redeployment.

Base

PI / time-series data from your historian. Process-domain diagnostics and prognostics. Detects efficiency loss, compression changes, and some bearing conditions.

Plus

Base data plus one high-frequency vibration sensor. Precise mechanical and bearing diagnostics. Valve leakage and oil whip detection.

Premium

Multiple high-frequency sensors and modalities. Maximum diagnostic precision: individual valve identification, crosshead wear, rider band wear.

Asset Coverage