The alert that arrives while the machine is still running.
Every asset carries a live health score built from the data your team already records. When readings drift away from an asset's own baseline in a shape the engine recognises, an alert is raised with a confidence level and a likely window - not a vague warning.
Detected pattern
Bearing wear, drive end
Confidence
87%
Window
9-14 d
Score
38
Evidence
- - Vibration complaints logged on 4 of last 6 shifts
- - Motor current 11% above the 30-day baseline
- - Interval between drive-end repairs shortening
- - Runtime 1,840 h since last bearing change
Every alert states its confidence level and the evidence behind it
Scores are compared to each asset's own baseline, not a generic average
No alert fires without a named action and the part it will consume
In the product
Score bands decide what happens next.
Health scores are not decoration. Each band maps to a specific response, so technicians know whether to watch, plan or act.
Healthy85 - 100
Readings inside baseline. Routine PMs only.
Watch60 - 84
A trend is forming. Logged, not yet actionable.
High40 - 59
Pattern matched. Part flagged, job proposed.
Critical0 - 39
Failure likely inside the stated window.
CD-04
Bearing wear, drive end
87%
Critical
AC-03
Air filter restriction rising
74%
High
PU-05
Seal weep pattern forming
61%
Watch
IM-06
Heater band drift on zone 3
58%
Watch
HP-02
No pattern detected
-
Healthy
Capabilities
What this module handles.
Per-asset baselines
A press that runs three shifts is measured against its own history, not against a plant average that flatters it.
Pattern library
Bearing wear, seal degradation, filter restriction, belt slip, lubrication loss and heater drift are each matched by their own signature.
Stated confidence
Alerts carry a percentage and a time window. Low-confidence findings are logged as watch items, not escalated as emergencies.
Repair-interval analysis
Shortening gaps between repairs on the same component are treated as a signal in their own right.
Routing by severity
Watch items stay in the queue. High and critical alerts notify the named owner for that asset group.
Feedback loop
Technicians mark each alert as confirmed, partly right or false. Those outcomes tune detection for your plant.
Setup
From first login to first useful result.
- 01
Load the asset register
Import or enter your machines with tag numbers, line, criticality and install date. A spreadsheet is enough.
- 02
Backfill repair history
Even twelve months of past breakdowns and parts replaced sharpens the first scores considerably.
- 03
Log readings on rounds
Operators record runtime, meter values and any symptoms on a tablet or phone during normal rounds.
- 04
Review the first alerts
Expect watch items first. Confirm or reject them so the engine learns which patterns matter on your floor.
Questions
Asked on the floor.
Do we need vibration sensors to start?+
No. Scoring begins with runtime hours, operator symptom reports, meter readings taken on rounds and repair history. Sensor feeds can be imported later to sharpen specific assets.
What happens when an alert turns out to be wrong?+
Mark it rejected with a short reason. The engine weights that pattern lower for that asset class, and the record stays visible so nobody re-litigates the same call twice.
How is confidence calculated?+
It reflects how closely current readings match a known pattern and how much history exists for that asset. Thin history produces lower confidence, and the alert says so rather than hiding it.
Can an asset be excluded from alerting?+
Yes. Low-criticality assets can be set to monitor-only, so they still carry a score but never raise notifications.