Spot the unusual before it becomes a breakdown.
A fixed limit only catches a reading once it is already too high. Anomaly detection learns what normal looks like for each machine, and flags readings that do not fit, even when they are still inside the limits.
Two stages, both learned from your machine.
| Stage | When | What it does |
|---|---|---|
| Statistical checks | From day one | Compares each new reading with the machine's recent readings. A value far from its usual range (measured with standard statistical tests, z-score and interquartile range) is flagged. |
| Machine learning | After about 14 days of history | A model (Isolation Forest) learns the machine's normal patterns from its own history and flags readings that do not fit them. |
A score, not just yes or no
Each reading gets a score from 0 to 1. Higher means more unusual.
Severity
Scores are grouped into low, medium and high, so people see the important ones first.
No alert floods
A run of unusual readings from the same sensor is grouped, so people get one message, not fifty.
Per machine
Each machine's normal is its own. A hot oven and a cool press are not judged the same way.
Built into the screens you already use.
Anomaly Feed dashboard
A ready-made template that lists unusual readings as they happen.
Anomaly score widget
Add a live score to any dashboard.
Alerts
High-scoring anomalies can raise alerts through the usual channels.
AI Copilot
Ask "how many anomalies on Press 4 this week?"
See Nisthora on your own machines.
Book a demo. We will show you live data and talk through your machines, lines and shifts.
Book a demo