Furnace power, motor current, vibration, cycle time — the signals that explain a loss are already on your machines. Munshi reads them off a panel it fits, and holds them on the same clock as the reason your operator logged. The join is the point: a reading on its own is a number; a reading next to a logged cause is an explanation.
Your ERP logs the outcome - tonnes, pieces, energy booked.
It never sees the signal underneath - the furnace holding, the wheel drawing high, the current climbing for the same job.
Why the gap existsWhat it does
Power, current, vibration and cycle time, sampled continuously off a panel fitted to the machine — the readings that move before a loss shows in the count.
A signal is held against the operator’s reason on the same clock and the same asset. The excursion and the “crane wait” that explains it sit together, not in two systems.
Each channel resolves to one asset and one parameter at the point it is bound, so a current reading is this wheel on this line — not a number that could be anything.
In the plant
Furnace power held flat while the metal waited. The reading was there; Munshi put it next to the crane-wait the operator logged.
A blast wheel drawing higher for the same job — blade wear, weeks before it would have been a stoppage.
Zone temperatures that drifted while the line stalled, next to the soak that burned the surface into scale.
What Munshi surfaces
Teal is normal, maroon is flagged. Illustrative data.
The shaded stretch is metal at temperature with nowhere to go.
A wheel drawing harder for the same job — wear, weeks before a stoppage.
The outliers the signal already explains.
Cycle time creeping for the same batch — a machine no longer performing as it did.
Drifting past where it should return. The afternoon nobody logged.
Get started
See this on your own floor.
Bring a month you already understand. The useful test is whether Munshi finds what you already know — and then what you didn’t.