A fan in a server and a fan in a treatment plant are the same device at different sizes. So is a booster pump and a midstream mainline pump. PRISM learns signatures, not one kind of asset.
Strip the industry away and the critical rotating assets are the same population: motors, drives, pumps, fans, compressors, and the loops they sit in. Those features generalize across scale, and the models read them in production every day.
Remote sites drop off for a week then dump it all at once. Tags carry a contractor's name from 2009. Half the fleet has no failure history. Most tools require you to fix that first. These models came out of clinical records, which are messier still.
Mainline and booster pumping, transfer and injection pumps, compression and the drives behind them. An unplanned station outage is throughput lost against a nomination, and the sites are hours from the nearest crew.
Dewatering pumps, slurry and tailings transfer, thickener and process water circuits. Duty is brutal and wear is constant, and a dewatering failure is a production stoppage rather than an equipment problem.
Transfer and circulation pumps, cooling and chilled water loops, compressed air, vacuum, and the fans and drives across the utilities block. The utilities nobody owns are usually the ones that take the line down.
Purified water and WFI loops, clean utilities, HVAC serving classified space, chillers and process cooling. An excursion here is not only downtime; it is an investigation, a deviation, and potentially a batch.
Whatever the sector, the shape is the same. PRISM reads the historian and SCADA telemetry already on site and delivers the decision into the system your crews already use.
One-directional ingest in, decisions out. Nothing in the product can change a setpoint, close a valve or start a machine.
Multivariate signature detection, the two-layer architecture, tolerance for gapped and mislabeled telemetry, and the motor-driven rotating asset population. All running in production today.
Which modes matter in your sector, what your telemetry actually contains, and the sensitivity setting that matches your real cost of intervention. That is the work of an evaluation.
We have a production record in water. We do not have one in your sector, and we will not quote you a savings figure until it comes out of your own history.
Design-partner terms in a new sector are materially better than list, because the first deployments teach the models something we cannot buy. If you are willing to share history and be a reference, say so on the call.
We build a model on your telemetry and score it against failures you already lived through, with misses counted as loudly as hits.
No cost to evaluate. No live connection required. No obligation.