The value proposition

Earlier warning, from the historian you already run

For utilities and plants that keep years of readings in a process historian but have no team to build and tune a model for every asset: one watch across the whole fleet, alerts that explain themselves, and repairs planned before anything fails.

Two isometric scenes of the same fleet of transformers, pumps and tanks. Today: every asset has its own model and settings to tune, an alarm fires late, and a crew is called out in an emergency. With Patternode: one threshold plane spans the whole fleet, a single alert shows why it fired, and a crew is scheduled ahead.
TodayWith Patternode
CoveragePredictive models are built asset by asset: each mapped to its historian tags, trained on that asset's own history and tuned tag by tag by trained specialists. Templates shorten the work but do not remove it, so monitoring stops at the critical equipment.The whole fleet watched at once, with one alert threshold and no model per asset.
WarningA limit alarm fires once a value crosses its line, by which time the fault has been developing for days.Designed to warn while the numbers still look normal, from the shape of the signal, not its value.
TrustSet tight enough to fire early, alarms fire constantly and are acknowledged and ignored.Alerts held to a budget the control room can live with, each shown beside the nearest normal day.
CostAbout US$4,600 per asset a year at list for a leading product, with a 50-asset minimum on a three-year term, before the implementation project.Priced for the fleet, with no modelling project to fund before the first alert.
StartingNew sensors, a connection into operational networks, and a project before anything is learned.An offline export of data you already keep. Nothing connects to your network to try it.
OutcomeEmergency call-outs, outages and bursts.Repairs planned, with the parts and the outage window chosen.

The value proposition, drawn as an illustration, not a result: it shows no data, and no lead time or catch rate is claimed. Patternode is proving the method on public benchmarks first, then on a utility's own historian data. "Today" describes predictive models trained on historian data, as their vendors' product documentation sets them up; the price is a leading product's published list price, checked September 2026. No vendor publishes the engineer-hours per model.