The trouble with a failing bearing is that it can be both obvious and invisible. Obvious, because when it finally seizes, a line stops, a shift scrambles and a maintenance manager gets the sort of phone call that permanently changes the texture of 3 a.m. Invisible, because weeks earlier the warning may have been no more than a small, peculiar tremor buried inside a machine that already vibrates for a living.
Waites makes a business of that interval between tremor and trouble. The Cincinnati company attaches wireless sensors to motors, pumps, fans, gearboxes, conveyors and other industrial assets. Those devices collect vibration and temperature readings around the clock. Machine-learning models search the stream for anomalies. Certified vibration analysts then inspect important signals and send maintenance teams a diagnosis in ordinary language: what is changing, what may be wrong and what to check next.
It is a small distinction with large consequences. A warning that says health score: 63 creates another research project. A warning that says a drive-end bearing is showing signs of lubrication trouble creates a work order. Waites is designed around the second outcome.
A stethoscope for machinery
The company dates to 2006, though its origin story begins five years earlier, when industrial sales veteran Andrew Waites met software entrepreneur Rob Ratterman. Waites had watched plant workers make manual rounds to check equipment. Wireless mesh networking suggested another possibility: machines could report their own condition remotely and more often than any practical inspection route.
Ratterman, now chief executive, brought a software founder's instincts. The resulting company sits in an unusual seam of the industrial market. It is part hardware maker, part vertical software provider and part outsourced reliability department. That mix matters because condition monitoring is not one problem. Sensors must survive heat, dust, washdowns and bad radio conditions. Software must distinguish a useful change from ordinary mechanical noise. Someone still has to decide whether the pattern points to imbalance, misalignment, looseness, lubrication or bearing damage. And a plant team has to act.
Waites now offers a tiered sensor family. The flagship SM7 is built for critical assets and hazardous environments. It carries an IP69K enclosure rating, can be configured over the air and uses ultrasonic demodulation to catch early bearing and lubrication problems. It can also monitor slow equipment down to one revolution per minute. The smaller S4B brings high-frequency analysis to tethered, space-constrained or washdown applications. The compact SV5 handles tri-axis vibration and temperature where dependable coverage matters more than the SM7's full diagnostic range.
The physical devices feed a dashboard and mobile apps used to manage asset health, alarms, action items and reported savings. Waites says its encrypted cellular connection can run independently of a customer's core IT network and programmable logic controllers. In industrial sales, that is more than a technical footnote. Every avoided integration meeting can move a pilot closer to a working deployment.
“A sensor can notice the whisper. The product earns its keep when a technician knows which wrench to pick up.”YesPress observation
The four-step promise
That third step is Waites' most revealing product decision. In a market enchanted by full automation, the company advertises its humans. Its AI has been trained on more than 13 trillion readings, and Waites reports 99.92 percent defect-detection coverage. Yet CAT II through CAT IV analysts still review meaningful alerts, contact customers and stay involved until an issue is resolved. Algorithms bring tirelessness and scale. Analysts bring skepticism, context and the ability to ask what changed on the floor yesterday.
This arrangement also answers a common problem with industrial alarms: trust is easy to lose. A maintenance crew that repeatedly tears down healthy equipment because software cried wolf will learn to ignore the software. Human verification is not merely a service layer. It is part of the interface between a statistical model and a person whose next action costs labor, parts and production time.
Selling avoided disasters
Waites does not publish a menu of prices. Its public sales motion is enterprise B2B: assess critical assets, run a proof of concept, deploy sensors and expand across plants after savings become visible. Customers buy a combined package of hardware, connectivity, software, installation and expert monitoring. The economic unit is less a sensor than an avoided hour of downtime.
A cracked shaft that had time to introduce itself
At a Belgian plant, Waites detected irregular vibration on a 40-year-old ball mill. The customer reported 5,376 hours of avoided downtime and more than $11.2 million in avoided production, repair and labor costs. Early notice mattered because replacement components carried long lead times.
That Owens Corning episode has become Waites' signature case. The building-materials company had evaluated 40 condition-monitoring options and used a 90-day proof of concept before selecting Waites. After the ball-mill warning, it expanded the program across 24 facilities. The eye-catching dollar figure is customer-reported, but the operational lesson is sturdier: predictive maintenance is valuable when warning time matches the realities of sourcing parts and scheduling production.
Other deployments fill in the picture. Domtar says it avoided 483 hours of downtime in one year while monitoring more than 600 assets. Buzzi Unicem USA reported more than $1 million in savings and planned to extend monitoring across its cement plants. A global e-commerce operator reported 40 percent fewer lost production hours across 13 fulfillment centers, with $15.2 million in savings and 143 percent first-year return. At Jensen Precast, 48 sensors went live in a day to watch mixers and conveyors.
The customers are reliability engineers, maintenance supervisors, plant managers and operations executives in businesses where physical flow is revenue: automotive, logistics, mining, cement, pulp and paper, food and beverage, pharmaceuticals, energy, airports and commercial facilities. Waites cites an average deployment of more than 1,500 sensors. At that scale, manual inspection is not disappearing, but it becomes more targeted. A technician can spend less time hunting for a problem and more time fixing the one the network has already narrowed down.
Where Waites fits
Routes and reaction
Periodic handheld inspections, calendar maintenance and emergency repairs. Familiar and flexible, but faults can grow between rounds.
Coverage plus follow-through
Continuous sensing, automated detection, expert review and documented action across many sites and asset types.
Data without a department
Lower-touch hardware and software can suit expert in-house teams, but the customer owns more interpretation and response.
Broad platform, heavier lift
Large automation vendors can connect condition data to wider plant systems, often with more integration and implementation work.
Competitors range from focused platforms such as Augury, Nanoprecise and Petasense to long-established industrial vendors including SKF, Emerson, ABB, Fluke and Siemens. Waites' argument is not that vibration monitoring is new. It is that its particular bundle is unusually deployable: rugged sensors, a separate communications architecture, broad asset coverage, a proprietary data set and humans who turn an alert into an intelligible next step.
The December 2025 partnership with MaintainX pushes that argument further. A verified Waites alert can now create a prioritized work order automatically, carrying an analyst's summary, recommended action and a link to deeper vibration history. Comments synchronize in both directions. It sounds mundane, which is precisely the point. Predictive maintenance has spent years improving the prediction. The remaining leak is execution - the distance between knowing and doing.
There are limits to every neat ROI story. Avoided losses are estimates about an event that, by definition, did not happen. Wireless sensors still need sensible placement. Models need good context. Plants need parts, permission and time to respond. Waites' own structure acknowledges these frictions. It does not promise that data alone repairs a motor. It builds a chain intended to make the right repair more likely before urgency takes over.
The maintenance department as a network
The company's scale creates its own flywheel. More installed sensors produce more examples of healthy machines, strange machines and machines approaching recognizable faults. More examples can improve detection. Better detection gives customers a reason to cover less obvious assets. Waites says it adds about 10 billion readings every day - roughly 116,000 per second on average - across six continents.
Its stated mission, “Creating a world where nothing breaks,” is knowingly impossible in the literal sense. Things break. Grease dries, shafts crack, belts wander and bearings wear out. The useful ambition is smaller: make failure less surprising. Give a plant enough notice to turn a crisis into Tuesday's maintenance plan.
That is also what makes Waites interesting beyond the factory gate. The company is an example of applied AI at its least theatrical. There is no robot foreman and no glowing general intelligence. There is a sensor the size of a small puck, a line on a chart, an analyst looking twice and a maintenance worker arriving before the smoke.