AUGURY

Somewhere on a production line right now, a motor is failing. It looks fine. It sounds fine to anyone walking past. But buried in the hum is a frequency that has started to climb, the mechanical equivalent of a cough that turns into pneumonia. Augury heard it three weeks ago, flagged it, and told a maintenance crew exactly which bearing to replace. The line never stopped. Nobody made a headline out of a disaster that did not happen.

That is the strange business Augury is in: selling the absence of catastrophe. You cannot photograph the explosion that never occurred, or invoice for the eight hours of downtime a plant did not eat. Yet for 500-plus manufacturers - the companies that bottle your soda, mold your toothpaste, and wrap your chocolate - that quiet is worth a great deal. Enough, it turns out, to make Augury the only unicorn in its corner of the industrial world.

An AI company that lives on the factory floor

Augury is an industrial AI company headquartered in New York, though its roots run back to Israel, where co-founders Saar Yoskovitz and Gal Shaul started it in 2011. It does two things that sound simple and are not. First, Machine Health: it bolts sensors onto the spinning parts of a factory - pumps, motors, fans, compressors - and uses machine learning to predict when they will break. Second, Process Health: it watches the production line as a whole and finds the hidden reasons a batch came out wrong or a yield slipped.

The premise underneath both is almost old-fashioned. Machines have always told us when they are sick. They get hot, they shake, they whine at frequencies a trained ear can catch. The problem was never the signal. The problem was that nobody could listen to ten thousand machines at once, around the clock, without getting bored or going home. Augury's bet is that software can.

Downtime: the most expensive word in manufacturing

Ask any plant manager what keeps them up at night and you will eventually arrive at the same word: downtime. An unplanned stoppage is not just lost production. It is spoiled material, idle workers, missed shipments, and the special misery of discovering the failure only after the line has gone silent. For decades the industry's answer was to either run machines until they died - cheap until it wasn't - or to service them on a fixed calendar, replacing perfectly good parts out of pure anxiety.

Both approaches are, in their own way, a confession that nobody actually knew what the machines were doing. The first guesses too late. The second guesses too early. Augury's founders looked at this and saw an information problem dressed up as a maintenance problem. The data existed. It was leaking out of every bearing in the world as heat and vibration. It was simply going unheard.

A stethoscope for machines

The first version of the idea was charmingly literal. In 2014 Augury launched Auguscope, a handheld device that plugged into a smartphone and let a technician press it against a machine, like a doctor with a stethoscope. The phone would listen and offer a diagnosis. It was a clever party trick and a serious wedge. Every reading became training data. Every diagnosis sharpened the model.

By 2017 the company had something bigger: a whole category. It called it Machine Health, and it was no longer about a person walking around with a phone. It was permanently installed sensors streaming data to the cloud, where prescriptive AI did not just say "something is wrong" but "this bearing, this fault, fix it this way." The bet had matured from a gadget into an operating system for reliability.

What it actually does between failures

Strip away the category names and Augury is a pairing of hardware and software. The hardware is a family of sensors called Halo. The R4000 line captures triaxial vibration, temperature, and magnetic data - the full vocabulary of a machine in distress. The U2000 ultrasonic system handles the awkward cases: equipment so slow it barely turns once a minute, like the giant kilns in mining and cement, where ordinary sensors hear nothing useful. Augury's answer is to sample fast, up to 100kHz, and let the math find the flaw hiding in machinery that looks, to the naked eye, almost motionless.

The Process Health half arrived by acquisition. In 2022 Augury bought Seebo, an Israeli process-optimization company, and folded its technology in. The logic was clean: knowing a pump is about to fail is useful, but knowing why the entire batch keeps coming out two percent under spec is a different and equally expensive question. Owning both lets Augury argue it watches not just the parts but the whole.

The numbers that make skeptics quiet down

Predictive maintenance has a long history of overpromising, so skepticism is fair. The counter-evidence sits in the customer list and the growth curve. Augury says that since its 2021 round, revenue grew roughly five-fold and its base of Fortune 500 manufacturers tripled. The names attached are not pilots-that-went-nowhere; PepsiCo scaled the technology across its network and wrote publicly about the lessons of doing so.

The investors reading those numbers are a tell in themselves. Beyond the financial firms, Augury's cap table includes strategics with real skin in the industrial game: Baker Hughes from energy, Schneider Electric's SE Ventures from automation, and Qualcomm Ventures from the chip side that makes edge sensing possible. When the companies who could build a competitor instead write checks, it usually means something.

People and machines, working the same shift

Augury is careful, in its own language, not to frame this as machines replacing people. Its stated vision is a world where the combined work of people and machines makes life better. In practice that means the AI does the tireless listening and the human does the judgment - deciding what to fix, when, and whether the plant can afford to wait. The values it lists internally - People First, Own It, Question It - read like a company that knows its product only works if the technicians on the floor actually trust it.

There is a sustainability argument folded inside the maintenance one, too. A machine that fails catastrophically wastes energy, material, and often the entire batch running through it. Catching the fault early is, almost incidentally, a way to waste less. For manufacturers under pressure to cut both costs and emissions, that two-for-one is part of the pitch.

The factory that fixes itself

The direction of travel is clear in Augury's newest product, the Industrial AI Workforce - software agents that do not just flag a problem but help orchestrate the response. The logical endpoint is a plant that detects, diagnoses, and increasingly decides on its own, with humans supervising rather than reacting. Whether the industry gets all the way there is an open question. Whether it wants to is a harder one.

But the trend pushing it forward is not subtle. Skilled maintenance workers are retiring faster than they are being replaced, and the institutional ear for what a healthy machine sounds like is walking out the door with them. Software that has listened to millions of hours of vibration is one way to keep that knowledge in the building. Augury is betting it is the way.

So return to that motor on the line, the one that was quietly failing. In the old world, it ran until it didn't, and someone got a very bad phone call. In Augury's world, it got a work order three weeks early and a fresh bearing on a Tuesday. The line kept moving. The chocolate got wrapped. And the most interesting thing that happened that day was, gloriously, nothing at all. That is the product. The hard part was convincing an entire industry that nothing is worth paying for.

Interviews & product demos