AI Listens to Machines to Diagnose Mechanical Problems

Augury has more than seven million recordings of machine sounds as part of an effort to create the factory of the future.

IoT-enabled wireless sensors gather mechanical data from critical equipment before Augury’s proprietary machine learning algorithms compare these unique machine signals to thousands of similar recordings to diagnose and translate the collected data.
IoT-enabled wireless sensors gather mechanical data from critical equipment before Augury’s proprietary machine learning algorithms compare these unique machine signals to thousands of similar recordings to diagnose and translate the collected data.
Augury

Augury is a full-stack IoT company with headquarters in Israel and New York.

The company was co-founded by Saar Yoskovitz, CEO, and Gal Shaul, CTO, after Shaul took a business trip while working for a medical device company. He was sent into the field to inspect a machine that wasn’t operating properly. 

When Shaul arrived, he could tell that the fan on the machine needed to be cleaned simply from the sound it was making. Shaul told Yoskovitz about his trip and they developed the idea of using software to listen to machines to diagnose mechanical faults. 

They officially started Augury in 2012 out of a basement in Israel. Today, the company has headquarters in Israel and New York and serves as a digital transformation partner for industrial organizations that seek to increase productivity and reduce unplanned downtime by optimizing machine health and performance.

In this interview with Jonathan Biagiotti, Product Marketing Manager at Augury, we discuss the company’s technology; why it has more than seven million recordings of machine sounds; and what the factory of the future will look like. 

IEN: How does the company's technology work?

Jonathan Biagiotti: Augury’s technology has rapidly developed since the early prototypes created by Saar and Gal. 

Today, Augury’s flagship technology is its HALO turnkey solution, which monitors critical assets by combining vibration, temperature and magnetic data with advanced artificial intelligence. 

Our IoT-enabled wireless sensors gather this mechanical data from critical equipment then Augury’s proprietary machine learning algorithms compare these unique machine signals to thousands of similar recordings to diagnose and translate the collected data into powerful and actionable insights. 

The company has a dedicated team of PhDs in physics, electrical engineers and vibration analysts who tag data and modify Augury’s algorithms to fine-tune mechanical diagnostics. Stakeholders are notified of what’s wrong with their equipment, possible causes, and how to fix them.

IEN: What are the primary industries that you serve?

Biagiotti: Most customers are part of the Fortune 500, spanning the consumer-packaged goods (CPG), food and beverage and pharmaceutical manufacturing industries. Augury also has a presence in the automotive and wood products industries.

IEN: How long does a typical implementation take?

Biagiotti: Since we sell a turnkey solution, not just a monitoring tool, Augury takes a consultative approach to each new customer relationship. Our engineers visit customers at their facilities to conduct an audit that helps determine what the networking layout should be, and to determine how to best monitor critical equipment to improve machine health. 

These site surveys allow us to write a customized proposal for each of our clients that maximizes their return on investment. This approach allows us to be sure that the HALO solution is built specifically for the customer’s application. Augury invests in preparation for efficient installations, so actual installation can take as little as a few hours or at most, a few days.

Typically, from day one customers start to see value in this investment – as they’re collecting data from their machines in a more efficient matter than ever before. Within two weeks of implementation, our AI has enough data to understand each machine and its unique modes of operation. Most customers see a return on investment within six months of installation of our HALO sensors, and all of them usually do by the end of the first year.

IEN: What is the hesitation for companies that have yet to become more tech savvy? Is it cost? Company size? 

Biagiotti: The largest issues companies are facing when it comes to machine health have to do with the issue of scale and a knowledge gap around data management and data analysis. 

Today, most businesses rely on basic telemetry data from the machine’s sensors. This allows stakeholders to know when a machine is on or off, its temperature and pressure readings, or other raw sensor data.

However, there are countless cases where the control panel or HMI shows that a piece of equipment is fine, but there is actually an underlying issue that can’t be seen with their existing data. This is part of the reason why the industry is moving away from raw data and towards data-driven insights, which is what our continuous diagnostics solution provides. 

Business leaders are aware that machine learning algorithms can help them manage and sort the enormous amount of data they collect. However, they don’t know how to turn that data into actionable insights. Cost is rarely the issue. Installing the HALO solution is an investment that reduces the cost of machine repairs within the first year, and pays for itself quickly.

IEN: Can technology help small- and mid-sized manufacturers compete with larger companies? 

Biagiotti: Yes. With remote continuous monitoring and diagnostics, small and mid-sized manufacturers can reduce the amount of costly last-minute repairs they need to make to machinery at their manufacturing sites. 

Instead, they can schedule repairs at early stages when it’s cheaper for everyone involved to fix and without interrupting production runs. Increased uptime and reduced waste caused by machine faults all help increase their margins and improve their agility, which is essential for survival when competing against larger firms with greater resources.

IEN: How many machines has Augury diagnosed? 

Biagiotti: We’ve made more than 70,000 diagnoses across more than 6,000 facilities. The sheer number of encounters we’ve had speaks volumes. To date, we’ve also built a library of nearly seven million machine recordings.

IEN: In your opinion, how far has the industry come in the switch to IoT/Industry 4.0? And how far does the industry have yet to go? 

Biagiotti: Many businesses are piloting IoT tools, but because they still rely on basic telemetry data, they are still trying to understand the services or tools they need. 

Certain industries, such as pulp and paper or oil and gas, are further along in this process. They’ve already implemented machine health monitoring technology because of how expensive their machine investment is. 

In the next five to ten years the type of predictive insights that our continuous diagnostics generate will become the standard expectation of customers when it comes to IoT machine monitoring technology.

IEN: What are your expectations for the company over the next year? 

Biagiotti: We’re experiencing phenomenal growth with our Fortune 500 and mid-market clients. The majority of our customer relationships are at the point where they’re expanding from a few plants that have standardized our machine health and performance solution to now being the global standard across the entire manufacturing footprint. 

This growth coincides with clients’ desire to compare performance across plants around the world to industry- and company-wide performance benchmarks, and Augury’s ability to do that within a cohesive platform that drives adoption through actionable insights. Within the next five years, we will transition from the inconspicuous leader in machine health and performance solutions to the undisputed industry standard-bearer.

IEN: What does the factory of the future look like? 

Biagiotti: Lights out factories are the future. However, this won’t be achievable without addressing the blind spot of machine health and performance. 

Many processes will occur autonomously within factories, and systems across the business will communicate with one another since everything will be connected due to the widespread implementation of IoT technology. 

Data will go from siloed and too overwhelming to manipulate, to easy-to-analyze and highly interpretable thanks to AI. 

By leveraging IoT integration, autonomous machinery, and artificial intelligence, businesses will experience tremendous productivity gains. As a result, more employees will be free to spend time acting upon the insights that the newly integrated data will provide.

IEN: From a state of the industry perspective, what keeps you up at night? 

Biagiotti: More connectivity leads to the potential for more security threats. 

Today, Augury doesn’t control any of our client’s systems, we just monitor data and deliver insights that help them make better business decisions. 

Because we take security so seriously, all the data we store is highly secured and backed up through cloud services. 

However, as all IoT service providers continue to grow, and eventually integrate with other IoT systems, security will become more complicated. 

It’s a challenge Augury is already anticipating and we’re taking all the necessary steps to ensure our solutions never become a point of vulnerability.

More