IIoT Startups Exploit Their Small Size to Deliver Big Data

Chen Linchevski

My partner Raanan Manor and I realized that there was a problem in translating the behavior of the machine to a language that a computer can understand. We’ve worked very hard to make this complicated process very simple in terms of user experience. I believe that by combining the knowledge of plants, structure, and behavior with data analysis will enable us to predict failures more accurately and efficiently. This is how we came up with Precognize.

What is your background?

Prior to founding Precognize, I was the CEO of Opcat, served in leading positions in Ayeca, and worked with GE, the Israeli Electricity Company, Elbit, Electro Optics, SilverlakeAxis, and Elta Systems, specializing in conceptual modeling of complex systems. I have been a guest speaker in the “Systems Architecture and Lifecycle Design: Principles, Models, Tools & Applications” course at the MIT Professional Institute.

Additionally, I was a contributor to the development of agile methodology for intensive software development together with representatives from Nokia-Siemens Networks, Philips and F-Secure. I hold an LLB from the Hebrew University in Jerusalem.

What does a company stand to gain by implementing predictive maintenance in their factory?

The way maintenance is done today is extremely outdated—100 years ago, factories were responding to issues, rather than preventing them. That responsive (instead of proactive) approach is expensive and interruptive.

Predictive maintenance is revolutionizing the industry and transforming the entire way of doing operation. This technology predicts problems in a factory before they even happen, changing the paradigm and preventing disasters. The technology’s proactive monitoring also provides significant savings to a factory.

With time, ROI will become increasingly evident from less downtime, more production, lower maintenance budgets, and better reputation due to fewer accidents. In addition, capital expenditure investments can be reduced in redundant equipment, companies can pay for less repairs in a crisis, reduce environmental damage (flaring events), and cause less damage from “hot shutdowns” (equipment and consumables).

How does machine learning save money for a factory?

Machine learning does a tremendous job detecting anomalies. But at the same time, it creates a lot of noise, making it hard to differentiate anomalies in the data and find true problems in a plant.

Bridging that gap with the right solution results in immediate savings from production loss. Predictive maintenance could lower shutdown costs by 50%. In addition, predictive maintenance ensures long-term savings, such as cutting down on maintenance costs due to lengthening maintenance intervals and getting rid of redundant backup systems.

Ranaan Manor

How do engineers work with the technology?

The process engineer, operation manager, or maintenance engineer models the plant through an easy-to-use interface, which allows the engineer to describe the structure and behavior of the plant, down to the measurable sensors, ultimately connecting the machine, the process, and the data.

It took a lot of hard work to make the interface intuitive, but the brain behind it is obviously complex, allowing the software to imitate the way the engineers think. We believe that incorporating the engineers’ intimate knowledge of the plant into the technology is key to predictive maintenance, while also making their knowledge an integral aspect of the new software.

What should be considered during the installation process?

Prior to starting with a company, it is important to ensure that they have collected historical data from the thousands of sensors in their plant(s). Not all data is saved, or kept longer than a short period of time before being erased. Ideally, companies should store data for at least 3 to 4 months up to a year.

With this technology, companies that store their data don’t need to install new sensors when implementing a new solution. Rather, they can work with the ones they already have.

Why should these companies partner with a small startup with no background in the industry? What are the benefits?

Instituting changes can be a challenge in some organizations, especially those that have a well-developed way of doing things. Innovation requires the ability to change quickly. But change isn't easy, especially in a complicated and highly regulated work environment where things have been fine-tuned to work in a certain way.

Big companies, like GE and Siemens, are investing billions in this technology, but are facing competition from startups who may be more agile, attentive, and better suited to deal with changes. This is a learning process for the organizations themselves, causing them to change their entire maintenance paradigm, so their needs must be top priority. Where large companies could have layers of bureaucracy and approvals, startups have a shorter chain of command that provides a faster response rate. In an era where markets, regulations, and costs change in an instant, you often need to think like a startup.

What’s next for Precognize and the IIoT industry?

As for the future, we are looking to develop a disruptive business model for the industry. We want to engage with insurance companies to offer predictive maintenance as a means to reduce the cost of risk. This way, plants gain reliability due to less shutdowns and failures, and hence should expect lower insurance premiums. In addition, insurance companies save huge amounts by not paying for the preventable damages, especially those related to third parties and business interruptions.

The industry is beginning to use these technologies, but it will take a few years for companies to adapt to these changes, as the entire way of thinking needs to change before becoming mainstream. Companies who want to be ahead of the curve should jump on these technologies now.