“It is not a ‘nice to have’, but a ‘must-have’” –AI-driven Solutions for industry plant optimisation

Ronja Schrimpf

Startup Stories

30 percent reduction of maintenance costs, 80 percent reduction of failures and up to 25 percent increase of profitability – With these numbers, it is no surprise ReliaSol is seen as a ‘must-have’ rather than a ‘nice to have’, not only by its founder Mateusz Marzec, but also by its growing customer base. For industry plants, ReliaSol can make a difference that draws consequences for the whole company, its employees and the environment. In an interview with 5-HT, Mateusz Marzec explains what makes ReliaSol unique and different from their competition.

Founder Mateusz Marzec
Founder Mateusz Marzec

“If we are able to predict failure …” What is ReliaSol?

“We developed an AI-driven predictive and prescriptive maintenance platform that firstly predicts critical machinery failures. Secondly, it identifies what are the causes and costs of that predicted failure. Thirdly, it provides all the information which is necessary in order to avoid this failure or minimize its consequences (prescriptive maintenance).” As Mateusz explains, it is not all about failure itself: “When we talk about production plants, the cost of downtime can be 250,000 Euro per hour on average. That is because with a failure, you are not able to produce and the amount of goods that are not produced together with costs of repairs are equal to that value.

If we are able to predict failure, for instance 24 hours before, we have the time to organize company resources, bring technicians to the plant and replace the part or parts which are going to fail. With that preparation, you are able to continue.”

“When you monitor all machines of a production plant …” What use is ReliaSol?

“Our approach minimizes the number of failures by 80 percent, costs of maintenance by 30 percent, significantly improves safety and also improves production efficiency from 20 to 25 percent.

“Our secret sauce and our biggest advantage is that we know how to analyse massive amounts of data efficiently. In order to analyse data and support decision makers, you need to have a lot of sensors data to dive deep into it and get insights.

With AI algorithms, we are able to find what data patterns usually occur before a failure.” For example, Mateusz explains, if a company had a certain failure before, AI can find the pattern that always happens 24 hours before each failure – like increased pressure and a drop in temperature in the same time – and identify a pattern, that ReliaSol’s software can look for in real time. With this, ReliaSol is able to identify a pattern of when failure is to happen the next time.

The technology implemented by ReliaSol is especially interesting for the energy and chemical industries, as ReliaSol’s website reveals: “Chemical and petrochemical plants are characterized by a multitude of technological installations, complexity of production processes and a large variety of end products. Despite this diversity, the common objective is to conduct processes in the safest, the most efficient and the most profitable way. Chemical installations are equipped with a variety of critical machines and devices: industrial furnaces, reactors, columns, tanks, silos, compressors, blowers, turbines, driers, centrifuges, pumps, extruders, heat exchangers, chillers, evaporators and much more.” – All of these are endangered by failure. ReliaSol could be the solution for predictive maintenance for chemical industry plants.

“The crème de la crème of what we do …” What are ReliaSol further advantages?

“What I was talking about is predictive maintenance, but we also provide prescriptive maintenance to our customers. Compared to predictive analytics, prescriptive maintenance is relatively new. The main goal in predictive maintenance is to show/indicate that something will happen, while prescriptive analytics goes further and analyses different problem-solving scenarios in order to find the best one for the company and prescribe them a solution with minimal costs and losses.

The crème de la crème of what we do is process optimization. With deep learning, we are able to build a model, a digital twin of the whole installation or even a plant. With this model, we can check what would happen if we changed certain factors in the production line. Our algorithms will check these different factors and show , for example, that by changing certain factors you can minimize the pollution or maximize the production efficiency or the quality of the product and so on – and all that in real time!”

“ReliaSol was the first company in the world that …” Who is ReliaSol?


“When we have production plants with hundreds or thousands of sensors, each sensor can send values each millisecond. In the end, we have terabytes of data. In order to analyse this data and find patterns, super powerful algorithms and huge computational power are needed.”

ReliaSol started in 2015 with the NVIDIA inception program, a deep learning and KI-Startup incubator program of the international producer of graphic cards NVIDIA. “Because of that cooperation, we have a technology which is able to predict failures super accurately.

I established Relia Sol with my PhD supervisor – Tadeusz Uhl, a professor for the mechatronics department. What sets us apart as a company is that our Technical Team consists of not only Data scientist, but has a strong background in mechanical, electrical and process engineering as well. This allows us to truly understand the problems our customers have with their machines and, very importantly, speak their language. In the current team, our Chief Technology Officer is Piotr Lipnicki, who has a great background in AI but he also understands machines. When it comes to commercial parts, our CCO is Kasper Huisman, who is trying to optimize our business model and puts a lot of attention in new areas of business. Who is also important is our CEO Régis Casenave, who is a former top Manager for an international electronic company. Another important member is our CFO Mark Elgers. We create a nice team together.”

“We are more and more recognized in Europe …” – How can 5-HT help you?

“Firstly, we want to improve our market visibility. This is crucial because we are building our recognition in other European countries like the Netherlands, Belgium, France and it's time for worldwide expansion. Cooperation with 5-HT will give us the possibility to reach more companies who are interested in implementing AI based technologies and improve their performance. We want to share our knowledge and experience and reach the right people in companies. Secondly, access to these companies would be helpful. It can be crucial that someone refers you to the right person in a company when it comes to getting an introduction and door opening.”

“I think we are the best because …” How does ReliaSol stand out from competition?

“The market is crowded, but our solution is the best in class. There is still a lack of knowledge when it comes to p     redictive maintenance. Predictive maintenance is also often understood in a wrong way. It is a buzzword – even if you only have a sensor that measures vibrations of a machine which can predict failures, people sometimes call that predictive maintenance.[1]  But this is not comparable with analysing terabytes of data and looking how everything interacts within this machine, how the environment and other machines interact with it. What we do is not that hardware approach but a data driven approach. This data driven approach is emerging quickly.”

Since it is difficult to compare ReliaSol to hardware approaches of predictive maintenance, Mateusz only compares themselves with data driven approaches. “I think we are the best because we automatized the whole machine learning process. If we have a lot of data, we have to prepare the data, solve all problems in the data like missing values or outliers, transform the data and use different machine learning methods and parameters. At the end, we can compare the accuracy of the models. Usually, when it comes to data driven approaches, analysts try to build the best possible model. In our case, we automatized this process. Thanks to that automatization we are able to check so many different options, that analysts would not be able to in a whole year what we are doing in one day.

Thanks to automatization, we are also able to decrease deployment time twice. Thanks to automatization, we decrease our implementation costs.” This means that ReliaSol is also more scalable, faster, cheaper and much more accurate than the competition – “And that is what counts for companies’ managers”, Mateusz explains.


“It is not a ‘nice to have’, but a ‘must-have’” – What are your future plans?

“In 2020, we moved our headquarters to The Netherlands to support our goal to grow internationally and make it easier to close deals in our new markets.

Our plans back then were to explore new markets, but COVID obviously impacted the whole business. Interestingly, while COVID may put investments on hold for certain companies, we also have conversations with many plant owners and managers that, because of COVID, now want to implement predictive maintenance as it allows them to monitor their operations remotely and have less employees on site.

I also would like to change the way people think about data driven predictive maintenance. People often are underestimating the power of this technology. It is not a ‘nice to have’, but a ‘must have’ technology. People should try it. It is not a matter of if, but of when they try it.”

5-HT Digital Hub Newsletter

Want the latest tech and industry news, events, relevant info from the ecosystem and more?

Subscribe to 5-HT Newsletter now Subscribe to 5-HT Newsletter now

Become part of the 5-HT Digital Hub Chemistry & Health

Exchange ideas with innovative startups and future-oriented companies in our ecosystem. We look forward to meeting you!