ARTIFICIAL INTELLIGENCE

Christina Cardoza: Hello and welcome to “insight.tech Talk,” formerly known as IoT Chat but with the same high-quality conversations around IoT technology trends and the latest innovations. I’m your host, Christina Cardoza, Editorial Director of insight.tech, and today we’re going to explore digitizing the manufacturing supply chain with experts from Relimetrics and iProd, but as always before we get started, let’s get to know our guests. We’ll start with Kemal from Relimetrics first. Please tell us about yourself and your company.

Kemal Levi: Hi, I am Kemal Levi, Founder and CEO for Relimetrics. We enable customers with a proven, industrial-grade product suite they can easily use to control and automate quality assurance processes across use cases with no code. And using our product our customers are able to build, deploy, and maintain mission-critical AI applications on their own, in conjunction with any hardware. This can be done both on-prem or in the cloud, and a key industry challenge that our product repeatedly succeeds in tackling is our ability to adapt to high production variability, which is commonly experienced in today’s manufacturing.

Christina Cardoza: Great. Looking forward to getting into that and how that is going to impact the supply chain or bring benefits to the supply chain. But before we get there, Stefano Linari from iProd, please tell us about yourself and the company.

Stefano Linari: Hello, I am Stefano, Stefano Linari. I am the Founder and CEO of iProd. iProd is an Italian startup founded in 2019 to create the first holistic tool designed for manufacturing companies of each size, accessible for free, and as a software as a service. Our user can leave tons of purely integrated software like ERP, Amira, CRREM, IoT platform and use just one modern cloud platform, our platform.

Christina Cardoza: Awesome. So, I wanted to start off the conversation just getting the state of things right now. Obviously a couple of years ago the supply chain was headlining in the news almost every day for weeks on end—just the challenges and the obstacles. But I feel like there’s been a lot of integration and advancements in the technology space, that those pain points we were feeling a couple of years ago we have been able to get over a little bit.

But I’m curious what challenges still remain or where are the pain points today. Stefano, if you want to talk a little bit about what’s going on at the manufacturing and supply chain level.

Stefano Linari: Yeah. This supply chain unfortunately is still purely integrated, especially for its more-than-medium enterprises where digital tools are not updated and easy to be integrated because they are legacy technology. We are far away from the concept of this so-called manufacturing as a service, where the manufacturing capabilities are accessible in a fluid way. This part of the ask for a highly integrated, multi-tier supply chain, able to digitally orchestrate and provide a custom-made piece optimizing cost, impact, and user resources.

Unfortunately, even on the other side of this supply chain, if you look at the OEM we face other issues. And the companies are not able to serve the new part of this for their industry that is the machine customer, where a product, digital product, is able to purchase autonomously spare parts and accessories from the OEM itself and even from third parties. For example, a turning machine that after digitalization can work a belt or a gear after several number of working hours. This is still far away from the reality.

Christina Cardoza: Yeah, you make some great points there Stefano, and one thing I want to discuss a little further is you mentioned a lot of the problem is that there’s still legacy systems in place, and I’m sure that’s creating a lot of silos that these machines can’t talk to each other. Data is not end to end.

So, Kemal, I’m curious from your perspective where are some areas that manufacturers can start digitizing aspects of the supply chain and how that’s going to help address some of the pain points Stefano just mentioned?

Kemal Levi: First of all, digitizing aspects of manufacturing helps to trace quality across the supply chain. As parts move along the supply chain, quality automation helps to identify anomalies before they get to the customer and risk downtime. So for the entire supply chain, and particularly for the OEMs, it is really important to trace the quality status of parts or products from a multitude of suppliers and also run data analytics to see which one is actually performing better and read out those vendors who are not performing well.

Now, digitizing aspects of manufacturing also helps to improve the bottom line. So as manufacturers ship products to their customers, they must identify issues with outbound transportation and logistics. So a magnifying lens looking at different points of the supply chain gives better visibility to improve margins, and in the case of the sectors that we typically serve to margins are often razor thin. So maximizing the number of items getting to the end of the manufacturing line that meet the required quality standards has a direct impact on the bottom line.

Another example is that digitizing aspects of manufacturing, helping to make better supply chain decisions and correlation of acquired data across the product life cycle—and this can be all the way from manufacturing to sales to service—enables continuous business intelligence. And a company that can trace quality in real time and do a better assessment on where quality issues originate can ultimately boost profitability.

Christina Cardoza: Yeah, absolutely. I’m glad you mentioned the quality-automation aspect of the supply chain. I feel like sometimes when we talk about supply chain challenges, we are often thinking about deliveries and shipments and getting manufacturing production out the door. But it also starts—it’s an end to end issue—it starts on the factory floor; it starts as you are developing these products, making sure that everything is high quality, that it can go out the door and can be delivered on time. So that’s a great point that you made, and then looking at the different points of the supply chain so that it’s really an end-to-end experience.

Stefano, I’m curious, as we look at quality automation and all of the different parts manufacturers need to be on top of in order to have this end-to-end digitized supply chains, what are the technologies that are being used? Or how can we start enhancing and optimizing supply chain efficiency?

Stefano Linari: From our side, all these things can start from the demand side. If we start to build intelligent machines that can be transformed in a machine customer, we can create a more predictable demand. We can avoid to rush, to produce spare parts and install it in a non-planned way, creating a simple condition to optimize the supply chain. So from our side in these months, in the last year, we are pushing this new part upgrade inside OEMs.

What we have created to support the OEM to handle a new generation of machines that we call “machine customer,” it’s to create a free and self-service interface in the cloud where each OEM can create their rules and their identity—the digital twin of every machine that’s built in a few minutes. Gartner in their last books name it, “When Machines Become Customer,” recognizes our platform as the first machine-customer-enabling platform in the world.

We are then creating the condition to digitalize the supply chain. Because when you speak about potential saving, entrepreneurs are interested, but they are engaged when you tell them about increasing revenue. And with our technology embedding new intelligence on board of the machine, we are transforming our production tool in point of sales. And this is a remarkable shift in the mindset of the OEM that can be easily understandable.

Christina Cardoza: So I’m curious, because we were talking about the legacy systems earlier in the conversation, is this a software approach that we can take to digitizing the supply chain? Or does there have to be investments in new hardware? Or can we leverage existing infrastructure?

Stefano Linari: We have to combine both, because for sure software platforms can make the interface and user experience simple, but we can’t forget that manufacturing tools and equipment, automatic warehouse and production machines are not yet intelligent enough to analyze their needs and try to simply the life of the end user and to the OEM. So we need a combined approach at the moment.

Christina Cardoza: Great. And of course when we are talking about adding intelligence and doing things like quality automation, AI comes to mind. AI seems to be everywhere these days. Kemal, you mentioned you were—you have an AI approach to being able to provide that quality automation and look at different parts of the supply chain. So I’m curious, from your perspective, what is the role that AI should be playing in these supply chain processes?

Kemal Levi: Well, AI in supply chains can deliver powerful optimization capabilities required for more accurate supply chain–inventory management. It can also help to improve demand forecasting, reduce supply chain costs, and this can all happen all while fostering safer working conditions across the entire supply chain. Traditionally the supply chain has relied on manual inspections and sorting.

So I would like to give an example that centers around smart inventory management. So this, this process—the inventory management process—can be labor intensive and prone to error, adding costs to the loss. So today AI-driven quality-automation tools like ReliVision can be deployed without requiring any programming skills or prior machine learning knowledge, and they can offer access to real-time information that can improve efficiency and visibility. Now, similarly, AI can also be used in conjunction with computer vision and surveyance cameras to monitor work efficiency and safety objectively, and provide data-driven insights for businesses to optimize workflows and improve their productivity.

Christina Cardoza: So do you have any customer examples? I know you just provided the inventory use case, but I’m curious if you have any customer examples that you can share with us: how, what problems they were facing, and how Relimetrics came in and was able to help them and what the results were.

Kemal Levi: A good example is renewable energy leaders which engaged with us to help them inspect their wind turbine blades before they’re released to customers. So, using our AI-based quality-automation and non-destructive inspection-digitization platform, our customer is today able to automate the inspection of phased array ultrasonic data and assess the condition of blades before they are placed in the field.

And the main challenge that our typical customer has is to digitize inspections, which is time-consuming and prone to errors, and improve traceability across their supply chains. And with our product our customers can rapidly implement AI-based machine vision algorithms on their shop floor, and they don’t need to write a single line of code while doing this, and they can share, train the models across inspection points and leverage existing camera hardware, irrespective of image modality. Whether it’s infrared, X-ray, or PAUT.

Christina Cardoza: I love the no-code approach that you guys are taking, because I know a lot of manufacturers, they see these benefits, they want to achieve them, but there’s obviously labor shortages happening in the area in their space, and they can’t always have the skills or be able to deploy these as fast but they’d like to get these benefits. So, love seeing how we can make it more accessible.

When you have these no-code solutions, who are the type of users that are able to implement some of these in practice? Do you need those engineers? Or is it really an operator or a manufacturing manager that’s able to take part in this as well?

Kemal Levi: Well, we would like to enable process engineers to be able to build AI solutions, and not only build but also deploy these AI solutions and then maintain them. So what we see is that maintenance of AI solutions can also be quite costly. So we are making it possible for non-AI engineers to be able to maintain AI solutions.

Now we can of course also serve AI engineers as well; we can help them just prototype their AI solutions faster and deploy them to the field. The maintenance piece, again, is typically an important aspect that AI engineers typically would like to transition to operators after they are successful in the field. And this is exactly what we do: we make it possible for maintenance of AI models and training of new AI algorithms for new products, new configurations to be done by non-AI folks.

Christina Cardoza: Yeah. It’s amazing to see how far technology has come, and how non-AI folks can be involved—especially since these people are the ones on the factory floor with the domain intelligence, so they can spot the quality issues or be able to train some of these models better than an AI engineer probably would if they don’t have that deep manufacturing experience.

Stefano, I’m curious, from iProd’s side, what are the solutions and products that you guys have on the market that are helping your customers in these different areas? And if you had any customer examples that you could share with us as well.

Stefano Linari: Yeah, we have several use cases of machine customers spreading from concrete industry, industrial filtration, and manufacturing. But I want to present you the most significant case that was done with Bilia. Bilia SPA is the third-largest turning center builder, and their machines are sold to automotive companies and manufacturers of consumer goods and a lot of industry where metal parts are needed.

Most of those machines you can figure out to be installed in a shop floor, even in small and medium enterprises—you know that in Italy, but in Europe in general, most of the companies are under nine employees. So you can imagine that no expertise in IT can be found in the customer side especially.

So we have enriched, equipped, this turning machine with an external brain so we can go—it’s in a panel PC, technically speaking, but we like to describe it as an IoT tablet to make them more friendly for the end user—and with this tablet we have two connections at the same time: One with the CNC of the machines, and then we can acquire real-time data about usage and consumption of resources. And on the other connections, usually Wi-Fi or forward dealing, we are connected to the iProd cloud.

This solution—it’s a bundled solution, because we have to provide security and trust to the end user that no sensible data about their process and their secret sauce to create the perfect piece are not exfiltrated. Then, in the cloud, Bilia—the manufacturer—with their process engineer and maintenance engineers, using a visual approach as Kemal defined before, so even in this case, no programmer, no coder is needed, but you have a wizard in the cloud where you can simply drag and drop spare parts and services from the Bilia catalog to conditions that can be simple rules: every 1,000 hours, please change the filters or fill the oil. Or forward-looking AI and ML that can predict more accurately what must be changed.

The main point when we start this project is, “Okay, but why does the end customer have to accept that the turning machine will ask him to buy something? I have spent €200,000 for this turn, and every day he asks for more money? Why do I have to pay?” And so it was a bit scary, but the customer not only accepted the recommendation, but they ask the machine for more. They allocated a dedicated budget to the machine itself—usually in the order of €200 per month, no big budget, but in the most efficient area.  Because under this level the machine can automatically place the order, and you receive a notification on your mobile: “Hey Stefano, in a couple of days you will receive the new filter.” Or new belt, and so on. For €50, €60, because most of the spare parts are cheap. But we try to estimate the cost of placing the order and processing the order, and this is never lower than €50 for each side.

So the end user knows that if the machine never stopped and by autonomously the needed the spare parts, consumable, periodic service, he is saving money. And probably the same items purchased in an autonomous way are even cheaper, because on the other side I have to spend time to answer an email, answer phone, send a contract, and blah blah blah. So what was something that at the beginning sounds very difficult to do because the no skill, not very digital guys—it’s a real market success.

Christina Cardoza: Yeah. And I’m sure that is a common scenario in the industry: not knowing where to start, being worrisome of getting started, how much it’s going to cost, how complicated it’s going to be, if it’s going to be wasted effort. So it’s great to see how manufacturers can partner with companies like iProd and Relimetrics to be able to integrate some of this and really make improvements in the supply chain.

One thing that comes to mind—and I should mention, insight.tech, we are sponsored by Intel—but we’re talking about artificial intelligence and the cloud and real-time capabilities and insights into some of these things that I’m sure that you guys are working with other partners to make this all happen end to end, much like your customers. Sometimes we need to rely on expertise from other areas.

So, curious about how you’re working with partners like Intel, and what the value of that and their technology is. Kemal, I can start with you on that one.

Kemal Levi: In our implementations we are taking advantage of Intel processors and Intel hardware such as Intel^® Movidius^™ vision processing units, and we are also often relying on Intel software such as OpenVINO^™ to optimize deep learning models for real-time inferencing at the edge.

Now in the case of quality automation or digitizing visual inspections, customers are very sensitive about computing hardware costs, and they really do care quite a bit about smart utilization of CPU, so we use the Intel OpenVINO toolkit to minimize the burden. And also as an Intel market-ready solution provider we have access to a large community of potential buyers of our product.

Christina Cardoza: Great. We always love hearing about OpenVINO. That is a big toolkit in the AI space, like you mentioned, taking some of the burden off of engineers and just being able to easily run it once and deploy it on many different hardwares. So it’s great to hear.

Stefano, I’m curious from iProd’s end, how are you guys working with partners like Intel, and what are the areas that their technology really helps the iProd solution be able to benefit customers?

Stefano Linari: At the moment we use widely Intel-embedded mobile processors, because even if we haven’t done a heavy workload on AI and ML, what our customers want is for sure to reduce energy consumption at the edge. You have to consider that each IoT tablet is installed on top of each production machine and in a harsh environment, so we need a fanless processor with high computing power and low consumption for standby.

We also use Intel connectivity for Wi-Fi, because we need connectivity that can be reliable in EMC, in difficult spaces where you have welding machines and robots with high power, and this is what we are using now. OpenVINO and new processor with the Ultra core—Ultra is also in our ladder. We are starting to experiment with these features to accelerate especially ML and AI models to predict the usage, because we combine in the tablet—I didn’t tell before—not only IoT data that comes in from the CMCs, but from the cloud we receive even a schedule of the next batch to produce.

And what we are trying to do is to forecast the production, because you have to combine how many working hours this model will do if I will win this deal. Most of the calculations have to be done on the edge, because the customer doesn’t want to move outside their company sensitive information. For the manufacturer for example that produces a piece for the aerospace industry or high-end machines—a supercar like Ferrari, just to name a brand—their technology that is inside your software of the CMT machines, it’s all, half of the value of your company, and you don’t want to transmit even to iProd this information; you want to process all the information on the edge.

Christina Cardoza: Yeah, absolutely. One thing I love about these processors and toolkits is that this technology, it seems to be advancing super fast every day. Some things that a month ago that we were interested in are now becoming reality, and manufacturers, sometimes they have trouble keeping up with all of the advances and getting all of the benefits. But with partners like Intel and these processors they’re really making new changes every day to ensure that we can continue to keep up with the pace of innovation.

I’m curious, Stefano, how else do you think this space is going to change? What do we have to look forward to for the future of the supply chain?

Stefano Linari: I agree with even what Kemal told before: what we see, it’s a digital continuum—from the machines to the OEM to the supplier of the OEM to create a continuum of information. Because we don’t want to spend time in the order process. This is the piece that is considered a loss of time, and Amazon and other online stores are driving the user experience. Because now B2B requests inspire and are driven by B2C experience in the day-by-day life.

The second main point that is pushing the digitalization and will became mandatory in the next few years, at least in Europe, will be the ESG regulation and the so-called Supply Chain Act. So a company in 2026 has to present the ESG report, so they have to account for the emissions that each process in the company generates, and the main focus is on the manufacturing side obviously. And with the Supply Chain Act you have to provide this information, not only through the ESG report to the public, but you have to share points and data with your customers in real time or near real time. This means that the supply chain must be heavily integrated in the next few years.

Christina Cardoza: Great point. And you mentioned sustainability earlier, where we were talking about how some of these things can help worker safety. There are so many different areas that we can talk about, and we’ve only scratched the surface in this conversation.

Unfortunately we are running out of time. So, before we go, Kemal, I just want to throw it back to you one last time. If there’s any final thoughts or key takeaways you want to add—what we can expect from the future of supply chain management? Or how else AI is going to continue to evolve in this space?

Kemal Levi: Well, I think, as I said before, there will be a lot of focus on real-time data analytics and correlating acquired data across the product life cycle. And this goes all the way from manufacturing to sales, to service, to overall enable continuous business intelligence and help to derive better supply chain decisions.

And I think looking to the future companies will strengthen demand planning and inventory management in tandem with their suppliers. There will be data visibility at all levels, whether it’s from in-house manufacturing, suppliers and logistic partners, or customers and distribution centers. The supply chain will no longer be driven by uncertainty in demand and execution capabilities, and overall it will be characterized by continuous collaboration and flow of information.

Christina Cardoza: Well, I can’t wait to see how that all starts to shape out over the next couple of years, and how Relimetrics and iProd, how the advancements and innovations you guys continue to make in this space. So I invite all of our listeners to visit iProd and Relimetrics’ websites. See how they can help you digitize the supply chain from end to end and really get that continuum of information in all aspects of your business.

And also visit insight.tech, where we will continue to keep up with iProd and Relimetrics and highlight the innovations that are happening in this space. Until next time, this has been “insight.tech Talk.” Thanks for joining us.

The preceding transcript is provided to ensure accessibility and is intended to accurately capture an informal conversation. The transcript may contain improper uses of trademarked terms and as such should not be used for any other purposes. For more information, please see the Intel^® trademark information.

This transcript was edited by Erin Noble, copy editor.