From Sensors to Strategy: How AI Unlocks Industrial Insight

The promise of AI starts with the data beneath your feet.

In this episode, we explore how connected load cells, long viewed as routine industrial tools, are becoming strategic assets. Jorge Truffin, CEO at Unified Cloud Sensors, is using real-time IoT data to power predictive maintenance, automate diagnostics, and even influence market decisions.

This evolution unlocks new possibilities across industries like agriculture, logistics, and construction, including:

• Transforming sensor data into pre-emptive maintenance alerts
• Reducing downtime through AI-driven diagnostics
• Detecting material loss in silos and supply chains
• Tracking asset movement to optimize operations
• Linking sensor output to commodity market insights

Tune in to discover how rethinking industrial IoT inputs can reshape outcomes—and build smarter, more resilient business models.

Transcript

Intro:
You are listening to IoT Leaders, a podcast from Eseye that shares real IoT stories from the field about digital transformation, lessons learned, success stories, and innovation strategies that work.

Nick Earle: So, in this episode of IoT Leaders, we are going to go on a journey. We're going to go on a journey with Jorge Truffin, who's the CEO of a company called Unified Cloud Sensors. Now, he operates in a very, very traditional industry, an industry that's been around for hundreds of years, which many people don't really think about.

And it's a $4.5 billion worldwide market, and it's basically, what are called load cells. Load cells, in case you don't know, very traditional pieces of equipment that weigh things from low-value weight to hundreds of tons.

And what Jorge has done is, he’s embarked upon a strategy of using IoT to collect the data to go from reactive to proactive to pre-emptive maintenance, but then to go further to create a series of AI agents that starts adding business value.

And you'll hear him talk in the pod where he's now going into, for example, in agriculture offering people, futures exchange financial information to tell them when they should start trading the product. It's an amazing journey from weighing, getting data, digital data, analogue into digital data, to being a fully managed services company offering financial business advice. He's based in the Czech Republic.

And he is an Eseye customer, so that's how we know each other. And I think the main thing that this does is it shows the value of IoT data around things in making AI models even more effective. And you'll hear me talk about that at the beginning, but there are increasingly a number of companies who are using the asset that they're sitting on which is information about things to actually go faster on AI than other companies who perhaps have information from the public data sources, they have information from their internal documents, but they don't have the information on things, and this is one of those case studies. So, without further ado, let's get going, and I'll hand you over to my chat with Jorge Truffin. Here we go.

Nick Earle: So, Jorge, welcome to the IoT Leaders Podcast.

Jorge Truffin: Thank you, thank you, Nick. Also, thank you for having me here.

Nick Earle: You're very welcome. You are, I have to say, up front, full disclosure, you are an Eseye customer, so thank you for that, but that's not what we're going to be talking about on this pod. This is a continuation of a theme that has really risen up lately, which is traditional IoT companies aggressively embracing AI to solve new types of problems, but also… well, traditional problems in new ways but also, along the way, reinventing themselves higher up the value stack. And it really shows the link between the effectiveness of AI and the data, and in this case, it's the data from things. So that's what we're going to be talking about. And the things in this case are things that, I must admit,
I didn't know too much about when we first met, and you have helped educate me, and they go by the name of Load Cells, which is about a $4.5 billion worldwide market.

So, before we get into the basics of load cells, let's find out a bit about yourself, Jorge. A quick overview of your background, because I know you've had many years' experience in connectivity and IoT, and even before it was called IoT.

Jorge Truffin: Yes, well, we are coming from weighing industry, so we have more than 30 years of experience in the weighing industry.

The load cells, as you mentioned previously, is the element that, when it's under some product, or under some weight, has a linear signal coming out, electrical signal. So, what we do is we transform the force, or the mass into an electric signal that can be measured. So that is the load cell. And we have been in this business selling load cells, selling indicators, and selling solutions for weighing industry more than 32 years.

Nick Earle: And I remember when we first spoke, and I was asking you, giving examples of it, what I didn't realize is that you could be, you know, in the weighing industry, as you said, you could be weighing things that are very, very light, like in a laboratory.

Nick Earle: So very precise measurements through to, I don't know, a 32-wheel truck, or something even bigger. So, you're weighing everything across the spectrum. As you said, it's the whole weighing industry across all verticals, and it's a big global market, isn't it?

Jorge Truffin: Yes. We are more focused on industrial weighing. Industrial weighing, it's not so much related with laboratory measurements, it's more related to the practical aspects of the industry. So, yes, we have load cells with capacities of less than 1 kilogram, for example.

And then we have load cells with capacities up to 600 tons. So, we can weigh huge weights, and we can weigh everything in the middle, okay? So, from silos, from dozing machines, from track scales, so it's a huge amount of weighing systems.

Nick Earle: Great. So that's the layout of the industry and your background. I know recently, you formed a new company out of the company that you've been in called UCS, Unified Cloud Sensors. We'll get onto that.

Nick Earle: Because that's part of the journey for what AI can do for your business. So, let's, first of all, let's bring it, make it real. Make it real for customers. With load cells, what is the problem with it in terms of the data? You say you turn it into an electrical signal, but with load cells, do you traditionally, have you been storing the data over time, or is it just temporary storage? So, you take a measurement, you do something with it, and then you throw the measurement away?

Jorge Truffin: Yes, just normal, the weighing systems, any kind of weighing has been always produced in order to measure this weight.

And then, maximally, maybe, to make a kind of feedback. So, weighing was used for a momentaneous situation, okay? And now what we find out is that this data, which is very relevant in the moment of weighing, then it's not used anymore, it's simply lost. And that is when we decided, in Unified Cloud Sensors to get into it. And this data is gold. This data is important and relevant, so we do not want to lose this data. We want to use it later for more proposals.

Nick Earle: And why is, you call it gold, can you give me an example of, hopefully we'll cover several examples, but a first example of why keeping the data is gold, as you call it. Why is it valuable to keep the data? Can you make it real in terms of a use case?

Jorge Truffin: Well, yes, the data can be used for several purposes.

For example, for the technical aspects, maintenance, for example, this data is relevant because once we are keeping this data of this weighing system we can say that later on, when this data starts to change, start to have a different behavior, the system, we can detect the source of the problem and the problem itself, okay? So in this case, we say that all this amount of time that the technical staff needs to determine what is the cause of the problem can be detected much earlier with the analysis of the past data. So, for that reason, or for that… to use this, we use machine learning, for example, algorithms to detect what is the probable cause of the problem, okay? So, this is why the data is gold, because normally, you would have to spend hours, maybe more than one hour to detect the cause of the problem, and in this way, you can detect it much, much earlier. So this is one of the reasons that we say that the data is gold, but also for the end user, because the end user is the one that owns the asset that is being weighed, and then in that case the user, the owner of this asset can, exactly determine when this asset has been increased or decreased by the… for example, in the silo, the use of this material has been going into the silo, or has been going out of the silo, so he knows exactly what is the flow of the material. And so that is an important data for the owner of this material, yeah?

Nick Earle: And as I understand it, and I've undergone an education process, as I mentioned previously, I was looking at truck weighing systems. You drive over a metal plate and measure the truck. There are various reasons for measuring it, and we'll come on to the issue of the contents of the truck. But there can be… I think I'm right in saying that we're not just talking about one load scale. I mean, I believe that a truck… a weighing for a truck could have up to 16 load cells underneath it, and I guess when you're talking about monitoring and predictive maintenance one of the challenges for a human being, I guess, is trying to work out if there is a fault. It could be a fault anywhere on… you have to track down, first of all, which load center. And also, the cost of, if you have to do reactive maintenance, I guess that takes a long time. I mean, a person has to go out there, find out which of the 16 is faulty, and then fix it, so you have a lot of downtimes, which is expense. Have I got that right?

Jorge Truffin: Yes, exactly. Normally, the track scale has something between 4, 6, 8, maximally 12 load cells. 16 load cells are huge.

But normally, it's about 8 to 12 load cells.

But even in that case, it's very difficult to detect who is the source of the problem. So, what we really do is that we monitor each load cell, so we know the behavior of each load cell at all times, even when the track scale is not… the track is not on the way bridge.

And in this way, we follow some sensor signals, like the zero shift. And these are signals that are telling us the health of the track scale. So, we know exactly that some car… some of the load cells, some of these 8 load cells, is having some problems on the long run, because normally the problem is not immediate. It's normally… it's a long-term problem that is being added step-by-step.

Yes. It's very difficult, it's very difficult to detect the source of the problem, because it's a very slow motion of the problem, yeah.

Nick Earle: Yeah, yeah, yeah. And, and without the continuous monitoring and predictive maintenance. As you say, when things start to go wrong, they typically don't go immediately wrong, they go gradually wrong. You can predict, and therefore take pre-emptive, proactive actions. Without that, how did the industry work? Because that's key to the business case.

Is it that they would wait until it was clearly a problem, and then they'd have to call out a technician to arrive on site? Is that the way the industry… because it's a multi-billion-dollar industry, right? So, is that the way the industry has traditionally worked?

Jorge Truffin: Exactly. Normally, the client detects some problem, and then calls to the service company, and the service company normally doesn't know what is going on, because they are informed that something is wrong, but they don't know exactly what is the problem, so the technical personnel has to go to the site to check what is the problem, so you have to go, travel, then spend some time checking and discovering the source.

And the problem is that normally, when you arrive to the place, normally, there is no problem, because that is the typical situation. So, then you not necessarily have to solve the problem, okay? So, exactly, you have to solve a problem when it already is existing. So, that is the big problem.

Nick Earle: Oh, that's interesting the way you phrased that, actually. In order for the engineer to service technician to diagnose it. It's got to be completely broken. They can't diagnose something that is on the way to be broken, so you're solving two problems. You're solving speed, but you're, early recognition of a problem, an automated diagnosis.

And I guess, that then all translates into ROI. I wanted to ask you another question, because it is fascinating, a really traditional business, a hundreds-of-year-old business. The, is it just the weight, or if the plate is not… I call it the plate that the truck sits on, drives onto, if it is not completely level, does that also affect the measurements?

Jorge Truffin: Exactly. That affects a lot, the measurement, because the forces are not spread evenly on the load cells. So at the end, you are just pressing one or maybe several load cells more than the others, so this is creating a disbalance that is normally… maybe can be partially solved, but then you don't have the repeatability that you need. So, in fact, which is another of the source of the problem, and also the basement, because normally very heavy trucks just making a big pressure on the basements, where the load cells are installed. So, also, the failure of these basements is another source of problem. But all this can be solved, can be detected on time by monitoring these load cells, and that is what we are really doing. So monitoring is the essence of solving even this kind of problem here.

Nick Earle: So your sensors, as part of UCS, the Unified Cloud Sensors, name of the company.

The initiative is not only pure weights, but you can actually look at the incline the angle as well, and so you're correlating the data using, I guess, algorithms to actually really track down what's really happening, because it's a very… there's a lot of parameters in play, multiple sensors, the angle of the plate, different things are happening, and that's a service that you are monitoring and then saying, this is what I think is going on.

Jorge Truffin: Exactly. Yes, we can monitor even more sensors, okay? So, temperature is another relevant sensor, because temperature is correlated in the weighing industry with many things. So, exactly what we do is for our client that already has a track scale, imagine that the client is already having a track scale, we can approach.

And we can, even in the existing situation, we can start to monitor the track scale in a flexible way. If the loaders are digital, it's easier. If the loads are analogue, we can digitize these load cells. And, keep them, or change the analogue load cells for digital. We know that most of the load cells in the market are analogue, so one of the things that we're offering our clients is digitizing the existing load cells, so we don't need to change the load cells. We simply use a digitizing junction box to digitize each load cell, and then do the monitoring, as said previously.

Nick Earle: Now, I mentioned at the beginning. Sorry, if we can go into a very selfish advertorial from an Eseye perspective, but it is relevant to this story. I mentioned at the beginning that you are an Eseye customer and have been one for a while, and I want to get… in line with all of these podcasts, we're always trying to get to the business case.

I think that, from my knowledge of you as a customer, is that because you've got to transmit the data back to your service, your platform, you decide… you took a decision early on, not to use the client's network.

Because I'm sure there are networks, if you're in a yard, of a factory yard, or wherever, grain silo. You decided to use cellular, and could you just talk me through why you decided to do that, and then, just very quickly, why you thought that, given that you're based in the Czech Republic, but you have global reach. What Eseye gave you that perhaps some of the solutions didn't give you in regard to solving that problem?

Jorge Truffin: Yes, the idea of not using the client's infrastructure, communication infrastructure, was based on security, okay? Security from all sides. Security for continuous work of the system, so we want to avoid that the IT department of the client may disconnect this communication, but also, we wanted not to use the infrastructure of the client that we don't know what is the level of security that this client is having. So, our level of security is very high. We are using by the way, Eseye cards, so we're using your SIM cards.

And the level of security, we know that is warranted, so the data coming from our device, from our edgy device, we know that it's secure reaching our cloud infrastructure. Then there is another level of security.

And this gives us a very high level of confidence of the transmitted data. And thanks to the cooperation with Eseye that we have been from the beginning, right now, we are able to reach frankly speaking, almost any part of the world, if I can say. We are just working not only in Europe, but that is the main market right now.

But right now, we are trying to approach Latin American market, where we already know that we can work with reliability, and then the technical support in Eseye has been, let's say, for us as a company starting in this business, and on that moment, without big knowledge of this technology, was great. So, in this case, I want to thank you Nick in this way, because it has been a really great support for us.

Nick Earle: Well, thank you, and I think for what you're describing is a very common requirement, which is a single product SKU. So, I think what you just said is that you want to… you're based in the Czech Republic, you're expanding Europe, want to go to Latin America.

For example, Brazil, there's a lot of regulations to do with connectivity in Brazil, but the idea of what we can do is give you that single card, single eSIM, but then actually enable you to create a sealed device that can be shipped to, for example, Sao Paulo, Brazil be used turned on, and it will… it not only will connect, but it will connect in a way where you don't fall foul of the Brazilian IoT regulator, which says after 90 days, you have to leave, swap the SIM card over. So, the single product SKU is, in order to take something and then scale it globally, is one of the key advantages. So, let's leave the Eseye advertorial. And go to, the next case study, if we can, because this is a journey we're going to map out now, because it's going to ultimately, spoiler alert, it's going to ultimately end up with AI agents.

And we're on the path to describing that. So, on the second one, you were telling me about you mentioned the word silos before, grain silos. So, the second example is that somebody who owns the product. So now you've got somebody, it's their own grain silo, they're not taking a truck to somewhere else, and it gets weighed, but it's their own grain silo.

And you were telling me about the issues to do with monitoring. It's not just about weight when it comes to how much wheat, or barley, or whatever it is goes in the grain silo. It's quite a complicated process, isn't it?

Jorge Truffin: Well, yes, the product can be… because, as I told you, verticals, in our case, can be in agriculture, can be in chemical industry, and agriculture is one that, in this case, the silo with this kind of grain, it's very relevant in many markets.

And we realized that monitoring the silo contents, which is a huge amount of money inside, if you realize that the silo can hold up to hundreds of tons, for example. It's relevant to keep control on what it is on the silo, when the goods are going in, and when they are going out, so you can have absolute control of your assets.

And we find out that it's relevant that, we find out that in many applications, there is a lot of non-technical losses, which is, it's a problem to have this, this, unknown loss of material that you cannot detect at some point.

Nick Earle: And I have to Jorge, I have to jump in, because you used that phrase when we first met, and I was smiling, if you're watching this on video, you just saw me smile. I love the phrase, non-technical losses. I would call it theft, but of course you can't possibly call it that. It reminds me of a you know, phrases that people use for things of different words, like, you know, is it, SpaceX? Rapid, unscheduled disassembly, which means explosion.

And a non-technical loss, basically, I would call it… you can't call it theft, you're in the industry, I'm going to use the word theft just to explain to our listeners. So, you've got the hundreds of tons.

It's not just a weight problem, what you're describing, and I guess a humidity, a temperature, because it's a living product, if you like, you know, it's foodstuffs.

But also, there's the in and out. So now you're broadening to the process, right? You're looking at the… how much goes in versus how much goes out. And that's a difficult problem, isn't it? Because now you've got a tanker, or whatever, big truck.

You've got a silo.

And you've now… you're now having to, and there will be some loss, there will be some… I guess some stuff still sticks to the side of the silo, and this is a much more complicated case study than just wearing a… weighing a truck, right?

Jorge Truffin: Yes, exactly. Normally, when new material is arriving into the silo, it's being unloaded from a truck.

For example, and then all the owners of silo knows that not always the values are matching, okay? So, there is always… and the non-technical losses, not always are intended to be theft. It can be, for example, material left in the transferring structure, for example. That is forgotten, simply. This is a non-technical, or it's a technical, but the issue, the main issue on this case is that we know that when those people that are the ones to keep this material from themselves, for example, the driver, they know that the silo is under surveillance, that is monitored, so normally the non-technical losses disappear, okay? So, this is the classical advantage.

But also, there are some other issues, like, for example, sticking material on the walls of the silo. This is one technical big problem that some materials are having, so some material is sticking on the walls of the silo.

And you have to check visually. Visually, what is the problem, why the amount of material is not moving out of the silo, and it's because it is stuck. So, this is something that we are monitoring, so we are monitoring what is the center of gravity of the silo, so we know exactly when the material starts to stick on the walls, we try to determine on what part of the wall is this material and how much it is. So, we can give the advice to the owner of the silo that is the moment to free this material. And this is done mechanically, correct? So, this way of cleaning the silo is a mechanical procedure, but it has to be done at the right time, okay? So, we are telling the owner of the silo that this is the right moment to clean your silo and then take advantage of the right moment.

Nick Earle: I find it interesting, that phrase he used, the center of gravity. I just visually had the picture of a silo. If it didn't stick to the sides, then the center of gravity would drop in a linear relationship with the volume of product inside the silo, but if it was sticking to the sides, I guess what you're saying is that the level, the surface, would drop at a more rapid rate than the center of gravity, because you'd still have weight on the sides, so that's…

Jorge Truffin: Exactly.

Nick Earle: They don't know that, because opening up a silo and looking inside it is a big process, which they don't want to do, and of course you might contaminate whatever's in there anyway. It's a dangerous job as well.

Jorge Truffin: Exactly, exactly. So you know that normally the material starts to stick in one of the walls, or in a wall, and this wall is, for example, in between two feet, which are just on the load cell, so you can detect where exactly, in what part of the silo this material is located, and then you can take a decision of how to free this material, yeah.

Nick Earle: All right, so now we've had a really good base education on what you do, and load cell industry, and we could talk about lots of different use cases. We've talked about two.

Let's see if we can, talk about the role of AI and the future. So, let's, pivot, as we say.

Now, if we go back a bit, as I understand it, I asked you earlier about the… how the system used to work before the monitoring.

And you talked about when it went wrong, it would break.

And you call out the engineer, the engineer, I guess, would spend their time.

Well, like most maintenance engineers, they'd spend their time doing, I don't know, 6 or 7 calls a day, depending on the distances. So, it was a person, and then it was too late, because you really want to be pre-emptive. So is one of the basic ways in which you're using AI is to provide a level of proactive, pre-emptive monitoring services to your clients so that they avoid… you're actually avoiding the need for the break-fix process with the engineer. Is that the sort of base level of your AI work so far?

Jorge Truffin: Well, the AI can be used also by the service company, okay? So, what we really are doing right now, next year, we're going to implement the first AI agents. So, what you really need is to be checking continuously all the scales. Normally, one service company doesn't have one scale. They maybe take control of, maybe some hundreds, you know, scales. So, checking each scale in this way would be very, very difficult. So, what we are going to develop, what we're developing, is an AI agent that is doing this job. So, he's checking each element, each load cell, comparing the actual situation or the actual status of this load cell with the past.

And determine if this load cell and this weighing system are okay.

And base it on this analysis.

It's like streetlights, okay? If it is green, everything is fine. If it is yellow, maybe you should check some elements, because some of the elements are just moving strange, or something is outside of the border, and then if it is red, it's broken. Something is wrong and fix it. So, imagine that the service boss arrives to the company early in the morning, at 7 o'clock, and then he finds an email from his agent, and his agent spends maybe… 20 minutes doing this job for his 100 scales, and then he has the result. So, he's telling you what has been detected by night, and what should be fixed that day, or maybe the next day, or planets, if it is a yellow light.

Nick Earle: So now what we have is… I'm just, you know, going over the conversation that we've had. We've gone from pure weight, turn it into an electrical signal, weigh one parameter, the weight.

Then we talked about diagnosing it across 8 to 12 different load cells, and identifying which one it is, and then doing proactive pre-emptive maintenance and fixes, rather than get the technician out after it's broken.

Then we started talking about expanding to different parameters, like temperature, or the… whether the plate is level. Now what you're talking about is another change, which is, based on the agent, you're… yes, it's an agent, but from a business model point of view, you've now gone into a managed service, the way I would describe it.

If I'm a customer of yours, I am actually buying a managed service off you to monitor and, give me a report. Like you say, it's like someone, every night, inspects every load cell and all the information, and then when I come in in the morning, says “here's your report on the status of your assets, and what action you need to do.” Your agent has done the work overnight, so now you’re adding more value to your customers with this one agent. Have I got that right?

Jorge Truffin: Exactly. We suppose not to have only one agent, okay? We can have several agents, so imagine you have an agent that will be checking you, for example your track scales, okay? But as I told you, normally there are many types of scales that you can have in a company, so you can have dozing scales, you can have belt scales, and you can even have a different system, because for us, it's the same if we measure the signal from a load cell or from a pressure cell, from a pressure sensor. So, we can have different types of agents that may be checking different types of sensors, and that is what we can do, different sensors, and then correlate these different signals to find out some situations that normally you do not realize they are existing, okay? So normally, what we call is the sixth sense, okay? That is something that is based on experience, and this experience, we put it on a digital weigh, and then with the help of machine learning, which is a kind of artificial intelligence, and these agents, we can find out some data that is hidden, that is, that is not disclosed to the eye, and find out some correlations.

So, for example, we can correlate, and we already are doing it, in some cases, we can correlate, for example, PM10, which is the dust in the air, and imagine that a silo can be compromised. The enclosure of a silo can be compromised. So, in some cases, it is a very small fissures on the silo. So, we can detect, measuring the PM10, which is the particulate matter, on 10 microns, we can determine that this silo has its own fissure, and some material is escaping, okay? At the beginning, this is very small quantities of material, so we can detect this situation, and then determine that something is even wrong on the silo. So, in some cases, when the material is… because of its parameters of this material, we can even predict that the silo, which is a structure, it is nothing to do with weight directly.

But we can start to detect, for example, this kind of problem, okay? So, there are several types of sensors that we can measure, correlate them, and then determine if the information is valuable or not, for technical reasons.

Nick Earle: And you said something, very relevant, interesting, in the middle of that, was that typically, this would be done by someone who's extremely experienced, and may have 30 or 40 years’ experience, and so you go to the specialists and the specialist looks at the data and says, I've seen this before, almost like a doctor, right? You know, I've seen this before, I think it might be this, and that person is a very valuable resource, but a very scarce resource?

And also, it's not scalable. What you're saying is the agent is learning.

All the time, and the agent is increasingly playing the role of the human specialist for very complex, multifactorial, multi-data point, problems to diagnose exactly what's going on. So, the agent is becoming the specialist.

Jorge Truffin: Exactly, exactly. We all see that there is a lack of specialists, we see that. And then, you know, getting knowledge, it takes you a lot of years. So, the specialists are just maybe having 30 years of experience.

But now, you have to replace this person with somebody who has the same experience. So, this is a very big problem, and then what we really are doing is learning in a very fast way, the agent can learn this. If we have enough data, if we have enough data, and we are monitoring right now, from now, so then we can expect that in some time, we will have an agent that will have all this knowledge. So, what we're doing is training these agents with some knowledge that already is available, and we have it.

But the best solution is to train the agent for the specific situation of a weighing system that is existing. So that is why it's very relevant to start monitoring right now. It's very relevant, because all this data that is being collected now will be used for training this agent.

Nick Earle: I said at the beginning of the podcast, about one of the things that's becoming very, popular, hot right now, is a growing realization that IoT is essential for AI, and it hasn't always been the case. I think we've had IoT, and it's existed for years, and then there was AI, and they were seen as completely different things, and… and what's now happening in the last 3 or 4 podcasts that we've done, of which this is one of them.

People have started talking about not just training, you know, your AI agent can access public information which has been scraped off the web, and etc. And then you… and then it was like, oh, well, what you really want to do is train it on your company policy documents. So, people talked about agents, where it could access your expense policy, your employment policy, your behavior, standards of business contact. Policy, your customer sales funnels data.

But now we're talking about training the agent on data from things.

And it hasn't been talked about as much, but we had one from Volvo where the guy said, “my job is to connect 500 million things on Volvo's 150 factories.”

And we were saying, why are you trying to connect 500 million things? He says, because the more data I have from real things, the more I can actually ensure that the factory lines never ever stop.

And now what you're saying is, yes, you've got these agents, you've created the agents.

But the fact that you started with IoT, you have an asset, which a lot of, perhaps, your competitors don't have, you have connectivity to thousands of things, and you have knowledge of the correlation between the data points from the different sensors.

And you're feeding that into your agent. So IoT, basic IoT connectivity, is a fundamental enabler of the agent. And I absolutely believe that is the next iteration of AI is that it's going to be, what about all the data from the things, as opposed to all the data that is based in company policy documents, or is based in company databases. And it seems to me that you could then take that and go into, as a company, from a business strategy point of view, Unified Cloud Sensors could go into providing more than just a managed service. I mean, you could go into the area of telling people about, and I think he mentioned this previously, saying to them, well, the asset that you've got in that silo has a value.

And there's a timing issue as to when you sell it. You're not just putting stuff into the silo; you've got to sell it. But you don't immediately sell it, because it's based… there's a futures exchange, isn't there, on futures grain prices, for example. So, is that something that you're looking at, actually adding even more value? Because now what you've got is you've got the data.

So, you could actually… and you've got the scheduling, and you know the trucks, and when they're coming in, and how long it takes, and the process is mapped out. Is that something that you're considering moving into, adding even more value on the financial side of things?

Jorge Truffin: Yes, yes, you are right. Exactly. Once we know the data in value of weight, then we can evaluate this in money, okay? So that is relevant. So, in silos, for example, in agricultural products.

We can check what is the value, what is the market value, the stock value of this asset.

And then, just pairing this data with this, the actual data, with this real amount of material in the silo, then we can help you, we can help our partners to establish the strategy. The strategy, this is the asset, it's valued in so many tons, but also in so many dollars, or euros, or whatever it is. And then what is the next development? Because I can check what has been the past value of this same amount of tons, and what is supposed to be my next value of this asset in the next future. So, I can take decisions on if I'm going to be moving this material out immediately, or maybe a little bit later, so I can, at least I have the knowledge of how much money is on my silo, okay? Base it on real values and online is because what we do is that we get the data from the different sources, which is normally… or open source, or maybe they are paid, but it's not relevant for the client, because we arrange that. So, we can every… for example, in the EU, every Friday, there is an update of the market here in Europe.

In the United States, it's probably… it's more flexible, there are more sources of the stock, but for example, if we want to go here in Europe, you can, every Friday, know exactly what is the value of the agricultural values of different products, of different grains, and you can then take a decision, okay? It's possible to go to a stock market and then take the data from the stock market so you know every day what the actual value is of what you have on your salary. So, yes, exactly, we are trying to link the data from different sources. It can be stock exchange; it can be all the sources.

For example, construction is very interesting. We have a client in construction, and if he's having in his reactors he's using, you know, this special kind of material for floors, okay? So, this is expensive, and so we say, okay, what is the construction right now? What is the construction market in this moment? So, you can now establish a flexible pricing of your product based on the construction, the construction rate that is being right now in the market. So, this is another one: pricing.

Nick Earle: It's a long journey, isn't it? It's a great example, I think, of how IoT data plus AI, and in this case, agents, enables a reinvention of a business. Because you couldn't do that, what you've just described, unless you had the base data. We did one recently on a hospital, Turkish example of a hospital.

They've got the data from the patients, they've got the data from all the assets in the hospital, they've got the data from the staff in the hospital, and because they have those three sources of data, I called it, it's like a big Sims game, where everything's moving in a hospital, and they know exactly the correlation.

They can provide a whole new value-added set of services to optimize, for instance, to reduce the time… the waiting time in accidents and emergency waiting rooms, which is a big issue. But what you're saying is, you know, you think about this last 40 minutes, we've gone from weighing something like a truck.

To now, looking at futures exchange values for grain and becoming almost like a financial advisor to companies.

Because you have data that other people don't have.

And the people have the futures price every Friday of grain, however that futures market works, but they don't have the base data that you have, and the input-output, you know, trucks arriving in, trucks arriving out, the non-technical losses, as you call it.

And everything. So by having that data, you now have a much wider choice of what type of value you add in the future, and you're going to be a very different company in 2- or 3-years’ time than the company you were 2 or 3 years ago, and that's the promise of IoT linked with AI. And that's what we're trying to do in this podcast, is shine a light on these.

To, to say, you know, any business, I really do believe it's any business could actually totally reinvent their value proposition.

If they follow this similar journey of starting off with basic connectivity and then aggregating it.

Putting it into a layer, going from reactive to proactive to pre-emptive and then adding a series of value-added services on top, because you're then becoming a managed services provider with much more value than just giving them data on weight. It's a very different value proposition than 3-4 years ago, and if you can do it in load cells, it's a great example that can be applied to anything in the industrial sector, as you said.

Jorge Truffin: Exactly, and then every day that passes and we learn from the clients, because the owner of this data is the client, okay? We supply them all the data, the data is of our clients.

And then we try to learn what the client is… what the client really needs more. So, we try to learn, and then we try to understand his business, and then once we understood everything, we try to add these values, okay? So sometimes we must learn… we are not in agriculture only, but we can learn from any other industry.

And then we observe, we try to learn what the AI really needs and then aggregate more and more functions.

Nick Earle: So, you could take the learnings from agriculture into the chemical industry, where the numbers are even bigger, for example. Well, it's a great story, and we've come to the end of the timing, but it is a rapid transformation, a very interesting model, not just of AI but creating agents based on increasing value, and as you said, and then saying to the customer, well, what other sorts of value could we do?

And then you're now being totally customer-driven by what other problems can we solve for you. And it's only because you've got that base data, because everything is connected, which, given that we're a connectivity company, makes us feel good.

As well. So, Jorge, I’m going to leave it there, but I'm going to ask you one last question. I just thought…

Some of the companies that we talk to, they say, well, the problem is that there are certain industries that aren't seen as sexy, my words, not yours.

And it's hard to get people into the industry, but it seems to me that young people, it seems to me that that's not something that you're worried about, because you're actually… your goal is to get the agents to do all the work. The agents become the specialists. You're almost trying to eliminate the people.

Jorge Truffin: Yes, no, of course, there will be never a possibility of getting rid of all the people. Good people are always in need. The problem is that they are scarce, okay? And for avoiding this, what we really are doing is implementing these agents. These agents are really helpful. This agent can do things that normally you would need a lot of experience, and that is something that young people are missing. Young people are not missing knowledge, for example, of implementing AI, so that is good.

But they are missing the experience, so this experience is given by this machine learning, by this AI, and that is great. So yes, it's, the agents are helping a lot in this case.

Nick Earle: So, to put a bow on it, to finish, the agents are going to be training the next generation of employees.

Jorge Truffin: Maybe, yes.

Nick Earle: The agents are going to be training the people with the knowledge.

Jorge Truffin: Exactly, exactly.

Nick Earle: Interesting. A great way to finish. Okay, thank you, thank you very much for that. If people want to find out more, want to contact you directly, the company is, Unified Cloud Services, is that…

Jorge Truffin: Sensors.

Nick Earle: Excuse me, sensors, Unified Cloud Sensors. What is the website, or if they want to go on and find out more, is it UCS?

Jorge Truffin: Yes, www.unifiedcloudsensors.com

Nick Earle: There you go. Great. Well, thanks for being my guest on IoT Leaders, thanks for being an Eseye customer, and thanks for telling such a great story about how you're rapidly reinventing yourself and climbing the value stock. I'm sure that'll be very relevant to our listeners who are all over the world.

And every country that they're in will have similar problems and lots of load cells. It's a multi-billion-dollar industry, so I'm sure they'll be very interested in the story that you've told. So, thanks very much for being on the pod.

Jorge Truffin: Thank you, Nick, thank you very much also.

Outro:
You’ve been listening to IoT Leaders, featuring top digitization leadership on the frontlines of IoT. We hope today’s episode has sparked new ideas and inspired your IoT and digital transformation plans. Thanks for listening. Until next time!