Expert Talks with Paolo Dalla Tor: innovating the machine with apps, cloud and IoT

Paolo Dalla Tor's journey: from teaching to leading R&D

Daniele: Good morning, I'm Daniele Delle Case and this is Expert Talks, the Things5 podcast that talks about technology and smart products with industry experts. With us today is Paolo Dalla Tor, R&D manager at Icematic. Hi Paolo, how are you?

Paolo: Fine, welcome to Castelmac.

Daniele: Paolo, I know your career is a very long one, spanning more than 30 years. I'd like to really understand what your current role is and what challenges you face every day in that role.

Paolo: I'm the technical director of the ice division at Castelmac, which has been manufacturing and designing ice machines for more than 50 years.

In the last 5 years there have been really major changes, because beyond the impact of adopting eco-friendly refrigerants, which completely reshaped the product lines, we made a choice on the electronics/IT side to adopt the cloud.

This has profoundly changed the way the machines work and has added new opportunities for ice machines. Today, compared to the past, development times for the electronics/IT side have increased considerably; in some cases they reach 50% of total project time.

Daniele: In your nearly thirty years of experience, I noticed you started out in teaching. I'd like to understand how, in your daily work, that kind of experience helps you in your role as a manager, as director of research and development.

Paolo: It wasn't a very long period, but it was very intense. What stayed with me from that experience was the students' curiosity about understanding new things in the world. Now, even though I'm getting on in years, I've kept that same curiosity and attention to the changes happening in our HoReCa sector. So when we finish a project, we're not satisfied; we're already trying to think about what to do next and about the next challenges, also because by now there's no shortage of ideas. That curiosity the students showed has stayed with me, you know.

Daniele: Now, bringing that experience to your current role, which is also that of a trainer for your team, how do you keep these skills up to date, and how do you keep your team up to date so you stay competitive in a world that's evolving so quickly?

Paolo: Let's say that on specific topics we also run refresher courses, but my experience leads me to say that the winning choice is choosing the right suppliers, important ones that are successful in their activities and work. This also makes it possible, by osmosis, to pass on all the experience related to certain areas of technology.

Then, of course, it's important that if someone has this knowledge, they share it within the company. That's our primary goal: spreading the knowledge we acquire over time.

How research and development has changed from the '90s to today

Daniele: Tell me a bit about how the role of research and development has changed and evolved from the '90s to today.

Paolo: As I mentioned earlier, when I started in the '90s, even though I was an electronic engineer, I mostly dealt with mechanical and electromechanical solutions in ice makers. Little by little there was an evolution; the first electronic boards started to appear with fairly simple software, sometimes with no interface at all, and the development time for this part of the design was really small, about 10-20% of the total time.

Now our equipment is often compared to home solutions that have advanced electronics, color touch interfaces, cloud connectivity. While home appliances can make big investments because they have huge volumes, in our case we're asked to do the same thing, so to face the same technological challenges. We've accepted these challenges, even though it clearly weighs more on us because we don't have the home appliance market, and we're introducing similar solutions in our equipment.

Daniele: We know that today the product is no longer just a physical product, but has become a true digital ecosystem. How has your team's approach to project management changed?

Paolo: The software side is starting to take up a significant share of planning, so for projects that are increasingly complex you need to estimate the required time up front. Getting the timing wrong means not entering the market at the right time. So sometimes the difficulty isn't so much designing the mechanical part of the machine, but actually having an idea of how much time it will take to develop an app or a software solution and doing the proper checks.

How much do electronics and software weigh on the budget? The single-board choice

Daniele: If you had to give a percentage of the budget that's invested today in physical components and in software, how would you split that budget, and above all, how do you expect that ratio to evolve in the coming years?

Paolo: 2-3 years ago, in the ice division, we made the choice to unify the electronic solution into a single board. This allowed us to have horizontal solutions across all the ice product lines. If we'd had one board for each ice product line, we would have had really high development burdens and costs. That saved us somewhat, but in any case the share of the budget for electronics and software is now around 30-40%.

Unifying the solutions, which also benefits end customers, has allowed us to contain this cost.

Daniele: What were the main challenges you had to face in integrating hardware and software along the path of a digital product like an ice maker?

Paolo: Yes, as I said, we had six product lines, each with a different electronic board. The time had come to revisit the electronics, and we asked ourselves whether we should make new electronics or build a system. A system meant electronics communicating with an app that communicated with the cloud.

At that point it was easy to conclude that the more we simplified the electronics, reducing the number of part codes needed, the more effective, the more virtuous the system we were creating would be. So it was a considerable effort, because the development time for six pieces of software that were all new is significant, but the advantages we have now are so many that we can confirm it was the right choice.

Digital obsolescence: the app that rejuvenates the machine

Daniele: I'd like to think a bit about a product's lifespan, right? We know that a product like yours can stay in operation for at least 10 years. How do you manage the obsolescence of the digital product compared to the physical product?

We know that the digital product evolves much faster than the physical one.

Paolo: In the past we had the electronic board with internal software. It could rarely be updated, also because the machine was normally controlled by pressing a button and there were very few functions.

With the new board, we introduced an app. This app has many functions that an ice maker couldn't have before. So in practice the machine has the electronic board and the app controls the ice maker. This way, all the developments that come to the app rejuvenate the machine. In other words, even a machine from 3 years ago, if it uses next year's app update, will have functions it didn't have when it was sold. This allows the customer to always have an up-to-date machine with updated functions that doesn't age, or that obviously ages only mechanically. So this is also a winning point, because the customer will have a machine that gains value over time because it will have increasingly important functions that they'll be able to use.

External suppliers, patents and cybersecurity in software development

Daniele: Are these developments handled internally or also outsourced, and above all, what are the main challenges of working with external teams on these research and development projects?

Paolo: As we said, software is starting to weigh heavily both on the budget and on the company's know-how. We work with external suppliers who specialize in the various applications, and in the future we'll probably need to consider that all the know-how we put into it (because we have about 6-7 patents in the latest electronic board) will have to be protected somehow. Ideally, in the future the company would be self-sufficient, at least on the software side, precisely to be agile with changes, so to be fast, and also to preserve know-how so it doesn't become too public. At the moment, though, we work well with the suppliers we have, so there are no particular problems.

Daniele: We know that today this process of product digitalization is bringing new regulations and new challenges with it. In fact, we know the European Union is investing a lot in regulating cybersecurity. How are you dealing with these challenges related to cybersecurity and to these new regulations?

Paolo: In this case we have no alternative, we have to rely on our suppliers, although recently we've turned to external experts precisely to validate the

work carried out by the suppliers. It's a really important issue, even though it's a heavy burden for mid-sized companies because it requires a cybersecurity culture that they normally don't have. So the challenge has to be tackled together with the suppliers.

Technology and usability: the single-step rule

Daniele: Now, talking about users, we know that your research and development work is definitely tied to the people who handle sales and product marketing, and it very often also involves the end user in defining the requirements. How do you manage the balance between technological complexity and product usability?

Paolo: In general, we resist the temptation to make a video game. Many people develop video games with touchscreens, but that's not the point. There are big chains that have drawn up specifications for how an interface should be designed, precisely because a branch could be located, for example, in the United States and have an operator who doesn't speak English and has to use the machine. I've always thought, I won't say of a child, but close to it, when it comes to using the software. The software has to be easy to use and intuitive. The complexity has to be hidden underneath. So we always develop functions that have to be started by pressing a single button. A single step is a winner. Having more than one adds complexity, leads the customer not to use it, and to a system that maybe only gets used at 20%.

Daniele: Do you do usability studies?

Paolo: We follow some guidelines. The single-step rule is definitely important. We put the complexity into the algorithms underneath; we don't want to make life complicated for the end user, especially if they have little time, if they work in a kitchen, in a bar, and are in a hurry. Also because when there's a complicated interface, what you've done is probably wrong.

Daniele: Do you have focus groups with end users to get feedback?

Paolo: We started with some software and there was an introduction with some initial suggestions. By now it's been almost 3-4 years and we haven't received any complaints, so we think what we did was pretty much on target. Now we're revising the apps and we've introduced some new features that I hope will help our customers even more.

IoT in service: avoiding wasted service calls in 120 countries

Daniele: How complex was it, on the other hand, to introduce the IoT component into an established product like the ice machine?

Paolo: IoT is important. Consider that we sell equipment in more than 120 countries around the world. So if you look at the Italian situation, the local dealer who has to provide service might drive 40-50 km. There are places where they have to take a plane. Leaving the office to go repair a piece of equipment without any idea of what's wrong with it creates a lot of problems. It increases service costs, and the customer has a machine that can't be used for several days. We've made something that allows whoever goes out to do the service call to already have a clear idea of the job, so they bring the components needed to repair the machine and the tools that are required.

IoT makes this possible: it lets you monitor machines remotely, understand the type of alarm, the type of job to be done, and so it reduces service costs and machine downtime. Even with the app, which isn't connected via IoT but via Bluetooth or to the machine, the end customer can already tell whoever provides service what problems there are. This has really solved many problems we had in the past. In the past, someone would call and say "The machine doesn't work." You'd ask what the problem was and the answer was: "The machine doesn't work." There was a very heavy service problem, with visits being repeated 6-7 times until they managed to find the problem.

Daniele: I imagine this may be one of the arguments you then use to convince more conservative customers to integrate technology into their products. Are there other keywords you use?

Paolo: There's also the possibility of having online documentation for the machines being serviced, without having to go looking for a manual that the customer maybe lost a year earlier. And in any case there's the possibility of getting clear answers from us too. When they call service, we give them details on how the machine has worked over the last 24 hours, the alarms there are, and we provide immediate support on how to solve the problem.

Stepping out of the fence: why adopting the app first opens new markets

Daniele: From your point of view, as an expert on both the product and the market, are there unexplored areas of research and development that you think will make sense to pursue in the coming years?

Paolo: We're definitely tied to home appliances. If we hadn't made that leap toward using an interface other than the button, or rather a display on the machine, if we hadn't adopted the app, we would have been cut off from the future developments that are coming.

Artificial intelligence is knocking at the door; in some cases it's already present in some applications, so clearly that's where things will end up. We deal with people who don't want to learn how to use a machine; they want a truly minimal approach while still getting all the functions and performance the machine can deliver.

So in the future, having chosen to equip ourselves with a connection via smartphone or PC will allow us to implement all the functionality that artificial intelligence will give us. I always use the example of the "fence." We were inside a fence along with other ice machine manufacturers. Adopting the app first basically allowed us to step out of this fence: that means new sales opportunities, the chance to have new interested customers, and that's the future. We have to widen the fence to reach more of the requests that come to us from the market.

From testing data to the machine report card

Daniele: It becomes an element of differentiation from the competition. And when it comes to environmental sustainability, how central can IoT become in reducing a machine's energy impact, and do you as a company have metrics to monitor energy savings?

Paolo: The app is already used in our internal factory testing, because we want to collect a set of data to process in order to extract metrics on the machines' performance.

The same will be done with the machines on the market. Understanding, for example, whether certain machines we've designed deliver the expected performance in hot countries or in hot seasons is important.

We test some machines; we don't test every machine at customers' sites, so having an idea that in the Far East our machines behave in a certain way and meet the design requirements is reassuring. If we see a trend toward performance we don't expect, we can go back and adjust our specifications. In the future, it will be important to have the ability to collect data and process it.

Daniele: If you had to identify the areas where IoT has brought you the greatest benefits, what are they, and above all, what feedback have you had from the market on the adoption of these technologies?

Paolo: We haven't been using it for very long, but it's clear that an end customer feels safe if someone is watching over them and their machine. So having the possibility that, if a serious alarm goes off during the night, the service center can already call the customer in the morning and solve the problem is important, especially if the type of alarm is clear and allows a fast, decisive intervention.

Daniele: Do you have an example of how data allowed you to solve real problems in the market or to improve a production process?

Paolo: We depend on component suppliers, so component suppliers deliver components that may not fully meet the specifications.

There was a faulty supply of water valves, and through monitoring you could see the machine running at its limit because it wasn't being cooled sufficiently. The service technician stepped in right away, confirmed there was this problem, and we solved it. Especially if the customer reports a problem with how the machine is working and then it's actually a pressure problem with the machine, with the water in the plumbing system. In this case the problem can be monitored remotely.

Daniele: So it also becomes a point of connection between you and service.

Paolo: Yes, for us service is fundamental. When service receives a report, the first thing it does is look at the "machine report card," because in-factory testing, with all the input-output data from the sensors and the input-output of the loads, is recorded, and so we understand whether the machine left the factory correctly. We can then build a history of the machine over time, recording service work, recording changes; a software revision can be recommended (because software also evolves over time). We try to have a machine report card that lets us know its life.

Digital ecosystem and ROI: why IoT has become a must

Daniele: By now you've built what's called a digital ecosystem around your product, right? And what does it mean to have access to a digital ecosystem that governs the product's development from its very beginning?

Paolo: It provides much more information. Now you have a lot more data, you have a lot more certainty that a project was successful, it gives consistent results.

We test machines internally, we have statistics and we see that a model produces a certain amount of ice at that ambient temperature, at that water temperature. This testing existed before, but the data was in paper notes and it was impossible to do a quick analysis, because you could retrieve the data, but it took a long time.

The speed of access to data changes a lot, and when we have more data from the field, the analysis can certainly be improved.

Daniele: Within these projects you've been carrying out for some time now, how could you calculate or determine a potential return on investment, and what, in your opinion, are the KPIs that actually let you track whether the investment is paying off or not?

Paolo: These are new realities, and then a lot will depend on how AI and new technologies make an impact. At the moment, for structured organizations, I mean dealers

that have thousands of customers, it's very important, because it allows monitoring that would require high costs if it were done any other way. We can no longer say, "Do we choose IoT or not?" By now it's a must. You can modulate the data transmitted, you can limit it to alarms and some data or you can capture all the data, but I believe that if you want to provide a service, there aren't many alternatives.

Daniele: Did you face any resistance along this path of change? And if so, how did you handle it?

Paolo: We manufacture machines under our own brand. Initially, when the app was implemented, we published it as a proprietary Icematic app.

When the various dealers noticed the potential it had for service, even those who didn't carry the Icematic brand asked to have an app for service. So we created it. That means the choice was right, it was something the market was expecting, and so we've moved to almost 100% of machines leaving the factory with the gateway for connecting to the app.

How to monetize IoT: data included in the price and new services

Daniele: Earlier you mentioned that in the future there could be new services sold through data. Are there monetization strategies you're putting in place? What's your approach to IoT monetization?

Paolo: Well, so far we've been focused on our dealers who provide service, and so we chose to offer a service that's basically included in the cost of the machine, while still providing the minimum available data, which is the alarms. In the future it will be a matter of understanding how this sector evolves, that is, whether dealers, for example, after working on a machine, will want full data transmission for a period, say a week, to monitor the machine's efficiency. At the moment we're limited to the first intervention; in the future more data could be requested and provided.

Daniele: On this point, I remember that a few years ago we spoke and you told me about a project potentially tied to monetizing ice production?

Paolo: Yes, there are companies that produce ice and sell it in plastic bags, or dispensers for selling ice. Maybe we'll get to that, especially with chains, given the trend of tying the ice service to a previous purchase. We stopped on this point because we also prioritize hygiene. In the latest app, for example, there are automatic sanitizing cycles for the machine or cycles with ozone for sanitizing. Installing a machine in a public place would require a service that hasn't been requested yet.

Daniele: So the product seen from a servitization perspective isn't yet a real need in your market. In your opinion, how much is this data becoming an asset for your company, and how much is it really an asset for a manufacturer?

Paolo: We'd need to set up a structure that processes this data and creates models. When we have models, we'll understand how our equipment works in different environmental conditions or in different countries, and we'll understand whether there are aspects that need to be improved. There are examples of companies that removed components, for example in washing machines, because they discovered that in certain countries a cycle was never used. Through the models, it will be possible to understand what the market requires and to customize machines for certain markets or certain countries.

Generative AI and voice control: the lesson from the Alexa experiment

Daniele: And changing the subject a bit, we know very well that one of the most disruptive trends now coming to market is generative AI. What do you think about adopting generative AI in a product like yours, in processes, and above all, what do you think about conversational interaction? Could it be something that's part of your world?

Paolo: If they compare us to home appliances, and home appliances adopt certain ways of operating equipment, our customers will probably expect our machines to adopt similar systems too. Clearly the end point is voice control that lets you give commands to the machine without necessarily having to know how to use a touchscreen or an app.

Daniele: I know you've already run experiments of this kind with Alexa in the past. What did you learn?

Paolo: Voice recognition wasn't mature yet, and the costs were extremely high. I hope current developments will now make these functions more affordable.

Generally speaking, it's true that we try to find a solution for customers, but we have to make machines that can be sold. When the application is mature and reliable enough (because, remember, we sell in 120 countries), then it will be possible to integrate new functionality.

Daniele: With multilingual AI, because we know very well that the new models have this flexibility. How do you imagine future interaction with machines could be?

Paolo: Starting from the service installer, I expect they could receive a report from the board that says what happened, when and why.

Daniele: So basically, you expect a machine that talks to you and tells you exactly what it's feeling, as if it were actually a human being going to the doctor. Very, very interesting. Now, talking about other types of AI, have you experimented with anything predictive for maintenance or for other activities?

Paolo: No, at the moment there are algorithms in the software that flag a risk condition in advance. To do something much more sophisticated on the predictive side, you'd need to work on those models we were talking about earlier, to have a fairly significant volume of data in order to define criteria that make it possible to carry out much more specialized predictive actions.

Daniele: And as for the evolution of interaction with the machine, still using these technologies, how do you think it will evolve in the future from your point of view?

Paolo: It will be an experience that I don't think will be generalist, because every piece of equipment has its own characteristics, every component has its own characteristics.

Every machine is a system of its own, albeit within a range of performance and operation. If we have enough data to analyze the machine, we'll probably manage to get something out of it, but the model of my ice machine will probably be different from my competitor's model: either there's an AI that truly absorbs a huge amount of data, or else it will be something specialized.

Data security: the real risk is network access

Daniele: And regarding privacy related to data, do you think we should worry about the ownership of our data, about all those data-related aspects, or how should we approach this topic?

Paolo: First of all, we try to design machines so that, even if there were an external intrusion causing malfunctions, the machines have passive safety that doesn't create problems for the end customer.

The biggest risk is that someone gets into the network through my machine's Wi-Fi connection. That's the most serious thing. I doubt anyone is aiming to stop a bar from producing ice for spritzes. That's a lot of effort for a not-so-huge result, but clearly this doesn't depend only on us; it depends on whoever makes our software, the apps, the processors.

How to choose technology partners: reliability and a single system

Daniele: Moving more toward understanding your approach to partnerships and to the technology ecosystem: as a professional, do you prefer working with large companies, multinationals, or do you prefer smaller companies, maybe more flexible ones, that can support you more precisely on your needs?

Paolo: Every technology topic has its own answer. In general, it's not size that qualifies a partner, but how they work. We make an effort to find organizations that are reliable, that deliver on time and provide solutions.

As a company, we can't expect to have all the know-how in-house, so choosing the partner is fundamental. If you pick the wrong partner, you lose the market.

Daniele: How important is it that this partner, and above all its solutions, are interoperable within your ecosystem?

Paolo: The more integrated they are, the better. The reason we introduced the app right from factory testing of the machines is that we want something integrated horizontally from the beginning of the machine's life to its end. I prefer

to have a single system, not to manage the machine with three different applications. The customer wants this too. The system has to be a single one, because the people who work with us expect minimal complexity. That's our goal.

What the professional sector can learn from home appliances

Daniele: How do you see the change in the sector you operate in, the HoReCa sector? Are there trends you're seeing accelerate, are there interesting things you're seeing?

Paolo: I believe we're very much pulled along by the solutions that will come from home appliances. So thinking only within our HoReCa fence, which is very small compared to the home appliance world, would be a mistake. We always have to keep an eye on that market, with curiosity, to understand what can easily be transferred to our applications. Everything points toward customer satisfaction.

If we provide solutions that make things easier for the customer, make them waste less time, let them do things quickly and well, then success is guaranteed. I believe that in home appliances, where we're talking about huge numbers, whoever designs a project is probably pursuing the same goals.

Daniele: If you had to analyze what's happening in home appliances and bring it into the professional world, what do you think are the two or three trends that will actually start appearing in products in the coming years?

Paolo: Clearly it depends on the products. There are products that have their own complexity, and there are products that are simpler. I'll give the example of ovens that have huge touchscreens with menus and submenus. In fact, carrying out a task very often takes a long time, you need very in-depth experience with the machine, and sometimes you may not even fully know how those commands translate into an effect, into a result. AI should skip all these

steps. You should be able to say "I want to make crispy duck à l'orange," and the machine should interpret the request and set itself up to deliver the result.

It's a major challenge, but in our sector too, people are less trained, and this makes people less curious, they get less training, they throw away the manual the day after they receive the machine. So I have to take this into account too, and if I add functions, complexity to my machines, the customer mustn't see that as a problem. We'll probably get to a verbal approach with machines, because otherwise we'd have to install huge touchscreens to have all the commands visible.

Paolo Dalla Tor's proudest innovation and his advice for those starting out

Daniele: I completely agree; we're also investing a lot in this area because we believe interaction with the product should be a natural interaction, like with a human being, and we believe this is the future.

And one last question, more of a curiosity really: you've been working in this field for 30 years now; if you had to single out one of the innovations you've brought, or something that really makes you feel proud of the work you've done, which one would you pick?

Paolo: Not because it's the most recent, but because it really was a major effort: consolidating all the boards we had in the ice division into a single board and interfacing it with an app. I think that changed the face of our products.

Today there are also other competitors that use an app, but more for service activities. We did ensure service activities too, but at the same time we added functions that the end customer with a similar machine from a year earlier didn't have. This allowed us to expand, to step out of the fence we were in. I believe this really is a true change of vision. We didn't renew the machine by making it a bit better-looking, maybe with a few parameters, an updated board; we wanted to change the vision of these machines: not just undercounter machines, but machines that could give something more, more functions, and that could be used much more simply than in the past.

Daniele: Let me ask you one last question, which is more of a personal curiosity of mine. What advice would you give to a Paolo Dalla Tor who's just starting to work now and who, 30 years later, has to approach the world of work?

Paolo: Someone starting from scratch is unlikely to have all the knowledge, all the skills the market requires today, but the most important thing is exactly what I said at the beginning: don't stay inside closed fences, have constant curiosity about the world around us. By looking at the world, you can understand many things and get many ideas for the equipment we design. If we take an existing machine and just renew it without looking around, we won't get any results. You mustn't have prejudices. That's what lets you have new ideas. Many of the patents we've filed came from other fields that had nothing to do with our sector. We talked with people who made technical solutions completely different from ours, an idea came up, and we applied it to our equipment. Curiosity, no preconceptions, and then a strong desire to grow.

Daniele: Sure, the desire to grow is definitely one of the most important drivers in our world. Thank you, Paolo, for your time; it's been a real pleasure talking with you.