We are with Angelo Tracanna of Before Digital Industries and today we meet about digital industry, decision science, and decision making in your sector. I’m really glad to have you here, Angelo. Let’s start simple—can you tell our listeners more about Before and about yourself?

Sure. Before is a company specialized in industrial IoT, especially in predictive maintenance applications. It’s a relatively new field, and I like to say we’re pioneers. We’ve developed a patented algorithm that compresses data, and the key point is—we do everything ourselves. Our software, our hardware—it’s all developed in-house. We offer a complete platform called IoThing. Normally, if you want to monitor a machine with IoT devices, you need to buy the sensors, then integrate them with software to manage the data. That integration part can be a mess. At Before, you get a one-stop solution.

We have two business lines. One targets manufacturers directly—companies that want to monitor their own plants. But most of our revenue comes from machine builders. Big names like Danieli, for example—they use our technology under their brand. These are custom systems designed to fit specific machines and analysis needs.

We’ve been doing research in this area for eight years, and we’ve deployed thousands of sensors in plants all over the world—from Italy to the U.S., Bangladesh, Taiwan, Sweden, South Africa, and more. Right now, we’re expanding into India, which looks promising—if no more wars break out, of course.

Yes, we can all relate to that. One thing that really intrigued me is your ability to reduce the data load with your patented technology. Can you walk us through how that works?

Absolutely. Our core tech reduces data using edge computing. For predictive maintenance, you usually sample vibrations from machines. These vibrations can reveal the machine’s health. But vibration data is huge—if you sample 100 points for one year, you can generate 56 terabytes of data.

With our patented method, we can monitor 2,790 points over 20 years using only 8 terabytes. That’s the power of our compression. And since we use LoRa—a long-range radio protocol—we don’t need cabling. LoRa can reach 3 km indoors and 18 km in open spaces.

That’s impressive. But I’m curious—why not just license the patent and let others do the rest? Why did you also build your own devices and create an end-to-end system?

Because we had to! Back then, I went to a local shop to buy a LoRa sensor, and they didn’t have one. So we built our own. And we realized it made more sense, especially for our niche.

We’re not selling millions of units; we sell thousands. It’s not a consumer market. 3D printing helped us prototype and then manufacture housings and components. We now have our own production with multiple 3D printers. If a client needs a specific sensor, we just build it.

It seems like you’re naturally aligning with the servitization trend. You’re offering continuous value beyond the initial sale. Would you agree?

Definitely. Machine builders can actually earn more from service than from the machines themselves. But traditional business models don’t support that. A machine’s life has six phases, and most interactions with customers are painful: during the sale, delivery, and breakdowns.

It’s like a marriage where you forget the good times and only remember the fights. Servitization changes that. And it needs enabling tech—like our IoT systems.

We started with predictive maintenance, but now we also help clients design full servitization strategies—documentation, FMEA analyses, maintenance workflows. It’s a fast-growing part of our work.

Interesting. But let me ask something more critical. Industry 4.0 made big promises—smart factories, deep transformation. But often the outcomes are fragmented or shallow. What’s holding this back?

That’s a big topic. Industry 4.0 is mostly a European concept. It started in Germany in 2011. But when rolled out across Europe, it mostly came as tax credits. So many companies created projects just to get the incentives. In Italy, we know how to play that game.It helped a bit—new machines, new plants—but many people didn’t know how to use the tech they bought. There was also a market issue. The real drivers of innovation should be the machine builders. But they weren’t empowered or incentivized enough.Instead of replacing all old machines—unrealistic—we should have focused on revamping them. Machines can last 50–70 years. You need to modernize them, not throw them away.That’s why I prefer Japan’s Society 5.0 vision: distributed intelligence powered by IoT. For example, if a sensor detects your wall needs better insulation, someone can offer the service proactively, and you save money by preventing the damage.Europe can’t compete with the US in large language models. But we can win in IoT. The US shifted to services, Russia to raw materials. China and India are mega-factories. But Europe still builds. That’s our real economy, and we should protect it.

That’s a solid point. And our diversity across EU countries could actually be a strength in that vision. Let’s shift now to innovation itself. You’re someone who spans materials, systems, AI, and business models. What’s the next innovation step you see coming?

The next frontier is AI trained on real-world data from IoT devices. That’s where we’re heading. There’s been a huge focus on the Green Deal, but sometimes energy-saving isn’t even possible. What we should focus on first is saving resources.I often joke—people buy super powerful computers with embedded AI that can write poetry and generate art. And what do they do? Go to work to pay for the computer. We need to exit the AI bubble and ask: how can this enable real human services?One example—we’re working with Cube Labs to analyze tremors directly through vibration, using our Before sensors. Instead of relying on a doctor’s interpretation, we measure it directly, with raw, accurate data. That’s the kind of real innovation we need.

As always, you have a way of stepping outside the frame to solve problems. Let me close with the last question we ask every guest on The Node. How do you personally make decisions? As a person, a scientist, a business owner—what’s your process?

First, I believe indecision is often worse than a wrong decision. So I try to decide quickly. I gather facts, consult with colleagues, and ask: what do we know for sure? Then I decide.If the situation is complex, we use structured methods. One that I like is FMEA—Failure Mode and Effects Analysis. You evaluate risks based on probability, severity, and detection. You get a score and you know what to address first. It brings logic to chaos.