Europe’s SMEs (Small and Medium-sized Enterprises) account for more than half of the continent’s jobs but are now reeling from soaring energy costs. A new AI–driven platform will deploy ‘agents’ to crunch industry data, identify new solutions, and keep Europe’s factories humming.

Article by Massimiliano Saltori


There’s an old saying in Europe that small businesses are the backbone of local economies. On paper, that seems reasonable. In the EU, micro, small, and medium firms – meaning companies with fewer than 250 employees – account for as much as 64% of total employment.

Besides the promise of a lower tax burden, their popularity is usually attributed to their inherent “flexibility”. Indeed, these small enterprises do tend to be less bogged down by layers of management and rigid processes compared to larger firms. It’s what allows them to pivot quickly when the market shifts after a crisis.

However, Europe’s SMEs, especially those in the manufacturing sector, now face soaring energy costs and sluggish productivity gains, as their size often prevents them from having the resources to improve in both areas. This is especially true for micro and small firms, which alone employ half of all EU workers.

As the geopolitical climate evolves rapidly, Brussels is increasingly turning to AI to optimise production and energy use for businesses of all sizes. In this landscape, the Dynamic Energy & Process Management Platform, developed within the FLEXIndustries project, stands out as a promising AI-driven ecosystem for orchestrating various industrial processes – from the automotive sector to the pharmaceutical industry.

We spoke with Project Manager Nicolò Bertozzi and Software Engineer Anna Geraci, both from Fondazione Links, one of FLEXIndustries' partners, about how this modular platform is redefining factory floors. The foundation, active in the field of applied research and technology transfer, is a joint initiative between the Politecnico di Torino and the Fondazione Compagnia di San Paolo.


What is Link Foundation’s role in FLEXIndustries?

Bertozzi: Our role begins with collecting, managing, and storing data, both from consortium members and third‑party sources. That data is then used to build simulations and optimisations of industrial processes and ultimately integrated into an overarching strategy. So, we provide these models with all the necessary tools to run on the platform, and then we coordinate them via an orchestration mechanism.

Geraci: That's the core of it, yeah. We work on the industrial digitalisation side, plus all the IT heavy lifting. Essentially, the platform's back-end and infrastructure.


How does data exchange take place?

Bertozzi: We use smart sensors provided by third‑party vendors. We don’t install them ourselves, but once they’re up and running, we process all the raw feeds. Right now, much of the data comes from three partner factories: Suanfarma in Italy, Fertinagro in Spain, and Ford in Turkey. The raw data we collect is then stored in a self-contained environment, so it can be delivered on demand whenever a module or user needs it.


That must be a lot of data. How do you keep it safe?

Geraci: It’s pretty straightforward, actually. When you send in your data and then you want to pull out the results, the platform first checks who you are, keeps a full log of what’s happening and handles all the behind‑the‑scenes services. Both developers and operators are aware of what’s happening and can trust the system every step of the way.


Can you walk us through a typical process a user might request?

Bertozzi: Sure. Let’s take Suanfarma’s pharmaceutical plant in northern Italy, for example. Say a user wants an analysis of the energy produced by their solar panels versus the power consumed by a fermentation line. First, we gather all the raw readings. That comes from the sensors on PV arrays and fermenters, for example. But a lot of data can also come from weather feeds or market trends for energy price forecasting, all available online. The platform could run them through data-quality checks, filter out bad values, and hand the clean data to our algorithms. The plant operator then receives the requested recommendations on the FLEXIndustries dashboard.

Geraci: So, the operator may get a suggestion like, “Store PV energy now and run fermenters using power from the grid while electricity prices are low.” That way, the plant will use the stored PV energy later, when grid prices are high.


What can you tell me about the Dynamic Energy and Process Management Platform itself? Is it something that already existed, or was it developed for this project?

Bertozzi: Well, the platform itself really stands on the shoulders of past EU research we worked on. We had been experimenting with these concepts in projects long before FLEXIndustries. Each initiative teaches us a lesson, so to speak.

Geraci: In fact, the core of the platform goes back to an earlier European project called Reclaim. In that case, it was about predictive maintenance, using data to decide whether an old industrial machine should be refurbished or replaced outright. So, we borrowed the Reclaim platform, called REPLICA, and expanded its functionalities for FLEXIndustries.


What should we picture when you say “agents”?

Geraci: Think of our platform as a team of specialist bots – each one tackles a small AI task and then passes its result along. They grab data from our database to run their optimisations. They’re called “agents” because this model follows something we call “AI agency”: instead of one big monolithic model like GPT, you have lots of small AI modules that collaborate.

Bertozzi: The specific module for each task is developed by one of the project’s partners, such as FBK in Italy, the CIRCE team in Spain, CERTH, or the University of Patras, both in Greece.


Seems like AI is delivering on its promise.

Geraci: Well, from a technical angle, we’re all still riding the whole GPT wave. But the focus for us, of course, is not to produce a flashy user interface. This is tailoring generative AI for industry‑specific tasks.


And you think AI will help Europe’s SMEs optimise their processes and energy consumption?

Bertozzi: Absolutely. For one, FLEXIndustries comes within the Green Deal, especially in terms of reducing our industry’s dependence on natural gas. Almost every company we work with is cutting back on gas, because while you can still negotiate electricity rates, you can’t escape gas dependence. But you can still reduce your consumption, for example, by electrically preheating water instead of burning fuel.

Geraci: And our platform itself is quite practical for these SMEs. No in‑house data teams required. The system does the legwork and delivers clear recommendations without adding complexity.




Photo credits: Google DeepMind via Pexels

Contributors: Nicolò Bertozzi, Anna Geraci

Project website: Homepage - FLEXIndustries
LinkedIn: FLEXIndustries
Mastodon: @FLEXIndustries