Mecspe 2026: what the manufacturing industry confirmed to us about AI

The most frequent questions at Mecspe 2026 were not about AI itself, but about legacy systems, unused data, and process control. Aidia brought its proprietary solutions—AVA, HR Agent and Aidia Control—to the twenty‑fourth edition of Mecspe, the benchmark trade fair for the Italian manufacturing industry. What emerged from three days of direct conversations is not a list of impressions: it is a precise snapshot of where Italian industry stands today with respect to AI, and what is truly needed to move forward. Four themes repeatedly surfaced in conversations at the booth.

1. The problem is not AI. It’s the systems that already exist

Almost no one at the booth asked “how does artificial intelligence work?”. The real question—the one that came up most often, phrased in different ways but with the same substance—was another: “Does it work with the systems we already have?”

Manufacturing companies have built infrastructures over time that work—ERP, MES, SCADA, CRM, production control systems—and replacing them is neither economically nor operationally feasible. What they seek is a way to add intelligence without breaking operational continuity.

The numbers reflect this tension well. According to the Invind survey by the Bank of Italy (2025), 27 percent of Italian companies with at least 20 employees now use AI tools, a figure that grew by 14 percentage points compared to the previous year. A real acceleration. Yet among those who have adopted predictive AI, only 20 percent use it extensively; for generative AI, the share drops to 12 percent. The technology has entered. But it is not yet fully embedded in processes.

Aidia’s role as a system integrator responds precisely to this point: we connect legacy stacks to new AI capabilities without requiring an infrastructure overhaul. AI is layered on top of what already exists, extending its capabilities without replacing it.

2. The data exists. The problem is that it isn’t being used

The second theme that emerged at the booth is subtler than the first, but just as recurring: many companies did not have a data collection problem. They had a problem of dispersion and fragmentation of large amounts of data.

Years of connected production systems have generated enormous information assets—machine logs, process data, order history, operational KPIs—that remain largely silent. Not due to lack of willingness, but because there is still no intelligent layer that queries them, correlates them, and transforms them into actionable insights for operational decisions. The paradox is documented. The DESI index of the European Commission places Italy ahead of Germany, France and Spain in the adoption of e‑invoicing and cloud technologies. Yet, according to the same data, Italian manufacturing SMEs still lag behind in the use of data analytics tools and integrated management software. The data is collected, but not processed. The raw material exists; the transformation does not.

Building that layer is exactly the kind of work Aidia does as a custom AI software house: not standardized products, but solutions designed around the client’s specific data and processes.

3. Automate, yes—but without losing control

The third theme generated the most in‑depth and often the most candid conversations. The fear is not that AI won’t work. The fear is that it will work in an opaque way. For a production manager or operations director, delegating decisions to a system that cannot be inspected is not innovation: it is a risk. Traceability, verifiability and operational governance are not optional requirements—they are the starting point.

This also explains a figure emerging from the Invind survey: AI adoption increases significantly with company size, exceeding 50 percent in firms with at least 500 employees, compared to 23 percent in those with 20–49 employees. Large organizations have more mature governance structures, making the perceived risk manageable. Manufacturing SMEs need something different: solutions that make automation transparent and controllable from day one.

Aidia’s technical response to this need is multi‑agent systems: architectures in which specialized software agents manage complex operations in parallel—document management, approval workflows, production monitoring, reporting—coordinated and traceable. Every flow is inspectable. Automation does not reduce visibility: it redistributes it more efficiently.

At the booth, Aidia’s three proprietary solutions demonstrated this principle in action:

AVA showed how a corporate virtual assistant can integrate into existing operational workflows, supporting teams and document processes without requiring structural changes to the infrastructure.
HR Agent addressed the growing need to automate HR management—from candidate selection to internal employee management—reducing administrative workload without removing human control.
Aidia Control, in combination with Alkeria optical cameras, demonstrated how visual quality control can become an intelligent, continuous, measurable process integrated directly into the production line.

4. Data sovereignty: a prerequisite, not a feature

The fourth theme, strongly emerging especially in conversations with mid‑sized companies, concerns an aspect that the industrial AI debate often treats as technical, when it is in fact strategic: where the data remains. For companies with proprietary processes, confidential production recipes, or sensitive job‑order data, this question comes before any functional evaluation. It is not distrust of technology; it is awareness of the value of what they own. And increasingly, it is also a regulatory requirement: the European AI Act, GDPR, and growing attention to the digital supply chain are making data governance a concrete organizational responsibility, not a formal compliance task.

Aidia’s response is structural: our architectures are designed to operate within the client’s corporate perimeter, compliant with enterprise standards, without dependencies on unmanaged external cloud infrastructures. For many companies that stopped by the booth, this was not a secondary feature. It was the condition that made everything else possible.

What Mecspe 2026 tells us about AI in Italian industry

The themes that emerged over these three days converge toward a single interpretation: industrial AI in Italy is no longer a matter of vision. It is a matter of method.

The acceleration is real: the +14 percentage points in AI adoption recorded by Invind in a single year is a signal that leaves little room for interpretation. But the gap with major European partners remains wide: in Germany, nearly half of companies had already adopted AI tools in the first half of 2024; in Spain, by November of the same year, the share was 31 percent. Italy reached 27 percent in 2025, with significant acceleration, but starting from further behind. And about half of Italian companies surveyed by Invind do not plan to adopt AI tools in the next two years.

The problem is not the availability of technology. It is the lack of adoption paths built around the operational reality of those who must use it—with the systems they already have, the data they already produce, and the constraints they cannot ignore.

Aidia leaves Mecspe 2026 with clear confirmation of the manufacturing industry’s needs, confident in the concrete value of its solutions. To everyone who stopped by the booth, even just out of curiosity, we extend our sincere thanks.

Want to explore how AVA, HR Agent or Aidia Control can integrate into your company’s systems? Fill out the form: https://aidia.it/contacts/.

Sources:

Bank of Italy (2025). Annual Report 2024 — Box: “The use of artificial intelligence in Italian companies”, based on Invind Survey (February–May 2025). https://www.bancaditalia.it/pubblicazioni/relazione-annuale/2024/rel_2024.pdf.
European Commission (2023). Digital Economy and Society Index (DESI). https://digital-strategy.ec.europa.eu/it/policies/desi.
European Parliament and Council (2024). Regulation (EU) 2024/1689 of 13 June 2024 establishing harmonized rules on artificial intelligence (AI Act). https://eur-lex.europa.eu/legal-content/IT/TXT/?uri=OJ%3AL_202401689
BolognaFiere (2026). MECSPE https://mecspe.com/