Why AI in the Factory Needs a Modern MES First

Industrial AI: High Expectations, Few Concrete Results

In recent years, artificial intelligence has become one of the most discussed topics in manufacturing.
Predictive maintenance, process optimization, automatic recommendations: the promises are many and often fascinating.
However, operational reality tells a different story. Many industrial AI projects remain confined to proof-of-concept stages, never becoming an integral part of daily production. In most cases, the problem is not the algorithm.

The Real Limitation: Lack of Context

Factories are not short of data:

• PLC signals
• Machine parameters
• Cycle times and counters

What is often missing is process context

An AI model cannot correctly interpret:

• a machine stoppage
• a deviation from the standard cycle
• a production anomaly

if these elements have not first been modeled, structured, and linked to the real process.
Without context, data remain just numbers.
And AI cannot make reliable decisions.

The Key Role of MES

This is where the MES becomes central.
A modern MES does more than collect information; it:

• transforms raw signals into process events
• governs operational states and transitions
• connects data, rules, and production context

In other words, it creates the logical layer that allows you to go from data to decision.
Without this layer, AI has no solid foundation to work on.

MES as the Engine of Industrial Workflows

The real value of a modern MES today is not in screens or standard functionalities, but in its ability to orchestrate industrial workflows.

Workflows that:

• represent real factory processes
• evolve over time without disrupting operations
• integrate IT and OT systems
• prepare data for advanced analytics and artificial intelligence

Only an MES designed as a workflow engine allows AI to move out of the experimental phase and become truly operational.

SFC4: Born in the Factory, Not the Lab

SFC4 nasce in fabbrica, non in laboratorio.

Its foundations come from real projects developed together with major automotive players, where requirements are not theoretical but dictated by production: operational continuity, process adaptability, integration with heterogeneous systems, and the ability to evolve without stopping production lines.

In these contexts, the MES is not a support system but a critical production element.
Every architectural choice must handle high volumes, organizational complexity, and frequent changes.

These real-world experiences guided the design of SFC4, resulting in an MES that focuses on processes, workflows, and events, not just screens or static functionalities.

An MES Designed to Enable Analytics and AI

SFC4 was designed with a clear principle: process first, intelligence second.

Thanks to a workflow-oriented architecture:

• events are structured
• states are explicit
• context is always available

This makes it possible to naturally integrate:

• advanced analytics
• reliable KPIs
• AI models that are truly usable

Not as isolated experiments, but as part of the factory’s operational flow.

Conclusion

Artificial intelligence does not replace the MES.
It only works if the MES is designed correctly.

A modern MES is not just a data collection system, but the layer that transforms production into structured, actionable information.

It is on this foundation that SFC4 was born.