---
title: "Putting data back in context: a key issue for quickly making industrial AI reliable — Marsanne Studio AI watch"
description: "Contextualising industrial data through digital twins would enable measurable performance and maintenance gains at Saverglass and Air Liquide."
url: "https://marsanne.studio/en/ai-watch/putting-data-back-in-context-a-key-issue-for-quickly-making-industrial-ce2307/"
lang: "en"
---

[Marsanne Studio](https://marsanne.studio/en/)

[← Back to home](https://marsanne.studio/en/ai-watch)

[← The AI watch](https://marsanne.studio/en/ai-watch)

Opinion · 15 September 2026

# Putting data back in context: a key issue for quickly making industrial AI reliable

Original title, in French: “Remettre la donnée dans son contexte, un enjeu clé pour fiabiliser rapidement l'IA industrielle”

**Publication**

Journal du Net (France)

**Author**

[Ilaria Michelizzi](https://www.journaldunet.com/account/ilaria-michelizzi-216275)

**Published on**

15 September 2026 · read in the page structured data

## Summary

The article argues that reliable industrial AI depends less on the volume of data than on its contextualisation (timestamping, links to equipment, maintenance history). It describes the evolution of the digital twin, which is moving from a static representation to a dynamic system that integrates engineering and operational data and knowledge graphs. Two industrial examples illustrate the point: Saverglass, which deployed a "Digital Factory" to centralise its production data, achieved a return on investment in under twelve months and reduced its energy consumption by up to 30% in some workshops. Air Liquide, through its "Smart Innovative Operations" programme, uses predictive analytics and digital twins to monitor critical equipment, having detected anomalies several months before failure and, in one case, avoiding more than €400,000 in losses.

Summary based on the full text of the article.

[Read the article on Journal du Net →](https://www.journaldunet.com/intelligence-artificielle/1554823-remettre-la-donnee-dans-son-contexte-un-enjeu-cle-pour-fiabiliser-rapidement-l-ia-industrielle/)

## Figures cited

-   12 months
    
    Return on investment of Saverglass's digital factory project
    
    Saverglass, production sites
    
    > « Dès la première année, le projet a atteint un retour sur investissement en moins de douze mois »
    
-   30 %
    
    Reduction in energy consumption in some workshops at Saverglass
    
    Saverglass, some workshops
    
    > « permis des réductions de consommation énergétique pouvant atteindre 30 % sur certains ateliers »
    
-   3 months
    
    Lead time for detecting a turbine anomaly before failure, at Air Liquide
    
    Air Liquide, Smart Innovative Operations programme
    
    > « Le système a par exemple détecté une anomalie sur une turbine trois mois avant les équipes, évitant plus de 400 k€ de pertes liées à un arrêt et à une intervention d'urgence. »
    
-   400,000 EUR
    
    Losses avoided thanks to early detection of a turbine anomaly, at Air Liquide
    
    Air Liquide
    
    > « évitant plus de 400 k€ de pertes liées à un arrêt et à une intervention d'urgence »
    
-   7 months
    
    Warning lead time before failure in another case at Air Liquide
    
    Air Liquide
    
    > « Dans un autre cas, l'alerte est survenue sept mois avant la panne, permettant aux équipes d'agir plus tôt, d'éviter des arrêts coûteux et d'optimiser le cycle de vie des actifs. »
    

## Classification

**Country**

Worldwide scope

**Company size**

Large companies

**Industries**

Industry and manufacturing · Energy, utilities, environment

**Companies**

Air Liquide (deploys AI), Saverglass (deploys AI)

**Keywords**

digital twin, predictive maintenance, industrial data, industrial AI, energy efficiency

Entry written by Marsanne Studio from the cited article, with the help of an AI model (claude-sonnet-5). The rights to the article and its illustrations belong to Journal du Net and its author. Entry translated from French by Claude Sonnet 5.5.