“Creating intimacy between business experts and data scientists”


After 39 years in the Air Force, Denis Mercier joined the management of the French industrialist Fives. For him, digital appropriation, especially of data, means wasting time in order to save time.

What similarities are there between the world of the military and the world of business in the appropriation of data? Denis Mercier, former Air Force Chief of Staff, now Deputy CEO of Fives Group, came to Big Data & IA Paris 2022 to talk about his experience in this field. To establish a culture of data, he recommends a field approach.

AdvertisingIs there a data culture in the weapon? It depends on the armies, some are more digital than others like the air force or the navy, but in any case, this culture of data is acquired through trust. This is what former Air Force Chief of Staff Denis Mercier explained during a plenary session on Big Data & IA Paris 2022. The one who was also Supreme Allied Commander for the Transformation of NATO (from 2015 to 2018) has been for 4 years the deputy general manager of the French industrial group Fives Group which designs precision machine tools and production lines for aeronautics. , automotive, rail, naval or construction. A company created in 1812, with a rich heritage of 200 years in industry, including 155 in steel.

To establish a culture of giving, Denis Mercier advises adopting a ‘bottom-up’ approach. We must not see data scientists as people above ground who do not understand business expertise, this intimacy between business experts and data scientists is built , he explained in a session entitled From the highest level of military command to industry 4.0, a story of transformation through data. The Deputy CEO of Fives Group recalls that it took some business experts more than 30 years to train. Conversely, if a data scientist leaves, we can recruit another one [NDLR : si tant est que le vivier de comptences en data science se renouvelle rgulirement].

Properly aggregating data, a complex step

If we want more digital in our professions, we need a strong industrial capacity , points out the former chief of staff. In the Navy, for example, a sailor understands that with more data, he can take on more responsibility, creating the conditions to do so himself. It is in this sense that intimacy with data specialists is extremely important, says Denis Mercier. This is first built by playing with the data, he notes. The data scientist must ask the business expert what are the right parameters, the right data to use, etc. This progressive approach requires the provision of a data management platform. The CEO of Fives Group cites as an example that of the French publisher Dataiku. It offers a studio that allows you to start very low, prepare the data, sort it and start getting into programming with pre-packaged tools. This is important, he explains, because it is the expert in the field who has the information.

Data can be extremely heterogeneous, with large amounts of data coming from many different sources. If we aggregate them correctly, we can create decision support mechanisms which, above all, do not decide for people, continues Denis Mercier, emphasizing the importance and complexity of this aggregation. Creating a global architecture is achievable, but I strongly believe in the first on-edge capabilities, artificial intelligence capabilities to deal with specific problems and acclimatize our sailors, our airmen. With that, I can start helping you make decisions, he says. Incidentally, Denis Mercier mentions Helsing, a German publisher that develops solutions based on artificial intelligence. These come to be integrated into the platforms of the armies to increase their operational capacities. A company that we will hear about, believes the former soldier.

AdvertisingWhen function takes precedence over rank

If it is necessary to tackle the aggregation of NATO data, for example, the task is immense. Not only do they come from 30 countries, but they are also complex and, by nature, highly confidential. Denis Mercier imagined a cloud bringing together all intelligence information and presented the idea to the ambassadors, who were unconvinced at first. After 18 months, the presentation of a first use case, very specific, revealed all the interest of such a project. But if it took 18 months for this presentation, it is because each country had to negotiate data transmission agreements with each other country. A long way. So I think we have to go step by step, hence the idea of ​​moving forward on unclassified data, continues Denis Mercier. You also have to know how to manage data of different classifications, but that, we know how to do it, if we want to, we can do it.

We have the impression that the situation calls into question the notion of hierarchy, suggests Engurand Renault, media and technology editor-in-chief of Le Figaro, moderator of the plenary session on Big Data & IA Paris 2022. It is often the opposite , replies Denis Mercier. His 39 years of experience in the air force allows him to recall the culture in force within this corps: function takes precedence over rank. In a project, it is the one who is the most important who, eventually, directs his leader. When the problem is dealt with, the hierarchy resumes, he explains. This notion of decentralization, the deputy CEO of Fives Group sees more, all things considered, in the military than in companies which are very hierarchical. If we want to improve operational efficiency, how can we organize ourselves so that we can give more responsibilities to an expert in the field, using the appropriate computer tools, explains Denis Mercier. We then have to think in a more general way and, here, it is not the field expert who intervenes, we take a ‘top down’ approach but if we use the latter from the start, we miss this digital transformation , warns the former chief of staff who calls for greater flexibility between the expert and the decisions that must be taken at a higher level.

A hub of data scientists supports business experts

And at Five Group, how did the digital transformation go? We found the same brakes as everywhere, we have to convince internally , recognizes Denis Mercier, adding that the bi-centenary company has great ambitions for this digital transformation. We have a hub and business experts. Like everywhere else, they don’t want data scientists to teach them their job. This is not the goal, the hub will only accompany them, insists the deputy general manager. From the start, there needs to be a close association between this hub of data scientists in Lyon and the business experts. And this organization must always be alive. And if we increase the expertise of business experts on data, then data scientists will be able to refocus on their own business. What is the return on investment? At first it is not important. You have to waste time to play together with the deals, it’s a partnership, insists Denis Mercier. It’s always a bet at the start, and this bet is never easy.

Finally, what measures have been put in place to measure the success of the data transformation? Within the group, we didn’t want to set up extremely strong KPIs,” explains the deputy managing director. I believe that it is the proliferation of use cases, once again purely functional or on business applications, that allow us to have an indication of our digital maturity. An indicator can be the number of field experts involved in the development of use cases.

Within the French group, the Fives CortX team – which brings together engineers specializing in software engineering, computer vision, data science and artificial intelligence – develops business applications for industrial connectivity, predictive maintenance, quality improvement, also covering the measurement of OEE (synthetic rate of return) on production.

Share this article

We would like to give thanks to the writer of this short article for this awesome material

“Creating intimacy between business experts and data scientists”

Explore our social media accounts along with other related pageshttps://www.ai-magazine.com/related-pages/