Companies Are the Next Grandes Écoles

Roxan Roumégas explains why.

There is a curious dissonance at the heart of the modern labor market. On one side, companies that struggle to find employees mastering their tools, their methods, their culture. On the other, a training system that produces graduates competent in the abstract, but often helpless before the concrete realities of a position. Between these two worlds, a gap that neither the grandes écoles nor universities seem in a hurry to fill.

It is into this interstice that Roxan Roumégas, co-founder of the Mentivis consultancy, has thrust himself. His diagnosis is simple, almost brutal: if school does not train for the real needs of companies, let companies become schools.

The End of Pedagogical Delegation

For decades, companies have externalized their training. They sent employees on courses, financed certificates, hoped that someone, somewhere, had thought to teach the right software, the right method, the right reflex. This model, convenient but passive, reaches its limits in a context of technological acceleration.

Giants like Airbus or EDF understood before the others: creating an internal training center means taking back control. Training on in-house tools, transmitting a company culture, retaining talent by offering them a learning trajectory, and incidentally delivering professional certifications recognized by the State, fundable via the CPF or work-study programs. The model crossed the Atlantic long ago: Google, Amazon and their peers have made internal training a pillar of their attractiveness.

It is not simply a question of cost, although the economic argument is real. It is a question of pedagogical sovereignty. The company that trains itself knows exactly what it transmits, to whom, and why.

The Diploma: Honorable Symbol but Perishable

To speak of training without mentioning the diploma would be a guilty omission. In France more than elsewhere, the academic title remains a powerful social sesame. But its value, once quasi-permanent, is undergoing erosion at an unprecedented speed.

An engineer graduated in artificial intelligence in 2020 must already retrain. A developer trained on 2018 technologies may be working on tools that have become obsolete. In technical trades, the question is no longer "what diploma do you have?" but "what have you learned in the last eighteen months?"

Professional certifications respond better to this reality. Less prestigious than the master's, they are more agile: they define precise skills, update easily, and correspond to concrete salary grids. They will not replace the diploma in the French collective imagination anytime soon. But they are already competing with it in fact.

The challenge, Roumégas emphasizes, is no longer to train someone once and for all, but to teach them to learn continuously. A pedagogical ambition far more demanding than it appears.

AI: Lever or Fracture?

Artificial intelligence obviously runs through this conversation as it runs through all others. Its impact on training is double, and both sides deserve to be looked at face on.

As a pedagogical tool, AI offers genuinely promising possibilities: personalization of pathways, adaptation to each learner's pace, immediate feedback, simulation of professional situations. The uniform classroom — same content for thirty learners with radically different profiles — has never been particularly effective. AI could finally make it possible to do otherwise.

But AI is also a factor of polarization. Professionals who master it gain in productivity; those who ignore it risk finding themselves left behind. Training in AI is therefore no longer optional. It is a question of equity as much as competitiveness.

One prerequisite remains, often neglected: mastering the fundamentals before delegating to the machine. A tool as powerful in untrained hands can produce mediocre results with excessive confidence, a particularly dangerous combination.

Teaching Whom, How?

Behind the great structural questions lies a more modest but formidable challenge: not everyone learns the same way. Some retain by listening, others by practice, still others by reading or silent reflection. Inductive pedagogies, project-based learning, resolution of real problems are all approaches whose effectiveness is known, and which we continue to apply all too rarely.

In companies, training is still often entrusted to human resources by default, for lack of a dedicated pedagogical role. It is a bit like entrusting a restaurant's kitchen to the cleaning team: good will does not replace specific competence.

A Promise, Under Conditions

If companies are indeed the next grandes écoles, they will have to accept the constraints. Training seriously costs time, money, and requires a pedagogical reflection that most organizations have never really developed. The risk is to reproduce, internally, the defects of the system one seeks to circumvent: standardized content, superficial evaluation, learning endured rather than chosen.

What is lacking is not so much technology as intention: that of forming individuals capable of adapting, questioning, continuing to learn long after the trainer has left the room. An ambition that, itself, has no expiration date.

*A big thank you to the Learning Club and Clément Mesclin for making this conversation possible.

Companies Are the Next Grandes Écoles

A Silent but Structural Transformation

The education system is entering a phase of deep recomposition.

Companies are no longer content to recruit profiles trained elsewhere: they are building their own training systems.

The Shift in the Center of Gravity

The production of skills no longer depends solely on educational institutions.

Companies:

• internalize training
• structure their own pathways
• align learning and operational performance

👉 Training becomes a strategic asset.

From Diploma to Operational Skill

The diploma is losing its function as the sole signal.

It is gradually being replaced by:

• internal certifications
• validation by usage
• performance in real situations

👉 What counts is no longer what you know, but what you can do.

The Company as an Educational System

Organizations are developing complete devices:

• internal academies
• structured career pathways
• continuing training platforms

👉 The company becomes an environment of permanent learning.

A Response to a Structural Crisis

This evolution responds to deep tensions:

• technological acceleration (AI, data, automation)
• rapid obsolescence of skills
• gap between initial training and real needs

👉 Training becomes a condition of organizational survival.

Toward a Distributed Educational Model

The education system recomposes around three poles:

• schools (fundamentals)
• companies (operational)
• intermediary actors (architecture and engineering)

👉 Education becomes an interconnected, non-centralized system.

Conclusion

Companies do not entirely replace schools.

They redefine what it means to "train."

Training is no longer a prerequisite for employment.

It becomes a function integrated into economic production.

To speak of training without mentioning the diploma would be a guilty omission. In France more than elsewhere, the academic title remains a powerful social sesame. But its value, once quasi-permanent, is undergoing erosion at an unprecedented speed.

An engineer graduated in artificial intelligence in 2020 must already retrain. A developer trained on 2018 technologies may be working on tools that have become obsolete. In technical trades, the question is no longer "what diploma do you have?" but "what have you learned in the last eighteen months?"

Professional certifications respond better to this reality. Less prestigious than the master's, they are more agile: they define precise skills, update easily, and correspond to concrete salary grids. They will not replace the diploma in the French collective imagination anytime soon. But they are already competing with it in fact.

The challenge, Roumégas emphasizes, is no longer to train someone once and for all, but to teach them to learn continuously. A pedagogical ambition far more demanding than it appears.