The same story is playing out in many organizations. Management purchases a handful of ChatGPT, Gemini, or Copilot licenses, announces during a company meeting that “AI is now a strategic priority,” and then hopes adoption will naturally spread across the workforce.

While a few curious employees seize the opportunity—experimenting with prompts and improving productivity on certain tasks—most teams remain at a distance. Some are skeptical, others are concerned, and many simply do not understand what is expected of them.

The result is often the same: six months later, the tools are available, but the anticipated transformation has not materialized.

Is this the employees’ fault? Tifany Clemenceau, who co-founded AI consultancy Eria with Melvin Duveau in 2025, does not think so. In her view, the issue is first and foremost one of internal AI culture and adoption.

Why Provide AI Tools to Employees?

A natural question arises: why invest in AI tools for employees when many of them are already available for free?

Because even companies that believe they can “wait before getting started” are often already dealing with AI usage behind the scenes.

Within the industry, this phenomenon is known as “shadow AI.” The term refers to employees who are already using ChatGPT or similar tools with company data, without official approval or governance frameworks—creating significant risks, particularly around data confidentiality and security.

The phenomenon is far from marginal. According to research by Boston Consulting Group, more than half of employees resort to alternative tools when their employer does not provide a suitable solution.

Providing approved AI tools is therefore essential. But as many organizations have learned, that alone is not enough.

The Trap of “We Gave You the Tool—Now Figure It Out”

“The real challenge today is changing how an entire organization works,” says Tifany Clemenceau.

The goal is to create an environment where employees understand why they are using AI, where it can help them in practical terms, and—most importantly—how to integrate it naturally into their daily workflows.

“A leadership team that simply says, ‘You all have Gemini now—go use AI,’ without training or dedicated time to experiment, is setting itself up for failure,” she explains.

There is a significant gap between discovering a new tool with curiosity and fundamentally changing the way one works.

Teams are rarely guided on the questions that matter most:

  • What information can be shared with AI tools?
  • Which tools should be used?
  • What outcomes is the company actually trying to achieve?
  • How can mistakes be avoided?
  • Where does automation make sense—and where does it not?

Without a clear framework, employees are left to navigate these questions alone, often leading to confusion and frustration.

Use Cases Don’t Come From the Executive Committee

In its white paper The AI Era: A Guide to Enterprise AI Deployment, published on June 3, Eria draws on feedback from ten French scaleups that have achieved AI adoption rates above 80%, including Qonto, Doctolib, PayFit, and Malt.

One key lesson stands out: leadership can set the vision, fund the tools, and establish priorities—but the most valuable AI use cases almost always emerge from the operational teams themselves.

“Executives don't know exactly what happens within every function on a daily basis. The real needs come from the teams.”

This is often where AI projects stall. When organizations select tools before identifying concrete problems to solve, AI quickly becomes another top-down initiative. Employees struggle to see the practical value of tools they are being asked to adopt.

By contrast, the companies moving fastest are those that encourage employees to experiment, identify repetitive tasks that consume time, and then share their discoveries with colleagues.

The mechanisms can be remarkably simple:

  • Weekly AI coffee sessions
  • Internal demonstrations
  • Dedicated time during team meetings to share lessons learned
  • Workshops and hands-on experimentation

The objective is not to turn every employee into a prompt engineering expert. Instead, it is about creating what Clemenceau calls “fertile ground for discussion,” where AI becomes progressively normal, visible, and collaborative.

She also cautions against treating AI training as a one-off exercise:

“Running a single training day is a good start. Believing that the company is transformed afterward is simply not true.”

AI tools and use cases evolve constantly. A cultural transformation of this scale cannot be achieved through a one-time initiative—it requires sustained commitment over the long term.

Managing Employee Concerns About AI

The Eria report also emphasizes the importance of acknowledging employee uncertainty around AI.

Behind seemingly technical questions—or what may appear as a lack of enthusiasm—often lie deeper and entirely legitimate concerns: fear of losing relevance, concerns about job displacement, or simply anxiety about rapid change.

“When employees don't understand why AI is being introduced, that's when we see the strongest resistance,” says Clemenceau.

For that reason, organizations should address the “why” before anything else.

Why is the company investing in AI?

What objectives is it pursuing?

How will this transformation concretely benefit employees and customers?

Until these questions are answered clearly, AI tools tend to be perceived as yet another corporate mandate.

The most advanced organizations invest time in communicating their vision transparently—not by promising an abstract revolution, but by demonstrating tangible improvements to everyday work.

A Layered Approach to AI Adoption

In practical terms, Eria recommends a gradual, structured rollout.

The first step is establishing the foundations: a clear AI vision, governance processes, and robust data security requirements.

From there, organizations should focus on creating an initial collective momentum before gradually expanding AI usage over time.

One effective strategy involves identifying internal “AI champions”—employees who already believe in the technology and are willing to help colleagues adopt it.

At the same time, organizations should develop a group of more technical profiles, referred to as “builders,” responsible for creating the AI assistants and agents used by the wider workforce.

Clemenceau offers a reassuring perspective:

“We will never have 100% of employees building AI agents, and that's perfectly fine. What matters is that everyone understands how AI agents work so they can incorporate them into their daily routines.”

AI Strategy Is Becoming an Employer Branding Issue

For organizations still hesitating to embrace AI internally, Clemenceau offers a warning: employees increasingly expect it.

“They want their employers to help them develop AI skills because AI is becoming a genuine employability factor,” she explains. “We regularly hear employees say, ‘My company isn't training me, and I'm worried about falling behind.’”

The issue therefore extends far beyond productivity gains.

It affects employer branding, talent attraction, and employee retention.

As AI becomes more widespread, workers are increasingly evaluating employers based on their ability to prepare them for new ways of working.

Companies that ignore this trend face a double risk: losing efficiency today while becoming less attractive to talent tomorrow.