Introduction

1The actual and expected repercussions of AI on organizations, jobs, and skills is particularly significant. More than 85% of companies expect AI to profoundly transform their processes by 2030,¹ and nearly 30% of work hours could be automated within the next five years.² A direct impact is that at least 60% of employees will need to be trained in AI by 2030.¹

2With Human Resource (HR) Directors, employees, and social partners raising this question, Veolia has to better understand the extent of the anticipated impact of AI and has launched an ambitious training strategy.

The AI Impact on HR and Skills

3A forward-looking study has been conducted, combining external quantitative and qualitative analyses with insights from internal experts. The aim was to develop, if not with absolute certainty, a clear position from Veolia on the impact of AI on jobs and skills, to provide shared understanding and guidance to support employees through this transformation. Three key themes have been identified through the study.

4First, AI impacts not only on jobs, but also on the way people work, interact, and evolve within the company.

5Second, support functions are evolving rapidly, both in the market more broadly and at Veolia. In these areas, external studies show a high potential for automation in the short to medium term.

6The third issue is how AI is reshaping the industrial world with new opportunities for collaboration.

AI and the Way People Work

7First, in every job, AI is transforming certain core tasks: some are augmented, enriched by AI, others are automated, replaced by AI, while new tasks may emerge or remain unchanged. The sum of these impacts on all tasks helps characterize a job’s exposure to AI and, consequently, the level of support required.

8Beyond the impact on individual tasks, the real challenge for companies today is to generate collective performance with AI, by integrating a range of changes into end-to-end processes. This trend, accelerated by agentic AI, will profoundly change everyone’s roles and, ultimately, how work is organized. We need to prepare for this transformation now, moving beyond isolated use-case experiments and rethinking processes.

9AI also fundamentally disrupts the traditional relationship with knowledge and expertise. In particular, the transfer of knowledge from senior to junior employees is being challenged. Today, junior employees naturally turn to AI for answers to their questions ‒ before seeking out a more experienced human colleague. At the same time, seniors increasingly delegate basic tasks to AI, thus depriving juniors of “on-the-job” learning opportunities that have traditionally been their main training pathway. These new practices can exacerbate generational divides and call into question how junior training is designed. Beyond the generational issue ‒ which recurs with every major technological shift ‒ all expertise-sharing interactions are being challenged by AI. If AI provides us with immediate answers, why take the time to ask an expert colleague from another department ‒ or even within our own team? The entire dynamic of workplace collaboration (and society at large) is evolving.

10Finally, as a logical consequence, AI is reconfiguring how we manage our teams daily, presenting managers with new challenges: overseeing AI use to ensure compliance and quality control, preventing AI-related risks (notably, around mental health), and maintaining a collaborative spirit. To put it simply, we could say that intelligence is not artificial but collective above all else.

The AI Impact on Support Functions

11In these functions, the potential for automation is very high in the short to medium term: over 65% of tasks in IT, finance, and customer service roles are expected to change significantly.³ At Veolia, we began experimenting early with use cases through the rapid roll-out of Veolia Secure GPT, our internal GPT, launched in 2023. Today, the ability to automate these use cases by creating agents ‒ still on this secure platform ‒ enables us to begin industrializing these applications. In parallel, the most common AI applications on the market are being integrated into Software as a service (SaaS) solutions already widely deployed for support functions. This will allow for rapid adoption. The main challenge in these roles is, therefore, to quickly train employees in good AI practices, to strengthen data culture, and to raise awareness of AI’s limitations. These fundamentals are essential for preparing for the transition² that will shape the future of work. Indeed, unlike generative AI, agentic AI can perform actions and act autonomously.

Focus: Measuring the Impact of AI on Jobs

12As part of Veolia’s forward-looking studies, three parameters were used to measure AI’s impact on various jobs:

1. The level of AI impact on job families (“AI exposure”) in terms of the potential for task augmentation and automation, based on the World Economic Forum’s work.

2. The potential value creation from these transformations, in terms of performance, growth, and human aspects.

3. The expected pace of transformation: fast (within the year), medium (by 2027), or slow (by 2030).

The way AI is Reshaping the Industrial World

13AI is reshaping the industrial world. This is already visible in industries that have led previous waves of digitalization and deployed large-scale data and industrial Internet of Things (IoT) infrastructures, such as automotive, aerospace, or pharmaceuticals sectors. The addition of conversational and agentic AI builds directly on these digital foundations. Veolia, like most companies in the energy and utilities sector, has already begun this operational transformation and deployed initial AI use cases in operations. These cases cover both new services and value creation opportunities for clients, as well as benefits for “augmented operators” ‒ notably regarding safety risk prevention, improved working conditions, and new career opportunities made possible by AI assistance for tasks like writing or translation.

14While the potential for AI in operations and Veolia’s ambition to be a leader in this area are clear, the path is longer than for support functions, as these are ongoing transformations that require significant time, financial, and human resources: setting up digital and data infrastructures, capturing and ensuring data quality, evolving business and contractual models, etc. Human support must continue throughout these changes, to best prepare operators for new ways of working. This means supporting the adoption of digital professional tools for some, strengthening basic data and cybersecurity skills, raising awareness of conversational AI, and gradually adopting concrete use cases.

AI Training Strategy: Genesis and Roll Out

15For a group like Veolia, with over 215,000 employees worldwide, training staff in AI is, of course, a major issue at the heart of our strategic GreenUp 2024-2027 program, which aims to make AI a driver for accelerating the ecological transformation. By 2027, the Group’s goal is to raise awareness of generative AI among 100% of employees. This includes not only “connected employees” (those with digital access) but also blue-collar workers currently more distant from digital tools, whose jobs will also be impacted by generative AI.

Genesis

16The trigger was the rapid rise of generative AI in the public sphere, especially with the widespread adoption of ChatGPT. Given the real risks posed by uncontrolled sharing of sensitive data on platforms like ChatGPT, the primary objective was to mitigate risks and manage practices by quickly establishing a common framework for all employees, clarifying and regulating AI usage in a professional context. This led Veolia to launch its internal conversational agent, Veolia Secure GPT in September 2023, under the leadership of Veolia CEO Estelle Brachlianoff.

17The main challenge was to raise awareness of both the risks and the potential of generative AI for everyone, in a context where its uses, technological and regulatory developments, and Veolia’s strategy were not yet clearly defined.

18The challenge was to spark interest and curiosity while establishing a framework despite an unclear and constantly changing context.

19To involve employees early in this large-scale transformation, an agile format was chosen: a face-to-face awareness workshop that could be adapted to different geographical contexts and business lines and easily updated.

20This deliberately “simple” choice ‒ counter to the more common e-learning formats for large-scale dissemination ‒ ensured agility and responsiveness in this context.

Engaging and Involving Employees in this Transformation

21A network of “AI champions” has been set up in each Business Unit (BUs) on a voluntary basis. These ambassadors were trained as a priority, tested new tools, shared best practices, and drove the dynamic within their teams.

22This is a central part of Veolia’s approach: The champions play a key role in the AI awareness and training strategy. They were the subject of a dedicated review to further refine their role and governance.

23A first pilot was conducted in a few business units in France and at Group headquarters. Training was then rolled out to all BU champions, remotely and in three languages (French, English, and Spanish). Finally, starting in June 2024, the awareness modules were deployed in BUs by the champions.

24In the end, 2024 saw significant achievements, with over 400 champions involved and 13,000 employees reached by this initial AI awareness workshop.

25Several changes have been implemented to address the major challenges of this type of approach:

• Continuous updating: Within six months, 30% of the content was already obsolete. Continuous updating has been implemented, with a wave of more in-depth training in early 2025, followed by another wave in late 2025.

• Combining overall coherence with local relevance: Training must be adapted to local issues and speak to different jobs, beyond simply being available in each language. Governance has been strengthened in this regard by leveraging the internal AI for GreenUp program, coordinating IT, HR, business, and innovation. The role of champions has been formalized and structured with appointments by management, while clarifying the respective roles of the central team, local players, HR, and AI Strategy Leaders. This approach has enabled the community to grow from an initial 400 champions in 2024 to more than 1,500 champions in 2025.

• Promoting a responsible approach to AI, in line with Veolia’s “raison d’être”: In the Champions Program, this topic was the subject of a dedicated module, designed to raise awareness of the ethical and transparent use of AI technologies in the service of human progress while preserving resources. The challenge is to preserve the intelligence of the action, integrate environmental aspects, and adopt a use focused on the benefits of ecological transformation. The European AI Act sets out a framework for action to build on, encouraging training for anyone who uses AI in their work, including on these aspects. Veolia is also a signatory to the AI Pledge and a member of the Coalition for Sustainable AI.

Next Steps and Training Priorities

26The focus now is on accelerating awareness, with clear targets: training 500 senior leaders and raising awareness among the employees, with the ambition of reaching 100% of the workforce by the end of 2027. An e-learning program is available in 12 languages, which, after the first wave of workshops, will enable a structured and consistent message to be widely disseminated.

27Beyond raising awareness, the adoption of concrete use cases must be structured to increase individual and collective efficiency. By the end of 2025, more than 800 business use cases had been identified, and some are beginning to be rolled out.

28Training is positioned to support the various business lines to help roll out and adopt these use cases. For example, as part of prompt engineering training launched by the Veolia Academy, a library of about a hundred exercises, organized by job family, was created and made available to champions, so that each could lead training in their area using tailored examples. Among the champions, coordinators for each business line have also been appointed at the Group level to promote the sharing of AI usage within their area.

Conclusion

29Far from being just a simple awareness project, AI for GreenUp is a strategic business program. Our approach to AI training is built on a systemic vision that combines safe usage, the adoption of practical use cases to improve individual and collective performance, and support for job transformation at all levels of the Group. To meet the Group’s ambitions, our training and awareness strategy is based on several key principles: a champions network, continuous iteration between experimentation and deployment, global and local coordination encouraging “copy & adapt” within and between professional communities, and specific support for the use cases deployed to ensure that all those involved locally receive training.

30Ultimately, our aim is to progressively support every region and every job as their roles and skills are profoundly redefined. This is essential to putting AI at the service of ecological transformation and building a more effective, sustainable, and human-centered industrial model.