With the arrival of agentic AI, digital agents are now capable of understanding and acting completely autonomously.
This silent evolution is gradually taking hold in open spaces, meeting rooms but also in university lecture halls. But to get the best out of these invisible employees, you still need to acquire the necessary skills and learn how to manage them.
Increasing adoption
In French companies, previously totally improbable scenarios are unfolding. A service plans its deliveries using an AI agent. The latter offers scenarios, detects potential delays, and updates the schedule in real time. Teams thus avoid spending hours recalculating. And these examples are multiplying. According to Deloitte, more than one in two companies plan to deploy AI agents in their processes by 2027 (1). Three years ago, there were ten times fewer. Gartner estimates that by 2028, a third of software applications will integrate these autonomous agents (2). In some departments, teams already collaborate with their “digital colleagues” as with full team members.
Free up time with automation
The economic promises are impressive: agentic AI could create up to $4.4 trillion in annual value by 2030 (3). Behind this potential, there are concrete uses. At ServiceNow, the time needed to process certain complex files has been reduced by half (4). Salesforce is following the same trajectory: its employees, freed from repetitive tasks, have more time to focus on strategy, consulting and customer relations. At Klarna, AI agents now manage a workload equivalent to 700 employees (5). But the key is not there: the value is found in the reorganization of roles. Human teams keep control of sensitive decisions and relationships. Agents absorb repetitive tasks and massive data flows. The center of gravity is shifting: automation does not replace, it complements.
Collaborators, AI agents, etc.: the role of the manager is transforming
In several CAC 40 groups, managers are discovering that their role is changing. It is no longer just about coordinating human teams. We must now learn to work with digital agents capable of executing, but also of suggesting decisions. A recent Salesforce study reveals that 75% of HR departments anticipate a profound overhaul of managerial skills (6). In some companies, training is emerging to teach how to set objectives for an agent, interpret their recommendations and decide when to take control. A team manager confides: “It’s like onboarding a new employee, except they don’t need a coffee break. But you have to know where to set the limits.” A new profession is emerging before our eyes: that of conductor of hybrid intelligences, where humans and AI work together, each in their area of expertise.
Educate new generations
On campuses, the transformation is palpable. Today’s students are no longer just preparing to use tools. They must learn to work alongside digital agents. The American economist Tyler Cowen recommends that nearly a third of university courses be devoted to understanding AI, its use and its limits (7). Experiments are already underway. The University of Toronto, with its “All Day TA” program, deployed an AI assistant capable of answering more than 12,000 student questions in a single semester (8). These digital tutors do not replace teachers. They personalize the support, detect gaps and suggest adapted exercises. The teachers focus on the essential: teaching students to reason, to doubt, to create.
Learn to control and supervise these new tools
Training must now include a central skill: “AI literacy”. A term which covers the ability to understand the mechanisms hidden behind the responses of the AI agent, identify biases, audit decisions and know when and how to intervene (9). In some engineering schools, workshops are already devoted to the ethics of autonomous agents. We learn not only how to configure these systems, but also how to supervise them. In this emerging future, technology does not replace human thought. She widens it. It frees up time, offers new perspectives and redefines the scope of skills.
The signal is clear: the future will not be one of substitution, but of increase. This change imposes collective responsibility. Integrating these new employees requires specific skills as it is essential to know how to supervise them well to get the best out of them. For the first time in a long time, technology offers us a promise: that of refocusing humans on what distinguishes them and fascinates them, and giving them the means, thanks to technologies, to create more value.
——————-
1. Deloitte – Deployment of agentic AI in businesses (2025) https://www.deloitte.com/fr/fr/Industries/tmt/perspectives/ia-agentique-autonomous-genai-agents-transformera-la-productivite-des-entreprises.html
2. Gartner – Forecast for the integration of AI agents into software applications (2025)
https://www.outsystems.com/1/gartner-report-generative-ai/
3. McKinsey – Potential economic value of agentic AI (2025)
https://www.mckinsey.com/capabilities/quantumblack/our-insights/seizing-the-agentic-ai-advantage
4. Business Insider – Productivity and AI at ServiceNow and Salesforce (2025)
https://www.businessinsider.com/generative-ai-evolution-software-companies-develop-ai-agents-workforce-2025-3
5. Babl.ai – Customer Service Automation at Klarna (2025)
https://babl.ai/ai-agents-poised-to-transform-society-but-governance-lags-behind-report-warns
6. Salesforce Research – Evolution of managerial skills (2025)
https://www.salesforce.com/news/stories/agentic-ai-impact-on-workforce-research/
7. Business Insider – Tyler Cowen on college programs and AI (2025) https://www.businessinsider.com/economist-tyler-cowen-college-students-trained-jobs-ai-work-2025-8
8. Financial Times – “All Day TA” Program at the University of Toronto (2025)
https://www.ft.com/content/daa0f68d-774a-4e5e-902c-5d6e8bf687dc
9. Graydi.us – AI literacy and digital skills (2025)
https://www.graydi.us/blog/gray-insights/agentic-ai-a-new-partner-in-progress-for-higher-education-and-beyond
