Artificial intelligence now occupies a central place in the thinking of professional service firms, but the advantage could shift from investment to the capacity to learn. This is the thesis put forward by Laurent Salanié, who argues that the advantage will go to firms capable of learning quickly, experimenting, and integrating AI into their practices. For over two years, these organizations have been wondering which tools to adopt and how much to invest.
This focus on technology is understandable, but it remains incomplete. Historically, major transformations in the sector have rewarded those who could invest, in sophisticated software as well as specialized teams. AI alters this balance: the best tools quickly become accessible to all, and access to them loses its strategic value. Everything will then depend on the ability to imagine relevant use cases.
A transformation that cannot be delegated
Many leaders still consider AI as just another technical topic, entrusted to innovation or business teams. This reflex quickly shows its limits, because AI is understood primarily by practicing it. It is no longer enough to select the right tools or sign the right contracts: one must concretely understand what they enable one to build.
Ethan Mollick, a professor at Wharton, argues that no one knows in advance what AI can bring to a given profession. According to him, the only way to discover it is through “disciplined experimentation.” A MIT study published in the summer of 2025 points in the same direction: it indicates that 95% of generative AI pilot projects in companies have no measurable impact on results, despite tens of billions invested. According to its authors, the problem lies less with the technology than with a learning deficit, both on the tool side and the organizational side.
For a firm, experimentation can focus on daily tasks. How is a technical note prepared? How is documentary research conducted? Each of these tasks holds opportunities that are only spotted by getting one’s hands dirty. It is now up to leaders to take part in the transformation rather than watching it happen.
The best use cases are built
Many still see AI as a ready-to-use tool, where one only needs to learn its features. The reality tells a different story. The first uses seem modest, like summarizing a document or drafting a letter. Then ideas flow: one test leads to another, several tools combine, an automation and then a specialized assistant eventually emerge to meet a specific need.
The process is less like installing software and more like a construction game, akin to Lego or Minecraft. You first assemble a few bricks, then, through trial and error, you end up building things you never imagined. Transformative uses rarely appear in the first few hours; they emerge after dozens of attempts. This is why leaders must “get their hands dirty,” not to become technology experts, but to develop the intuition that will allow them to spot tomorrow’s opportunities.
An opportunity for the most agile firms
This reality could have an unexpected consequence. If the advantage shifts from investment to learning, medium-sized firms possess often underestimated strengths. Structured enough to train their teams and disseminate best practices, they remain close enough to the ground to experiment quickly. In such a fluid environment, this adaptability could weigh more than size itself.
The issue extends far beyond the profession. Satya Nadella, CEO of Microsoft, summarizes it thus: “You can outsource work, not learning.” For the first time in a long while, the technological revolution demands more from firms than just investment. It demands that their leaders learn.
