For three years, businesses believed that the competitive edge would come from choosing the right AI tool. They were wrong. Now that the models are all on par, the only real difference lies in what humans can do with them.
Yet largely unnoticed, France Compétences had listed Product Builder among its six rapidly evolving emerging occupations as of December 2023… a prescient move?
The convergence of models has killed the competitive advantage of the tool
Companies that anchored their strategy on acquiring the “best” model missed the real question. The performance of AI models is now converging in dramatic fashion.
According to Stanford University’s AI Index 2025 report, the performance gap between the market’s best model and the tenth is down to 5.4%, from 11.9% the previous year. Even more striking, the gap between the two world’s best-performing models has shrunk to 0.7%. Normalization is touching all segments: open-source models have caught up with proprietary models, with the gap on reference benchmarks down to 1.7%.
ChatGPT, Claude, or Mistral now produce highly comparable results for 90% of common professional uses. Drafting a brief, analyzing a contract, summarizing a meeting: no tool stands out in a fundamental way.
Knowing how to use AI is no longer enough: you have to know how to build with it
Since the tool no longer makes the difference, the real lever of competitiveness lies in the ability to integrate AI into business processes through an emergent profession. The consensus on the need to “train in AI” masks a fundamental disagreement about what that entails.
Between the average ChatGPT user and the traditional data scientist, there exists a third profile, still largely underserved by the training system: the Product Builder.
This professional is capable of stitching together no-code tools and AI APIs to create functional applications, structure databases, and automate complex end-to-end processes, without writing a single line of traditional code.
The signal is unambiguous: according to Gartner, three out of four enterprise applications will be built without code by 2026 — this profile is not a niche, it is the direction the entire sector is heading.
A profile largely underserved by the training system
Companies today find themselves in a squeeze. On one side, traditional technology training is too long and too specialized to address the urgency: the country lacks data scientists and AI engineers, and these streams do not produce results in a matter of months. On the other hand, AI training offered to employees generally consists of short awareness sessions on prompt usage.
The MEDEF, in its 2025 proposals report, directly warns about this skills shortage: the employers’ federation estimates that 300,000 French people need to be trained in AI each year to maintain the competitiveness of French enterprises. Yet AI-focused programs currently graduate only about 17,000 students per year.
Professional retraining: the only credible lever to train Product Builders at scale
Becoming a Product Builder cannot be conjured up in a weekend. It is a full-fledged profession that requires product thinking, business vision, and technical mastery of data flows and automation. Serious training for this profile demands three to eight months of intensive practice. Neither short courses nor e-learning modules deliver this level of mastery.
That is precisely why professional retraining is a mass transformation lever. The new “retraining period,” effective February 2026, significantly eases the steps for employees wishing to change careers while keeping their jobs during training. The program is now open to all employees, with no age or qualification criteria. Operators like Transitions Pro are already tackling the issue with dedicated projects such as “AI Skills,” showing that financing mechanisms — CPF, Opco, regional subsidies — are ready to support this transition.
The aim is not so much to “train everyone in AI” as to design retraining pathways toward occupations that industrialize AI usage. The convergence of models has not ended the competition. It has shifted it: from technology to skills, from tools to the professionals capable of orchestrating them. Companies that grasp this first will gain a structural advantage over those that continue to chase differentiators in the licenses they subscribe to.
*Jonathan Pinet is the CEO of the European School of Digital
