Agentic development must not be reduced to a race for productivity, according to Jean-Charles Bordes, Group Chief Revenue & Offers Officer of the Smile Group. The time it frees up, based on AI agents (artificial intelligence programs capable of chaining tasks), must be reinvested in exchange, creativity, and innovation, provided that ESNs (digital service companies) and their clients make that choice together.
Companies have done the math: if AI agents write code, handle design, and produce documentation, why pay as much as before? ESNs and technology consulting firms hear this every week. According to the KPMG x Numeum report, October 2026, 75% of their leaders note that clients demand productivity gains thanks to AI. Two-thirds of them are revising their pricing models (fixed-price, subscription, asset monetization), and 37% see value-based billing as the most likely short-term outcome.
The risk, according to the author, is reducing a major technological revolution to a price war. “The real question is less about the time we save than what we do with it.”
Real gains, but not the only issue
Industrial history has long associated the factory with a loss of meaning in work. Agentic development, industrialized on a large scale, must change the game. It is an application development model that orchestrates specialized agents, framed by safeguards and steered by human experts. It leaves “the chain” to machines to fully restore “the workshop” to women and men, that place where ideas are invented, shared, and debated.
On projects undertaken, execution times are already cut by three. But speed alone guarantees neither quality nor value. Part of the freed-up time must be used to design the operational framework for the agents, and the rest must fuel what no algorithm can do.
Making room for exchanges and creative freedom
An agent invents nothing. Creativity remains the prerogative of teams, their knowledge of the field, and their interactions. What changes is the cost of experimentation: agents deliver a first version in record time, allowing three avenues to be tested instead of one, to fail fast, adjust, and start again, for design as well as technical architectures. The real opportunity is to stop deciding on hypotheses and build on concrete results.
The time returned can also be used to talk more. Still according to the same study, 62% of ESN leaders observe an increase in advisory needs upstream of projects. Yet how many steering committees spend time commenting on a progress rate, without discussing the value of what is being built? Spending half a day with business teams, understanding their daily work, explaining choices, and improving a project together can, according to the author, make the difference. “That’s often what distinguishes a delivered product from a truly adopted one.”
Rethinking roles and organization
At the ESN as well as at the client, business analysts, developers, testers, and project managers will produce less but will understand, communicate, and frame more: they will need to tell the agents what is expected, verify, and decide. The business analyst will translate the need into rules precise enough for the agents to execute, and the business expert will become an actor in every phase, provided they are available.
Chief Information Officers (CIOs) must also rethink the role and responsibilities of their teams. In their ten-year outlooks, behavioral skills take center stage: knowing how to communicate, adapt, hear what a client does not say. The tools will change, but these skills will remain.
For this transformation to be virtuous, choices are made together: it is up to ESNs to put agentic development at the service of humans, and up to CIOs and business units to give them the means. The ESNs that do this will be no less necessary tomorrow; they will be necessary differently.
