Olivier Rafal

Olivier Rafal, Consulting Director Strategy at «WeNvision», has been making the same argument since February 2024, and the argument has held. His CIO-Online tribune of 23 February 2024 claimed that generative AI belongs to product engineering rather than data science: building a foundation model from scratch takes months and millions of euros, so most organizations should hire integration engineers and cloud skills instead of data scientists, and treat the GenAI platform itself as a product funded by a monthly flow rather than a one-off investment.

Two pieces from June 2026 extend that frame instead of replacing it. On 1 June, «L'ingénierie logicielle à l'ère de l'IA : tout change... et rien ne change» sets out the paradox: no SDLC phase disappears under AI, each is intensified. Rafal reports that at Sfeir a business expert and tech lead pair now drives roughly 80% of the production chain, and prescribes three non-negotiable gates (specification, planning, delivery review). His formulation: "plus l'exécution est rapide, plus le cadre doit être strict". «vibe coding» is judged untenable in a company.

On 4 June, «Tokenomics foundation : l'ère du FinOps appliqué à l'IA est officiellement ouverte» moves the same reasoning to money: the token is the new unit of measure for technology spending, AI costs around 10 euros per hour per augmented workstation, and FinOps is about efficiency rather than cost reduction.

The constant across all three is the refusal to locate the problem in tooling. What Rafal keeps returning to is the operating model, which is harder to change than a vendor.