Norman Deschauwer

The through-line

Across these posts, Norman Deschauwer keeps returning to the same discomfort: the gap between what organizations claim about themselves and what they can actually measure or admit. Whether the topic is AI adoption, KPI dashboards, or "agile" transformations, he repeatedly asks people to check whether the numbers or the culture behind a claim are real, or just paint. In a room asked "qui imagine que 30 % du code de sa société est écrit par une machine ?" he notes "très peu de mains. Soit on ne sait pas. Soit on a peur de dire oui" [1].

Over time the emphasis shifts from agility-as-community-building (the origin story of the Journée agile, run for 16 years [13][17]) toward AI and vibecoding as the new terrain where this same honesty problem plays out. The vocabulary changes from KPI and rituels to prompts and curseurs, but the underlying question stays constant: are you measuring what's actually happening, or are you comfortable not knowing?

On measuring what's real

Deschauwer treats measurement avoidance as a form of denial. On AI adoption inside companies he pushes teams to name the share of code a machine writes, noting the range "entre 30 et 90 %" he saw circulating among Anthropic and Salesforce, and insisting "la tendance, elle, ne bouge pas : le développeur devient manager d'agents qui écrivent du code" [1]. On KPIs and OKRs, drawing on a guest with "près de 30 ans de conseil," he warns against the "indicateur pastèque. Vert à l'extérieur, rouge à l'intérieur," concluding that "tant qu'un chiffre sert à sanctionner, on le maquille. Le jour où il sert à apprendre, on arrête de le cacher" [6]. He applies the identical lens to agility itself: it is not the method that fails but the cover-up, since agility gets "utilisée comme une rustine sur un pneu crevé ou utilisée pour cacher les merdes sous le tapis" [12], and "la culture d'une entreprise va manger sa stratégie au petit déjeuner" when people are afraid to speak [12].

On delegating to AI

He frames AI delegation as a skill to be dosed, not a switch to flip. After moving past "copier-coller dans ChatGPT," he settled on principles like "sheet in, sheet out," coding "par intention, pas par instruction," and deciding "niveau par niveau ce que je lâche à la machine," concluding "la délégation n'est pas un interrupteur. C'est un curseur" [7]. He is candid that this changes his professional identity: "mon boulot n'est plus d'être développeur. Et ça me va" [7]. At the same time he relays a guest's caution that offloading "easy cases" to AI is not a clean win: juniors "arrivent directement face à des problèmes de senior sans être passés par l'étape junior" and "s'appuient sur l'IA pour des sujets complexes, et ces sujets sont moins bien traités," while seniors "s'entraînent moins" and "traitent moins bien les cas complexes aussi" [2]. On new model releases he keeps asking whether claimed gains like "moins d'interventions" are "un vrai gain de productivité, ou juste un argument marketing" [5], and stresses that access changes like Fable 5's rollout come with "garde-fous automatiques" and an admittedly "imparfait" model-switching mechanism [14].

On starting before you're ready

A recurring move is to reject the idea that you need permission, a business model, or full confidence before you act. Recalling the first Journée agile drawing only 32 people, he quotes the advice that kept it alive: "Non. C'est exactement comme ça que ça commence" [13], and draws the lesson "pas besoin d'être prêt·e. On ne l'a jamais été - on a juste commencé" [13]. The same instinct shapes how he reads a young entrepreneur's story: "sors un produit qui marche, le modèle vient après" and "sois ton premier client" [3]. It also shapes his read on team dynamics, quoting a colleague's line, "le problème, c'est quand tout le monde attend, et personne ne démarre," and turning it into a self-directed checklist rather than a complaint about others [4].

On culture over method

He separates surface-level agile practice from the substance underneath it. Quoting one accompanying voice, he defines the real starting point as returning "à qui on est et qui on veut," adding "je travaille beaucoup sur l'intention. Le processus suit" and the commitment to "reconsolider ensemble, et puis on prendra le temps qu'il faut pour que ça marche" [11]. He links this to personal accountability more broadly, citing "la plus grande chose qui m'empêchait de me réaliser pleinement, c'était moi-même" and "la liberté vient avec le choix, et le choix vient avec l'espoir" [10]. For him, agility surviving or dying is not about rituals but about whether "les gens ont le droit de parler, ou peur de le faire" [12].

From the stage

In podcast appearances he describes agility itself with a definition not found in his written posts: it means "remaining connected while decoupled," a form of "voluntary and governance-based practice" [17]. He also frames his own podcast's purpose more explicitly there than in his posts, positioning it as looking at agility "through a business efficiency lens rather than agility for its own sake," aimed at entrepreneurs, managers, and agile coaches, and addressing "practical value and financial impact" for "business owners and decision-makers who are often overlooked in agility discussions" [16]. He notes his consulting company may itself