{"id":3675,"date":"2025-12-15T11:49:26","date_gmt":"2025-12-15T10:49:26","guid":{"rendered":"https:\/\/2050.do\/?p=3675"},"modified":"2025-12-15T11:49:28","modified_gmt":"2025-12-15T10:49:28","slug":"ai-isnt-your-intern-anymore-its-your-unfair-advantage","status":"publish","type":"post","link":"https:\/\/2050.do\/fr\/ai-isnt-your-intern-anymore-its-your-unfair-advantage\/","title":{"rendered":"AI isn\u2019t your intern anymore \u2014 it\u2019s your unfair advantage"},"content":{"rendered":"\n

How 11 battle-tested founders deploy AI to build capabilities competitors can\u2019t match<\/strong><\/h2>\n\n\n\n

After publishing my first article<\/strong><\/a><\/em> on how to build credible business plans and investor relationships in a harder market, one question kept coming back from founders: \u201cAnd what about AI? How does it change the game for us?\u201d<\/p>\n\n\n\n

So I went back to the field.<\/p>\n\n\n\n

Over the past weeks, I sat down again with the same group of successful entrepreneurs  (****) \u2014 the same operators whose insights shaped the first chapter of this series \u2014 to pressure-test a related but sharper question: How do you deploy AI not just to save time, but to build capabilities your competitors simply cannot match?<\/strong><\/p>\n\n\n\n

I listened, I challenged, I compared scars. As in my first piece, what follows is a synthesis built horizontally \u2014 no theory, no top-down lessons, just collective intelligence<\/strong> from founders navigating the same storms. Clear convergences, useful tensions, and the sentences of hundreds of startuppers – which have participated in my podcast for the past 4 years –  that won\u2019t leave my head. (*)<\/p>\n\n\n\n

I began with a thesis: AI\u2019s value isn\u2019t automation \u2014 it\u2019s amplification of human judgment.<\/strong>  The founders confirmed it, but with an edge I didn\u2019t expect. \u201cIf you can\u2019t explain why your AI made a decision,\u201d Lo\u00efc Soubeyrand of Swile told me, \u201cyou don\u2019t actually control your business.\u201d<\/p>\n\n\n\n

Maxime Leroux at ClimateView added: \u201cWe run forty people with twenty to thirty AI agents \u2014 but every output is traceable, every decision has a human signature.\u201d<\/p>\n\n\n\n

What changed is this: AI has moved from experiment to infrastructure. The question isn\u2019t whether to adopt it, but how to do so without losing control of your product, your culture, or your ability to explain what happened when something breaks.<\/p>\n\n\n\n

\n

If you can\u2019t trace why your AI made a decision, you don\u2019t control your business.\u201d \u2014 Lo\u00efc Soubeyrand, Swile<\/em><\/p>\n\n\n\n

\u201cWe doubled our workforce with AI agents \u2014 but never lost the ability to explain our recommendations.\u201d \u2014 Maxime Leroux, ClimateView<\/em><\/p>\n<\/blockquote>\n\n\n\n

The 10x rule: start where AI unlocks, not where it optimizes<\/strong><\/h3>\n\n\n\n

The first pattern was sharp: don\u2019t chase 20% efficiency gains. Hunt for 10x capability unlocks.<\/strong><\/p>\n\n\n\n

\u00ab\u00a0If you’re 22 years old today, it’s the best opportunity to create a business,\u00a0\u00bb says Maxime Leroux, CEO of ClimateView<\/strong>, echoing Sam Altman’s provocative statement. \u00ab\u00a0You can start with agents, with artificial intelligence, as your first employees.\u00a0\u00bb (**) <\/p>\n\n\n\n

This isn’t hyperbole. For roughly $20 per month in AI subscriptions, a solo founder can now create professional websites, develop functional applications, analyze complex datasets, write compelling marketing copy, and execute campaigns that would have required a full team just three years ago. The barriers to launching have collapsed<\/p>\n\n\n\n

Hortense Harang, co-founder of We Trade Local (Fleurs d’Ici), put it plainly: \u201cAI doesn\u2019t just make things faster \u2014 it makes things possible that weren\u2019t possible before.\u201d<\/p>\n\n\n\n

At Fleurs d\u2019Ici, they analyse regional flower supply chains with a granularity that once required an army of consultants. \u201cWe can now offer hyper-local sourcing to florists who could never afford that depth of data. That\u2019s not a cost save \u2014 that\u2019s a new market.\u201d<\/p>\n\n\n\n

Rachel Delacour at Sweep frames the strategy: \u201cIf you use AI only to do existing work 20% cheaper, you\u2019re building on sand<\/strong>.\u201d<\/p>\n\n\n\n

Everyone can do that. \u201cThe moat comes from unlocking capabilities others can\u2019t match \u2014 because you moved first, collected better data, or built processes that compound.\u201d<\/p>\n\n\n\n

Mathieu Nebra at OpenClassrooms warns founders who confuse speed with defensibility: \u201cIt\u2019s never been easier to launch something with AI. But it\u2019s just as easy for someone else to disrupt you. The question is not \u2018Can I build it?\u2019 but \u2018Can I build something that lasts?\u2019\u201d<\/p>\n\n\n\n

Pascal Lorne, fresh from the GoJob exit, reframes the founder mindset: \u201cAI gives you superpowers and creates vertigo. The field of possibilities is immense. We succeeded by starting from real problems \u2014 not from shiny tech.\u201d<\/p>\n\n\n\n

Takeaway<\/strong>: map your processes not by cost but by potential impact if they were 10x faster, deeper, or wider. Prioritise AI use cases that create new capabilities, not marginal savings.<\/p>\n\n\n\n

The transparency tax you must pay<\/strong><\/h3>\n\n\n\n

The second consensus was unequivocal: you must be able to explain how your AI reached a decision.<\/p>\n\n\n\n

Soubeyrand at Swile was blunt: \u201cWe document every major prompt, version every model update, and require human review at decisions touching money or reputation.\u201d<\/p>\n\n\n\n

Not bureaucracy \u2014 risk management. \u201cWhen an AI recommendation backfires \u2014 and it will \u2014 you need to explain what happened, why, and how you fixed it.\u201d<\/p>\n\n\n\n

Leroux at ClimateView is even more structured: \u201cmaster prompts,\u201d no black-box code in production, human checkpoints at every critical step. \u201cWe double our capacity with AI agents, but we deploy nothing we can\u2019t explain to a city council or a climate scientist.\u201d<\/p>\n\n\n\n

Marta Sj\u00f6gren at Paebbl pushes this further with \u201cdecision journals\u201d \u2014 logs capturing not just what AI recommended, but why a human accepted or rejected it. \u201cWe review them quarterly to find systematic blind spots \u2014 both ours and the AI\u2019s.\u201d<\/p>\n\n\n\n

Axel Dauchez at Make.org ties this to culture: \u201cAI accelerates everything \u2014 including mistakes. Without built-in accountability, you lose the ability to course-correct when the model drifts or the market shifts.\u201d<\/p>\n\n\n\n

Lesson<\/strong>:  build transparency before scaling. Traceability isn\u2019t compliance theatre \u2014 it\u2019s operational survival.<\/p>\n\n\n\n

\n

\u201cThe era of geniuses is over. Now it\u2019s curiosity, resourcefulness, and judgment.\u201d \u2014 Pascal Lorne, GoJob<\/em><\/strong><\/p>\n\n\n\n

\u201cHire people who can have a productive conversation with a machine \u2014 and know when to override it.\u201d \u2014 Eric Carreel, Withings<\/em><\/strong><\/p>\n<\/blockquote>\n\n\n\n

Hire warriors, Not wizards<\/strong><\/h3>\n\n\n\n

A third thread: the skills that got people hired three years ago won\u2019t matter three years from now.<\/strong><\/p>\n\n\n\n

Nebra is direct: \u201cWe used to hire for what people knew. Now we hire for how fast they learn.\u201d<\/p>\n\n\n\n

AI-native adaptability beats credentials.<\/p>\n\n\n\n

Nicolas Reboud at Shine sees the shift daily: a senior engineer left; the new hire had half the r\u00e9sum\u00e9 but double the learning velocity. \u201cWe looked for people who can interrogate AI, not fear it or worship it.\u201d<\/p>\n\n\n\n

Eric Carreel at Withings insists on balanced skepticism: \u201cTreating AI as gospel is dangerous. Dismissing it is dangerous too. You need people who can challenge it, test it, override it, learn from it.\u201d<\/p>\n\n\n\n

Lorne crystallises the cultural need: \u201cCoders who once rode bicycles now travel at the speed of sound \u2014 but they still need to collaborate. Ideas come from humans, not models.\u201d<\/p>\n\n\n\n

Delacour offers the sharpest line: \u201cI don\u2019t need fragile divas. I want warriors \u2014 people with scars, who know how to fight and stay loyal.\u201d<\/p>\n\n\n\n

Implication:<\/strong> rewrite job specs. Hire for learning velocity, judgment, and the ability to collaborate with machines \u2014 not for static expertise.<\/p>\n\n\n\n

The junior crisis \u2014 And how to fix it<\/strong><\/h3>\n\n\n\n

Almost everyone raised the same concern: AI is erasing the traditional pipeline for junior talent (***)<\/strong><\/p>\n\n\n\n

Delacour captures it: \u201cIf AI writes basic code and drafts reports, what do interns do? We used repetitive work to build foundational skills. That work is gone.\u201d<\/p>\n\n\n\n

Reboud notes: \u201cWe have fewer entry-level roles. The old apprenticeship model doesn\u2019t fit the new economics.\u201d<\/p>\n\n\n\n

But several founders are already experimenting with solutions.<\/p>\n\n\n\n

Carreel at Withings redesigns junior roles around AI stewardship<\/strong> : juniors review AI-generated code, test it, refine prompts, and learn faster because they see more patterns.<\/p>\n\n\n\n

Nebra at OpenClassrooms formalised \u201cAI apprenticeships\u201d: juniors explain why they accepted or rejected AI suggestions. \u201cWe teach judgment through curation \u2014 not by protecting them from AI.\u201d<\/p>\n\n\n\n

Leroux rotates juniors across AI agents and projects with tight guardrails: rapid pattern exposure, safe learning loops.<\/p>\n\n\n\n

Consensus:<\/strong> don\u2019t abandon junior hiring \u2014 reinvent it. Teach judgment, not repetition. Pair juniors with AI under structured mentorship.<\/p>\n\n\n\n

Act now or watch the window close<\/strong><\/h3>\n\n\n\n

The final message was urgent: don\u2019t wait for the perfect AI strategy. The learning window is now.<\/strong><\/p>\n\n\n\n

Leroux is blunt: \u201cCompanies designing the perfect AI strategy will lose to those learning by doing.\u201d<\/p>\n\n\n\n

Nebra: \u201cPick one problem AI can solve, test it, measure it, learn.\u201d<\/p>\n\n\n\n

Small, fast loops beat grand plans.<\/p>\n\n\n\n

Delacour reframes adoption as culture: \u201cAI isn\u2019t a tech challenge \u2014 it\u2019s a change-management challenge.\u201d<\/p>\n\n\n\n

Carreel adds a market warning: \u201cTeams deploying AI systematically today will have 12\u201318 months\u2019 advantage. That\u2019s the difference between leading and following.\u201d<\/p>\n\n\n\n

Reboud is tactical: \u201cDon\u2019t hire consultants. Empower three people to experiment for 30 days. You\u2019ll learn more than from any deck.\u201d<\/p>\n\n\n\n

What to do next, concretely <\/strong><\/h3>\n\n\n\n