"We have an idea." When Aurélie van Peteghem and Sophie, from daphni, called Marc Simoncini with that sentence, he admits it himself: "generally, when someone opens like that, you should be wary." The idea in question: lock the founder of Meetic, Sensee and Angell in a room with a 24-year-old creator, Nicolas Mazurier - a.k.a. Aywen, a YouTuber specialising in AI, and ask them to build an AI-native company in eight days. Cameras rolling. Sceptical at first, Simoncini said yes.
Eight days later, by his own account, "we really had created a company". The experiment, told on LinkedIn on 19 June by Marc Simoncini and Aurélie van Peteghem, and filmed on Aywen's channel, is not an isolated PR stunt: it is a full-scale demonstration of what has changed in the way you launch a product.
"When I launched my first companies, it took months to get to what we did in a few days. From Minitel to Claude, wtf!" - Marc Simoncini
Forty years apart, the same energy
The setup dreamt up by daphni is almost too good to be true: "40 years separate them. 40 years of innovation. 40 years of technology. 40 years of ways of working", writes Aurélie van Peteghem. On one side, an entrepreneur who built some of the finest success stories in French tech. On the other, a creator who already builds with AI every day, for whom a coding agent is not a novelty to tame but an obvious working tool.
What the team thought it was filming - "an experiment around AI" - turned into something else: the meeting of two generations of entrepreneurs "driven by the same curiosity, the same energy and the same urge to build". That, in our view, is the real lesson. AI does not replace Simoncini's entrepreneurial instinct; it accelerates it. And the fluency of an AI native like Aywen only pays off because there is, facing him, a product vision and market experience.
Why eight days, and not eight months
What used to take months - prototyping, coding a first product, iterating, going live - now fits into one intense week. The tools have crossed a threshold. A founder can now:
- describe a feature and see it built straight away, without going through a hiring cycle;
- test an idea with real users before even having a technical team;
- iterate several times a day where you previously waited for the next sprint.
That is excellent news. The barrier to entry for building products has never been lower, and the daphni experiment makes it tangible for a wide audience. At daphni as elsewhere, Simoncini notes, the investor's job itself is "in full transformation": you learn, you test, you question yourself.
The flip side of speed: when you build that fast, two people, in eight days, you keep the product map in your head. But that mental map survives neither scaling up, nor the arrival of a team, nor "next month's you" who will have forgotten why a given decision was made.
Speed creates a new need: keeping the map
This is the paradox we see in everyone who adopts agentic development. The faster you write code with AI, the more features, branches and versions you produce - and the harder the question "what is actually done, in which version, on which branch?" becomes. Over eight days locked in a room, two motivated people hold that thread without tooling. On day nine, when it is time to convert and grow the company, the thread is lost.
That is precisely the conviction behind Survol: in an era where AI writes the code, the scarce value is no longer the ability to produce, it is the ability to steer. Keeping the functional view, knowing what is in test and what is in production, deciding as a team before launching the agent. The approved spec becomes the instruction - and the product's memory no longer evaporates when the filming stops.
The takeaway
- The rules of entrepreneurship really have changed: eight days are enough for a genuine start.
- AI amplifies experience and instinct, it does not replace them - the Simoncini × Aywen encounter proves it better than any speech.
- Moving fast is the new easy; staying in control of what you build is the new hard.
We are looking forward to seeing what becomes of this eight-day-old company. And, incidentally, to seeing how many founders try the exercise after watching the video. If they keep a readable record of what they build, day nine will be far simpler than day one.
