DailySurf began, loosely, in Bouznika, Morocco. I was there in June for a trip with Surf Progress, where I met Thibaut de Gueyer. Thibaut is a surf coach and the creator of Surf Progress. He has built a large online community around clear, practical teaching for beginner and intermediate surfers. He had organised the trip, and we got along quickly. We were both building things, and ideas came easily.
The actual project started a few months later, after I had returned to Paris. In November, I called Thibaut and asked: “What if we built a surfing app with AI in it?” His answer was immediate: let’s go.
By the following July, after many changes, we opened the beta.
The beta began around two simple jobs: help people explore places to surf, and keep a useful memory of their sessions.
AI helped us build and revise the app quickly. Things that would once have required more time and people became easier to try. But this speed can also be misleading. Building software is not the same as building a product people need.
The difficult part remained familiar. We had to talk to surfers, watch how they used the beta, and look at usage. We had to understand which parts were useful and, more importantly, what problem we were actually trying to solve.
The beta made this clear during a period of poor surf conditions. Usage dropped. The obvious explanation was that there was simply no good surf, but it also exposed a weakness in the product: the app was most useful when people already knew that a session might be possible and took the time to check.
That left much of the work with the surfer. They still had to inspect forecasts, compare spots, consider their level, check their calendar, and sometimes find out whether equipment rental was available. The app could present information, but it was not yet removing the search.
This led us to a new direction that we call Le Guetteur—the lookout. The idea is to search for suitable surfing windows on the user’s behalf, using the conditions alongside the surfer’s level, preferences, availability, and practical constraints such as travel time or rental. Instead of repeatedly checking the app, the surfer could be told when a realistic opportunity appears.
We are still testing this direction. We need to learn which signals matter, how personal the suggestions should be, and when an alert is helpful rather than noisy.
We started with a broad idea: a surfing app with AI. The beta gave us a more precise problem—helping someone notice the few sessions that genuinely fit their life and level. AI changed how quickly we could build. Finding that problem still came from talking to people and observing what happened.