How I work
With tech and design
At AÉSIO, I work with two contractor developers, a data engineer and a data analyst. With no designer or QA on the team, I design the flows and wireframes and run acceptance testing. I set data-quality rules before development and write user stories with their acceptance criteria. I also prototype in code with Claude Code: the prototype is for learning, testing and aligning. Production code stays with the tech team.
Facing sponsors
I start with a visible win. At AÉSIO, the MVP first automated the department reporting: 2 days of preparation cut to half a day, before any new request. After that, each request is ranked by its value for decisions, and only the indicators that drive decisions stay. At Biogaran, we started with the pharmacies most ready for digital: the MVP proves adoption before scaling.
When data contradicts intuition
Data tells you what, the field tells you why. On the BlaBlaCar case, 100,000 trips showed that photo and ID mattered; interviews nuanced it: they don’t make passengers pick a driver, they break ties at equal times. Auto-accept looked like a good idea; the data showed no gain, so we dropped it. And when my own product disappoints, I say so: Prisme v1 changed shape every day, so I rebuilt the pipeline rather than the prompt.
The products that drive me
B2B, data or AI products where discovery keeps going during delivery, with a tech team to make trade-offs with, and results measured in adoption rather than in releases.