Suggestions
Samuel Fillon
CEO | Sommelier du Parfum
Samuel Fillon is a professional who specializes in providing unfair advantages to composition houses and fragrance brands, such as Estée Lauder and Puig, through the strategic utilization of data and machine learning. He is known for empowering consumers with AI-powered fragrance advisory services that are deemed too good to be true. Samuel can be reached at samuel[at]sommelierduparfum[dot]com, and he prefers to meet over morning coffee. He pursued his education in Cursus Ingénieur, Economics, Management, and Finance at École nationale des ponts et chaussées, as well as School of Engineering at The University of Tokyo. Previously, he held leadership roles such as Chief Executive Officer at Omniscent.ai, Founder and CEO at Sommelier du Parfum, and Analyst at Avolta Partners.
Highlights
"😲 Wait, that biased?"
When I first started working in fragrance, years ago, I was shocked to see how consumer testing was well… everywhere.
On paper, it sounds great – I mean, the market is always right, so asking the market what it thinks is the thing to do.
Isn’t it?
So far so good. But let’s talk about the method instead.
When I explained a fellow engineer friend how these tests were conducted, he started laughing as we counted all the biases: 📍 Location bias: Recruitments are made at 1 or 2 locations (like Galeries Lafayette in France) 🚶 Idleness bias: Recruitments happen typically in the middle of the week (your target respondents don’t have a job?) 💰 Incentive bias: The incentive is minimal to keep costs low, so only people who value an hour of their time ~ 10 to 15€ are recruited
To make things even worse, people conducting those interviews are low-compensated, low-trained and incentivized by the number of interviews conducted (bye bye quality).
Everyone in the industry knows those methods are far from great but « hey, there are no other options out there »
With OmniPulse, we want to set the bar for fragrance testing to great and beyond: ✅ By removing the recruiting biases as we select and screen people online ✅ By ensuring that interviews are conducted in the best conditions possible ✅ By engaging with respondents in the comfort of their home, in a fun and interactive manner, far away from boring scripted interviews
After conducting multiple tests, we have refined our methods until reaching unprecedented predictive accuracy and now create video interviews with AI-assisted emotional recognition – and that’s just the beginning 🚀
📩 DM me for more info – Our team of fragrance experts and AI engineers and PhDs are happy to discuss your testing projects.
👇 Link to the OmniPulse website in comments!
De l'IA pour le parfum, mais quelle idée ? 😅
Design, systèmes de recommandation, IA générative… on rentre dans le vif du sujet avec Marc Sanselme dans Data Driven 101 !
Liens de l’épisode en commentaire 👇
Bravo Marc pour Data Driven 101 : pour ceux qui ne connaitraient pas encore, je recommande notamment les super épisodes avec :
Guillaume Lample (Mistral AI) Stanislas Polu (OpenAI, Dust)
