A doctor in chemistry and data scientist, Jean-Baptiste Coffin received the 2025 “Coup de Cœur” Award from the GDR Odorant, Odor, Olfaction (GDR O3) for his groundbreaking work conducted at the Institut de Chimie de Nice and within the start-up Perfumist.
His research, devoted to mass spectrometry assisted by artificial intelligence, opens new perspectives for measuring molecular similarity in the field of flavors and fragrances.
Fonds de Dotation Per Fumum: Could you briefly explain your research topic and what inspired you to explore the use of artificial intelligence in perfumery?
Jean-Baptiste Coffin:
“My research is titled ‘Comparing Perfumes by Combining Machine Learning and Analytical Chemistry.’
In concrete terms, I use analytical chemistry — particularly mass spectrometry — and apply artificial intelligence tools, or more precisely machine learning, to analyze the data. The goal is to detect an olfactory signature from mass spectra.
In the work I presented at GDR O3, I don’t focus on perfumes as complete entities (complex mixtures of raw materials), but rather on the molecules that compose them. Each molecule is analyzed in a mass spectrometer, which fragments it and produces a spectrum — a pattern of detected ions.
Normally, this spectrum is used to identify molecules. I use it differently: I leverage it to measure molecular similaritybetween different mass spectra, and thus attempt to infer olfactory similarity.
This work responds to a question raised by perfumers: how can we compare perfumes not through olfactory descriptors or pyramids, but directly through their chemical signal — that is, the molecules themselves?
My background suited this perfectly: I’m trained in chemistry and analytical chemistry, but I also have strong computing skills, which I developed from a young age. This dual expertise naturally led me to collaborate with Perfumist and to explore how artificial intelligence can help us better understand molecular similarity between aromas and perfumes.”
Fonds de Dotation Per Fumum: How can artificial intelligence help us better understand molecular similarity, and what is the goal of your research?
Jean-Baptiste Coffin:
“There are many ways to use machine learning to measure similarity between molecules.
The approach I chose relies on mass spectra, meaning on the signal itself, without prior knowledge of which molecules are present in a mixture.
Other teams have worked from molecular structures, creating “maps” of chemical space using algorithms. These maps position molecules according to their properties: floral, animalic, ethereal, and so on.
But in perfumery, formulas are confidential, so we don’t know the exact molecules. That’s why it’s so interesting to use direct analytical signals — mass spectrometry, infrared or ultraviolet light, nuclear magnetic resonance, etc. — to measure structural or olfactory similarities.
What’s unique about my PhD is that we’re no longer classifying odors (as in fougère, gourmand, etc.), but instead measuring a distance, in other words, performing a regression task.
Instead of saying “it’s identical / it’s not identical,” we quantify: similarity of 0.5, 0.8, 0.9, and so on. This allows for a finer, more nuanced approach.
I love this idea because perfumers are artists. Just as painters rely on a theory of colors, perfumers could one day rely on a theory of odors, which is still being built.
Our research, I hope, contributes a small but meaningful stone to this multidisciplinary edifice.”
Fonds de Dotation Per Fumum: Your background blends data science and olfactory creation. How do you plan to reconcile these two worlds after your PhD?
Jean-Baptiste Coffin:
“Machine learning has many applications in perfumery.
We can work on raw materials, on odors themselves, on quality or traceability. I don’t consider myself an artist, but rather someone who seeks to understand the ‘catalogue’ of odors, much like one studies a color palette.
During my industrial PhD, I worked with Perfumist, partly funded by the National Agency for Research and Technology (ANRT).
This allowed me to discover the industrial ecosystem of perfumery — professional events like TFWA Cannes, SIMPPAR, or the Barcelona Perfumery Congress.
And I realized how much AI, data, and research are awaited in this field.
In my view, the future lies in collaboration between composition houses, brands, and researchers — to better understand raw materials, detect fraud or counterfeits, and anticipate climatic disruptions affecting production.
The potential is immense — it’s just a matter of bringing the right people together.”
Fonds de Dotation Per Fumum: What have been the main challenges you faced during your research?
Jean-Baptiste Coffin:
“The main challenge — and also the greatest strength — is data.
At Perfumist, we have access to information provided by our partners, but formulations are often highly confidential. Since perfume formulas are not legally protected, they are subject to strict industrial secrecy.
To be able to work on the juice, we had to create our own specific database, without ever seeking to identify the molecules or ‘reverse-engineer’ perfumes.
That was an important ethical commitment for me: I didn’t want my algorithm to be used to create dupes (illegal copies). The goal is to generate value, not to destroy it.
Another challenge was multidisciplinarity.
My work brings together analytical chemistry, artificial intelligence, and the perfume industry. So we assembled experts from each field to cross our perspectives and move forward together.
I’d like to thank everyone I met along the way — especially at SIMPPAR and BPC, as well as the perfumers with whom I could exchange ideas and test the relevance of my hypotheses.”
Fonds de Dotation Per Fumum: What does the GDR O3 “Coup de Cœur” Award mean to you?
Jean-Baptiste Coffin:
“It’s a tremendous emotion.
I haven’t missed a single edition of GDR O3 since the start of my PhD — Paris, Dijon, Nice, and this year in Lyon.
It was the first time I could present a chapter that was no longer under embargo — the second one is currently being patented.
Receiving this award is a scientific recognition of the relevance of our work, both from an academic and industrial perspective.
I warmly thank the jury and the entire GDR O3 community for this honor.”
Fonds de Dotation Per Fumum: How do you imagine the future of perfumery in the age of artificial intelligence?
Jean-Baptiste Coffin:
“I believe AI can become a true ally of the perfumer.
It can enrich their palette, offer more subtlety and precision, while fully preserving their creative role.
AI can also help prevent unwanted interactions, anticipate allergies, or avoid problematic molecules.
Another strength of AI is its capacity to surprise. Sometimes, models find solutions we would never have thought of.
That’s what I love about this field — it pushes us to better formulate our questions, not just to find answers.
I’d also like to emphasize the importance of funding research, especially PhD programs.
Researchers, professors, and supervisors can’t do everything alone — PhD students are essential to scientific progress.That’s why I want to encourage the perfume industry to get more involved in CIFRE PhD programs, supported by the ANRT.
It’s through these collaborations that we can continue advancing our understanding of olfaction, scent, and perfume.”
