We look at the changing role of perfumers in an industry being revolutionised by artificial intelligence tools and data-driven algorithms
Until recently, perfumers would take months, even years to carefully formulate a fragrance. They meticulously pipetted each blend and waited for the oils to harmonise, repeating that process hundreds of times for experiments that differed by a couple of drops of an ingredient. As scientist-artist hybrids, they touted their work as both technical and artistic. And yet, compared to their sister industries of fashion and make-up, the fragrance industry has lagged behind in adopting technological innovations. However, thanks to growing pressure from niche small brands, as well as the ever-growing consumer demand for new fragrances, the fragrance industry is now starting to introduce and experiment with artificial intelligence tools and data-driven algorithms. The last time the industry saw this type of innovation was with the introduction of synthetics in the late 19th century, which forever changed the game in terms of new formulations. Chanel No.5, arguably one of the most famous perfumes to first feature synthetics as a main note, led the wave of these new scents. Over a hundred years later, a system upgrade is long overdue.
In April of this year, Givaudan, a major fragrance and flavour houses based in Switzerland, released an artificial intelligence-powered robot called Carto. Through an interactive touch screen system perfumers drag-and-drop colour-coded circles representing ingredient notes into trial formulations. These get sent to Carto’s instant-sampling centre that blends those ingredients faster than any human hand could. With the manual process out of the way, the company hopes that perfumers will be able to explore new ingredient combinations using the playful visual system backed by an extensive data library of fragrance formulations.
In this scenario, the perfumer still stands very much as the creator, and Carto acts as a digital supporter. “It allows us to experiment much more than we can today, to dose our formulas in the most performant way, and we, perfumers, bring the creative touch, the one most important part that can’t be replaced by any system,” says Calice Becker, perfumer and director of the Givaudan Perfumery School. Indeed, now that perfumers don’t have to do the menial tasks of blending, dosing, and measuring, their time can be spent on exploring new ideas and formulations. Technology lifts the burden of execution while elevating perfumers to creative directors.
Reducing time and labour involved in the fragrance creation process has important business implications for anyone hoping to play successfully in the rapidly growing industry. Analysts predict the fragrance market will grow from US$39 billion in 2018 to US$48 billion by 2024.
The challenge companies face is how to bring more fragrances to the market with less work, while also preventing the perfumer from olfactive burnout caused by the acceleration of the industry. As it stands, there simply aren’t enough perfumers to meet the demands of the consumers.
“Artificial intelligence is reshaping jobs and industries, not replacing them” Marion Costero
“If part of the work rather than the full creative work can be lightened up thanks to AI, so be it. We don’t have so many perfumers around to waste their time,” says Christophe Laudamiel, co-founder and master perfumer of boutique perfume company DREAMAIR. Having worked on iconic fragrances like Fierce by Abercrombie & Fitch and Polo Blue for men by Ralph Lauren, Laudamiel’s went on to launch his own line, The Zoo a few years ago. He’s a passionate advocate for the perfumery world and has donated hundreds of formulations to academic AI research to keep pushing the boundaries of fragrance. Even with the help of AI “there’s still plenty of work to do within the creation of a fragrance,” he says.
While Carto acts as a perfect apprentice to the perfumer, other technological innovations actually complete the full creative work. Created by IBM Research and Symrise, another major fragrance and flavour house based in Germany, Philyra does everything from start to finish, with the exception of a few final touches. Like Carto, this AI tool uses advanced machine learning algorithms to comb through hundreds of formulations and raw ingredients to identify any patterns or trends that lead to successful fragrances, within specific demographics. With that, it’s able to create a perfectly formulated fragrance that’s backed by data. Well, almost perfect since the perfumer will do a final quality check and edit the formulation with their tens of years of real-world experience.
“Artificial intelligence is reshaping jobs and industries, not replacing them,” says Marion Costero, senior perfumer at Givaudan. It’s certainly reshaping things. This month, Brazilian cosmetics brand O Boticario officially launched the world’s first-ever AI-designed fragrances, Egeo On, made in collaboration with Philyra. According to Tiago Martinello, research and development head at O Boticário, using AI reduced the creation time from three years to six months. The business implications of reducing production time to one-sixth are huge. With the help of Philyra, companies could create six times as many fragrances in the same amount of time it took to create one, and they would be hyper-personalised to specific demographic groups.
In theory, both Carto and Philyra could create a perfectly successful perfume without any human input. But perfumers and C-level executives are unanimous in their belief that these machines will not replace the nose – the term used in the industry to described perfumers. Not only do the robots not have a biological nose, making smelling impossible, they also can’t provide the emotional touches of a human. Being both an art and science, perfumery can only be reduced to numbers and algorithms to a certain extent.
“I don’t feel this is threatening perfumers at all,” says Laudamiel. “On the contrary, it can only give more buzz and consideration to the art of creating perfumes and will bring all kinds of new individuals to an industry well known to be rather hermetic.”
"If we look at it from an AI and data science perspective, we see a lot of unexplored territories" Frederik Duerinck
One such individual is Frederik Duerinck, founder of Netherland-based company Scentronix, which is focused on bringing perfumery customisation to the market not only for professionals but eventually for consumers. Turned off by the fragrance industry business model where, according to Duerinck, only a small group of people control what comes out on the market, Duerinck wants to bring control back to the consumer. “Let them decide. Let them create,” he says.
Duerinck’s AI software and compounding hardware machine – not so dissimilar from Carto or Philyra – use a variety of data to create a personalised fragrance for each person who interacts with the system. This is the star of his Algorithmic Perfumery exhibit that he’s been showing in Europe and North America.
“On the formulation side, a lot of things are basically following what's successful. What comes out is tweaked to what the marketing department came up with when they did their research on what has high potential, or when they tried to clone something that’s doing very well. If we look at it from an AI and data science perspective, we see a lot of unexplored territories which has the potential for a much broader scope of fragrance,” says Duerinck.
Having shown his work across the world, Duerinck’s next goal is to get these tools into the hands of the perfumers and consumers. He’s actually working with two perfumers who said that they never expected to work with these kinds of tools. And the cherry on top is that because the AI tools use data to formulate, these combinations can be electronically sent to other machines around the world for a virtual smelling session. “We have a perfumer in Switzerland, another in the Netherlands, and an evaluator in New York, and all three have a machine. They can do an evaluation of a fragrance through Skype without being in the same location. In a way we brought perfumery to the cloud,” says Duerinck.
The adoption of AI-powered technology by the fragrance industry has widespread implications: faster production times, never-before-paired raw materials, and the first steps toward digitization through data. All without replacing the perfumer themselves. “I think it is going to help expand the market greatly, and invent new ways of creating,” says Laudamiel. AI machines creating fragrances alongside humans: this is perfumery 3.0.