The perfume makers that can’t smell a thing
Do you need a human being to create a beautiful perfume? This is the question that is asked when artificial intelligence (AI) begins to infiltrate the perfume industry.
Companies are increasingly turning to technology to create best-selling and unique fragrances that can be produced in minutes.
Last year, Swiss-based fragrance developer Givaudan Fragrances launched Carto, an artificial intelligence-based tool to help perfumers.
Through machine learning (a way in which computers automatically improve results by learning from past results) Carto can suggest combinations of ingredients.
Using a touchscreen, the perfumer can bring together different perfumes using data from the brand’s vast library of fragrance formulas, a much more efficient process than using spreadsheets. A small robot immediately transforms fragrances into perfumes, making it easier for perfumers to test their new perfumes.
“It’s about finding a way to spend more time with the perfumer,” says Calice Becker, vice president of perfumery and director of the Givaudan Perfumery School.
“Perfumers can choose from 1,500 ingredients and put it in a bottle without touching the ingredients. It helps make sure you don’t waste time and look at your notebooks.”
Becker says that the perfumery process has evolved over the years and this is only the next step.
“Until about 40 years ago, perfumers worked with all the ingredients in front of them and took the ingredients and wrote down the quantities and names of the ingredients on a piece of paper.”
The 1980s saw the introduction of computers and perfumers would create their concoctions through a system that looked like an Excel spreadsheet, he says.
An advantage of Carto is that champions are created instantly, giving them a competitive advantage. “We can adjust the perfume almost live with the customer,” says Becker.
“It is a great advantage not only because we buy time but there is more intimacy when we connect in front of the instrument.”
What was their client’s reaction? “We have some of the early users, but some say they will never use it,” he says. “I think it’s completely normal. But it created a lot of buzz from customers interested in seeing how they can see the creations with it.”
The German fragrance house Symrise has taken a further step forward and has collaborated with IBM Research to create an artificial intelligence called Philyra, named after the Greek perfume goddess, who currently studies aromatic formulas and customer data to produce new fragrances.
Philyra was taught similarly to a perfumer apprentice, who can study for 10 years before making good new perfumes.
Like Carto, Philyra can’t actually smell anything.
Instead, the families of smells, including flowers, oriental and Cypriots, have been coded together with the different needs of products such as shampoos, deodorants and skin lotions.
Artificial intelligence was also taught on how much of each ingredient would be appropriate.
Claire Viola, Symrise’s vice president of digital strategy fragrance, is the first to admit that she hasn’t been flawless.
“It’s machine learning and sometimes the results have been wrong,” he says. “It’s still a project, the more we test, the more it continues to improve. It always needs training. You have to qualify each new material, so you understand the difference between different floral and oriental perfumes, for example.”
But, he says, the more they invest in training, the more precise it becomes.
“We taught that he is like a perfumer … the car never forgets [compared to humans]. The good thing is that the machine has a selection of interesting perfumes and combinations that you would never have thought of. “
Since the machine has a database of nearly two million aroma formulas, the potential for a wider range of perfumes and combinations is enormous. In 2019 the Brazilian cosmetic company O Boticário worked with Symrise to launch the first fragrance using artificial intelligence.
More business technology
A company is shaking the industry by offering consumers the opportunity to play directly with technology.
In Breda, the Netherlands, ScenTronix allows customers to create their own personalized perfume based on a questionnaire they answer when they enter its algorithmic perfumery shop.
After answering questions such as how you see your role in life and what kind of environment you grew up in, the algorithm analyzes the data to create unique perfumes for the customer in seven minutes. Customers can purchase five samples for € 30 ($ 33; £ 26).
ScenTronix co-founder Frederik Duerinck says he wanted people to be able to wear a perfume that was a reflection of themselves at the time.
“The perfume industry relies heavily on branding and adapting your identity,” he explains. “I thought it would be a good idea to completely change the dynamic and therefore the perfume becomes what you are and not the brand”.
Duerinck agrees that there is a great physiological barrier to overcome since people are not used to paying for a perfume that they are unable to test until it is produced.
However, he says they try to overcome this obstacle by having someone on site to help customers. “Although I would say that 75% of the time is perfect, we always have a professional there [to help],” he says.
Margaux Caron, global beauty analyst for colored cosmetics and fragrances at Mintel, believes that artificial intelligence is a powerful tool for creating original fragrances.
“Not only [AI tools] they identify the olfactory white spaces, but they also considerably optimize the speed of creation of the fragrance for perfumers.
“Technology and science are sometimes depicted and perceived as cold and rational, but the category of fragrances is showing a warm, emotional, human approach to it. The partnership between AI and perfumers is anchored in this philosophy.”
So the introduction of technology marks the end of the perfumer? Not according to the ones I talked to.
“It will never stop the role of the perfumer,” says Becker. “The computer will never produce wonderful ideas. But it can help make them come alive.”
Viola agrees, adding that it is a complementary support to their work, allowing them to experiment much more.
“It isn’t replacing the perfumer,” he says. “It is helping them to be faster and more creative and freeing them from tedious tasks. It starts again and ends with the perfumer. They are the ones with intuition, emotion and feeling and drive the car to better results.”
At least for now, as Viola says, “It’s a man-machine collaboration”.