How AI is unlocking the power of smell

News Room

This article is part of “Build IT,” a series about digital tech and innovation trends that are disrupting industries.

Smell is perhaps the most powerful human sense. Think about the scent of fresh lavender or an evergreen forest; these might bring you peace or remind you of something unpleasant. Either way, you don’t have control over how they affect you.

The connection between smell and emotion has intrigued scientists for decades. Every person experiences scents and feelings differently, which raises the question, why do we react a certain way to scents?

In her book “The Scent of Desire,” Rachel Herz, a neuroscientist, theorized that smell was the first sense to form in organisms’ brains to guide them away from danger. This cognitive development is now known as the amygdala, a part of the brain that interacts with our olfactory receptors to help detect peril and induce fear.

In a study conducted by Herz and a group of collaborators at Brown University, the team found that scents could trigger memories and affect mood. These memories emerge when the olfactory cortex processes a smell and communicates with the limbic system, which influences behavior and emotions, prompting us to respond in a particular way.


Fragrance companies capitalize on this science. Some are taking a technology-driven approach, harnessing artificial intelligence to create evocative custom scents for consumers.

“Creation is such a beautiful thing,” Frederik Duerinck, a cofounder of the tech-fragrance company EveryHuman, told Business Insider. “It’s such a fulfilling experience.”

But it’s not easy. To create custom fragrances, companies need to know how customers react to different aromas — and that’s hard to gauge when people can’t communicate abstract feelings.

“We have the same kind of difficulty using words to deconstruct our emotional experiences as we have articulating our experiences of smell,” Hertz said in “The Scent of Desire.

Using AI to decode fragrances

Givaudan, one of the world’s largest manufacturers of custom fragrances and flavors in the beauty and food industry, has for over a decade focused on using AI to decode the complexities of smell. The company works on the business-to-business side of the industry, designing and creating scents for various companies worldwide.

“Fragrance is really a kind of black box,” Johan Chaille de Nere, Givaudan’s director of digital transformation, said. “And AI is beneficial to better understanding and to better decode what the consumer really feels.”

Givaudan launched Myrissi, its AI tool, last year to demystify fragrance development. With a database of more than 25,000 consumer responses from surveys and product tests, the company’s technology creates associations between scents and colors to provide custom mood boards for brands developing fragrances.

NOS Emotiontech, formerly No Ordinary Scent, is another company using personalized consumer data to aid perfumers. At first, the Swedish company offered direct-to-consumer perfumes using an AI platform that analyzed three user-submitted photos.

Now the company focuses on its business clients, using an AI algorithm and existing customer data to create briefs that help brands put on multisensory experiences.

With the help of scientific data and its trained algorithm, the company can also figure out which “scent families and notes” can “evoke certain emotional responses,” Sandra Kinnmark, the founder of NOS Emotiontech, told BI.

For example, one video-game company commissioned NOS Emotiontech to fill its gaming studio with a personalized scent that helped participants feel at home and lose track of time while gaming. The fragrance focused on citrus and solar notes to boost happiness.

The Swedish company also partnered with the Hallwyl Museum, one of the oldest museums in Stockholm, fashioned out of a 19th-century home, to create a guided “scented tour” for visitors to “actually experience tales from this era,” Kinnmark said.

NOS Emotiontech designed five custom scents dispersed on strips at different parts of the museum. For instance, the fragrance in the dining room was heavy on floral and citrus aromas to draw out the smell of flowers and the exotic fruits that were popular at the time. The weapons-room scent emphasized wood and cardamom to highlight the tour guide’s discussion of colonial trade.

A new kind of signature scent

Some fragrance brands use tech to decipher scent preferences and create personalized scents.

EveryHuman uses its AI platform, Algorithmic Perfumery, to create three custom perfumes for each shopper. It works by gathering data from a survey that asks customers over 20 personal questions, including, “What are your favorite activities?” and, “Which color represents you the best?”

“It’s really a way for people to engage with smell on a much-broader level and take ownership of their own identity,” Duerinck said.

The three scents are generated based on interpretations of the data, he said. One perfume is created from the answers to the psychological questions in the survey. The second combines the psychological responses with the user’s demographic information, such as age and location. The third perfume is created by considering how customers interacted with the survey, like how much time they spent on a question.

Each resulting perfume is built from 46 “building blocks,” EveryHuman’s term for its specific blends of scents, with each block containing anywhere from two to 20 ingredients. Some blocks are more familiar scent profiles like “rose” and “grapefruit,” while others are more abstract like “sheer” and “metallic.”

The tool can create over 500 billion permutations of fragrances, Duerinck said.

“This platform is about opening up the medium of scent so people can relate to it a little bit more easily,” Anahita Mekanik, a cofounder of EveryHuman, said.

In November, EveryHuman partnered with The Fragrance Shop’s flagship store in London so that customers could try their perfume-making machine.

A fragrance reviewer who goes by TJ Talks Scents told BI he tried the AI tool and said that it felt less daunting than the trial-and-error approach of buying perfumes online. “I think it’s a good idea because some people don’t know what they like in a perfume,” he added.

He said the machine had some technical difficulties but that he enjoyed two of the three generated fragrances because they matched his preference for sweet and woody notes. He added that the scents made him feel warm while living in the UK, where the weather is often cold and wet.

AI could make its mark in the beauty industry

One of the biggest concerns about AI is that it could threaten people’s jobs.

Chaille de Nere told BI he didn’t want that to happen in the fragrance industry. He said he believed “AI could support and inspire our perfumers, not replace them.” NOS Emotiontech takes a similar stance; the briefs it shares with brands might be AI-generated, but the company has a staff of perfumers who create the scents.

But some experts worry if AI initiatives will be built to last. Chaille de Nere described personalized perfumes as a “consumer trend” resulting from the industry experimenting with the new technology. Kinnmark added she’s concerned that personalization “leads to less” because it adds more products to the beauty industry and works against contemporary movements to limit the amount of spending on multitudes of items.

AI could have a lasting impact on the industry’s environmental regulations. Chaille de Nere said AI could support sustainability measures by creating chemical compositions made ethically and with safe materials. He added that Givaudan had been using AI to replace ingredients banned by local authorities or regulatory organizations because of their negative carbon footprint. In 2022, the chemical compound lilial — commonly found in floral perfumes — was phased out in all European Union products over allergenic concerns.


While AI’s role in the beauty industry has yet to take shape, the fragrance industry is betting on the technology. Perfume is moving further away from its status as a luxury item, and AI could be the key to making fragrances with more kinds of scents for more people.

For Chaille de Nere, that could also mean rethinking fragrance’s role as a beauty product and a means of a sensational experience.

“The main power of a fragrance is not to be liked or disliked,” he said. “The main power is to create emotions.”

Read the full article here

Share this Article
Leave a comment