The future of fragrance: How AI and Olfactory Intelligence are democratising perfume

Would you wear a perfume designed by artificial intelligence? Thanks to Alex Wiltschko, you may already have without realising it.

The founder of Osmo and its perfume creator tool Generation, Wiltschko is reshaping one of the world’s most secretive arts, using what he calls Olfactory Intelligence (OI) to democratise scent.

“There are 600 perfumers in the world. There are more astronauts alive than there are perfumers, but it doesn’t mean that you and I can’t design scent and can’t participate deeply in the scent design process. We believe that everyone should have the right to be creative and in the medium of scent and fragrance.”

The premise is simple – harness AI’s computational power to create perfumes to any brief. And there’s space, he argues. “There are maybe 100,000 fragrances on the market. But how many photographs? How many paintings? How many songs?

Osmo's lab might look like any other perfumery, but the company is looking to revolutionise access to making scent. Photo: Osmo

“I grew up in south Texas, and my parents were scientists,” says Wiltschko, who has been fascinated by smell since childhood. His early attempts at being a perfumer were, by his own admission, “garbage”. The issue was not talent, but access. “The world of fragrances has kept them a complete secret for 300 years, and that always stuck with me.”

Undeterred, he pursued a doctorate in olfactory neuroscience at Harvard University to study how the brain processes smell. “I learnt very acutely that we don’t know that much.” He went on to Twitter, then Google Brain, where he founded parent company Alphabet’s olfaction group, before stepping away – with Google’s blessing – to digitise scent full-time.

For three years, Wiltschko and his team have been “deep in the weeds of research, development and science”, making the breakthroughs that underpin Generation. “We can now take a smell from one location, digitally encode it, then decode and reprint it as the same smell elsewhere. That’s what generates the data to train the OI algorithm that lets us build the experience.”

Wiltschko has spent much of his career at the intersection of technology and perfume. Photo: Osmo

It’s easier to demonstrate than explain. During our call, we set a brief: niche, feminine and nostalgic. Within seconds, Generation produces Childhood Reverie, with top notes of orange, mint and bergamot “to evoke summer air and outdoor adventures”, layered over violet, berries, peach and honey, recalling childhood desserts. Alongside comes a moodboard of pastel skies. Generation also outputs the molecular formula, with precise ratios enabling a lab to create it. This is the most important step, Wiltschko explains, that would normally take a perfumer months to formulate.

Each client receives three OI-generated samples before Generation’s master perfumer, Christophe Laudamiel, steps in to refine them for the market. “We have to smell it, and everything that goes out of the door, there are human perfumers involved,” Wiltschko stresses. “The AI doesn’t make the process happen, it just speeds it up.”

During our half-hour interview, Generation produces two distinct but complementary perfumes from the same brief. “For a master perfumer, that’s a year and a half of work,” he notes. The implications are vast.

The world of perfume-making involves a high barrier to entry, which Osmo is looking to break down. Photo: Osmo

Where fragrance was once the preserve of heritage houses and conglomerates, Generation allows emerging brands – even individuals – to launch bespoke scents quickly and at low cost. In theory, a boutique could brief Generation and have a fragrance ready within weeks, bypassing the layers of exclusivity that have defined perfumery for centuries.

Though it was launched only recently, Generation has already produced a scent for Seattle’s Museum of Pop Culture, inspired by a nine-metre statue of guitars. “It even smells electric,” says Wiltschko with a laugh.

For all the innovation, he is unfazed by competition. Isn’t he worried others will copy the idea? “I’ve been working towards this for nearly 20 years, and it’s by far the hardest thing I’ve ever done. If someone wants to replicate it, start your engines. Good luck.”