‘It’s crucial’: how AI is reshaping the fragrance industry - FT中文网
登录×
电子邮件/用户名
密码
记住我
请输入邮箱和密码进行绑定操作:
请输入手机号码,通过短信验证(目前仅支持中国大陆地区的手机号):
请您阅读我们的用户注册协议隐私权保护政策,点击下方按钮即视为您接受。
FT商学院

‘It’s crucial’: how AI is reshaping the fragrance industry

From hyper-personalisation to cutting costs, artificial intelligence is getting a hold on your perfume cabinet
00:00

{"text":[[{"start":7.65,"text":"Are you more conservative or unconventional? Which colour represents you the best? Do you see yourself as someone who has a forgiving nature? These are some of the questions EveryHuman will ask you before creating three “hyperpersonalised” custom scents in 5ml bottles for £42. The brand, founded in 2018 by Frederik Duerinck and scent designer Anahita Mekanik, uses artificial intelligence to match each shopper’s profile — created through an online questionnaire — to a palette of 52 accords, each comprising two to 20 ingredients. These are then mixed to create your signature fragrance."}],[{"start":47.4,"text":"On a quiet weekday morning at The Fragrance Shop in London’s Oxford Street, I recently watched as three small bottles moved on a conveyor belt underneath larger upside-down containers. The store is one of 11 physical retailers that host an EveryHuman “sensory machine”, which creates perfumes under your nose, adding a retro-futuristic touch to the experience (shoppers can access the same service online). According to one sales assistant, there is usually a queue at weekends to use it. “For us it is about making scent creation super-accessible for people and for brands,” says Duerinck. "}],[{"start":80.8,"text":"EveryHuman is one of a growing number of companies that are using AI to do so. Barcelona-based Scircle, launched by Álvaro Suárez in 2025, follows the same ethos with Aura, an AI-powered perfumer. There is no questionnaire here and the process starts by telling Aura, through an online chat, what kind of perfume you want. "}],[{"start":null,"text":"

"}],[{"start":null,"text":"
A machine making perfume
"}],[{"start":103.6,"text":"Suárez tells me it can be something as specific and personal (and, possibly, unappealing) as “the smell of my dog who passed away”. He explains, “In the past [bespoke] perfumers were only accessible to emperors and kings and royals. The perfumer would come and ask you questions until they understood what you wanted. We do exactly the same, but we democratise it.” A 20ml Scircle scent costs $95 and can be acquired online or through three physical locations in North America."}],[{"start":135.85,"text":"This kind of AI-driven application can feel gimmicky. But personalisation is a big draw for perfume shoppers, especially younger millennials and Gen Z-ers. According to market research firm Mintel, almost 45 per cent of under-45s use fragrances to express themselves. It’s a trend that has boosted the market for niche perfumes and spurred other fashions, such as layering, where shoppers buy and mix different scents to create their own. "}],[{"start":163.15,"text":"“It is for sure where customer trends are going and it enhances the customer experience,” says Claudia D’Arpizio, partner at consultancy Bain. But beyond personalisation, fragrance is “one of the industries that will be more transformed by AI in the way of working” behind the scenes, she adds. Its uses range from research and development to formulation, regulatory issues, testing and marketing, resulting in faster production cycles, lower development risk and — crucially — lower costs. "}],[{"start":195.5,"text":"The latter is one of the benefits of using Moodify, according to chief executive Yigal Sharon. The company, founded in 2017, started by digitising some 2,800 perfume ingredients and establishing how similar or dissimilar these ingredients were to each other. Then it formulated a mathematical model that could predict the smell created when different ingredients were added together. This allows Moodify to assist brands with the faster reformulation of existing fragrances to adapt to new regulations, or cut around 30 per cent of the cost by using cheaper ingredients. "}],[{"start":229.45,"text":"“Humans make perfumes with a lot of excesses, [such as] materials that could have been chosen better,” Sharon explains. “We are taking away the human inefficiency.” He adds that Moodify, which so far has raised $20mn from VCs and industry players including Procter & Gamble, can speed up the process of creating a new scent from the canonical 18 months to just three. "}],[{"start":251.6,"text":"Speed is an increasingly important asset for an industry that is now dominated by fast-changing, TikTok-driven trends. Equally crucial is ensuring that new launches are successful, something that has been elusive for perfumers for a long time, despite consumer panels and product testing."}],[{"start":268.6,"text":"“In fragrance there are a lot of launches that are not successful, but when they are they bring a lot of revenue and profitability,” says D’Arpizio. “Making the product right, reducing the time to market and reducing the risk of failure through AI is a big value enhancer in this industry where product development process holds a lot of value.”"}],[{"start":null,"text":"
"}],[{"start":null,"text":"
A series of images from New York which inspire a scent
"}],[{"start":289.35,"text":"FoodPairing AI, for example, focuses on proofing the success of products before they are launched on the market. The company, which works in the food, beverages and personal and homecare industries, has digitally mapped 15,000 aroma compounds and created digital twins of consumer types which, according to co-founder Bernard Lahousse, “can smell, taste and give feedback like a real consumer would do”. He adds that the correlation between what the digital twins say and what real consumers say is around 77 per cent accurate. “This speeds up the whole process enormously, because you don’t need a consumer panel, you don’t need to actually make a product [to test it].” "}],[{"start":328.25,"text":"From around 2018, sector giants including Givaudan, Symrise and dsm-firmenich, which make flavours and fragrances for consumer brands, have also started to launch their own internal AI-powered tools. For an industry that has long seen perfumery as an art form, this embrace of artificial intelligence can seem contradictory. According to Lahousse, who co-founded FoodPairing AI in 2009, interest from the big players really sped up post-Covid and with the release of ChatGPT in 2022. “Companies now see that if they don’t tap into AI, they will lose,” he says. It’s a view that is shared by Calice Asancheyev-Becker, French master perfumer and director of Givaudan Perfumery School, who says, more poetically, “It’s where the water is flowing. It’s not important; it’s crucial. We cannot avoid it.”"}],[{"start":378.05,"text":"Givaudan’s AI tool Carto, which was unveiled in 2019, comes with a touchscreen, an “Odour Value Map” of about 5,000 ingredients that perfumers can combine, and the possibility of creating instant samples. Asancheyev-Becker and Givaudan’s chief innovation officer David Kunau both stress that the tool’s main aim is to assist perfumers, not to replace them, in the creation and reformulation of scents. “There is no real digital nose that is working today,” says Kunau."}],[{"start":408.5,"text":"Creating one is among the big ambitions touted by Osmo, the AI start-up founded in 2022 by Alex Wiltschko, a former researcher at Google Brain, which has raised $130mn to date and wants to “give computers a sense of smell”. Leveraging breakthroughs achieved during Wiltschko’s time at Google, Osmo works with a “Principal Odor Map” that connects the smell of a molecule to its structure and allows the company to design new molecules to precise specifications. "}],[{"start":437.2,"text":"Wiltschko, who describes himself as a scent-obsessed computer nerd and has a PhD in neuroscience, believes that the digitalisation of smell will have life-changing consequences beyond the fragrance industry, including allowing machines to detect diseases through smell. Those goals, however, are “the North Star” for Osmo. Today the company works primarily with emerging indie brands and multinational groups, assisting them with anything from the creation of proprietary fragrance molecules to the ever-important optimisation of costs and timelines."}],[{"start":469.25,"text":"“Fusing artificial intelligence and scent is a very deep affair,” he says. “Really pragmatically today we help brands design their scents to go into products that people love and buy.” "}],[{"start":480.4,"text":"Follow us on Instagram and sign up for Fashion Matters, your weekly newsletter about the fashion industry"}],[{"start":494.65,"text":""}]],"url":"https://audio.ftcn.net.cn/album/a_1777873669_5958.mp3"}

版权声明:本文版权归FT中文网所有,未经允许任何单位或个人不得转载,复制或以任何其他方式使用本文全部或部分,侵权必究。
设置字号×
最小
较小
默认
较大
最大
分享×