What if computers could smell?

For decades, scent remained computing’s final analog frontier. Now, Osmo has accomplished what many thought impossible: digitizing smell. By training AI to decode the relationship between molecular structures and human perception, Osmo has built the world’s first olfactory intelligence platform—transforming scent into searchable, programmable data. While poised to reshape the $30 billion flavor and fragrance industry, its implications extend far beyond perfume, opening new possibilities across health, food, technology, and human-computer interaction.

Can AI smell you? This startup built a $130 million digital nose

Tech

fortuneaifeatures

Our mission is to give computers a sense of smell. For a century, we've been turning the world into data. Sight became pixels, sound became waveforms, but smell, the oldest sense no machine could digitize until now. For the first time ever, a teleported smell. With a single prompt, this startup says they can create a scent in just 48 hours. Completely reversing the model of how synth is designed and delivered a little bit fluffy, velvety. We see a lot of textures in perfumery too, but fragrance is just the beginning. The final frontier is using scent in the way that dogs can use scent to understand wellness or illness. This AI startup has raised $130 million to turn scent into data, perfume now, disease detection down the line. So what happens when you give a computer a nose? We're going to smell the whole world. For decades, 4 giants have dominated the $30 billion flavor and fragrance market. Now Ozmo's using AI to make it cheaper, faster, and more accessible. Neuroscientist Alex Wilschko created OSMO with a mission build the world's first olfactory intelligence, AI trained to predict what a molecule will smell like. What is OSO? Why did you create it? What we're trying to do is bring new, unique scents to more people in the world by removing the gates in the secrecy and frankly some of the arrogance that like, oh, only certain companies or only certain people can have a good scent. No, anybody can. We're democratizing scent. The idea of having shazam for smell, that you can walk into a room and just press a button and it just digitally can find the frequency of the music and match it to a correlated database, you should have that same feeling when you walk into a room and you smell a perfume or a cologne. You're like, what is that? Josh Wolf is a venture capitalist known for shaping cutting edge tech into real businesses. With a recently closed $1.5 billion fund, Lux Capital now commands more than $7 billion in assets. As one of the company's first outside investors, the firm helped raise $60 million for Ozmo's launch in 2023. How do you distinguish between what is scientifically possible and what is just hype? It's really sophisticated, hard. You need a PhD, and it comes down to three words. Does it work? If you're looking at something that is a hypothesis and you want to test it, then it's almost like a poker game, and you wanna ask how much capital accomplishes what in what period of time, and who will care. The bet is paying off. Osmo signed clients from indie brands to multinational CPGs. The company has authenticated sneakers by smell. All you have to do is get the shoebox close to the sensor, reduced animal testing, and received millions in funding from the Gates Foundation to support AI enabled insect control, altogether raising $130 million in just 4 years. How do you prove though that it's not just evocative, but it's also really a business? In our case right now, you just smell the results. I smell anxiety and 234 years of American capitalism. We took Osmo's tech to the floor of the New York Stock Exchange to see if a machine could actually capture the smell of ambition. To bottle a place like this, Osmo had to teach a machine to turn scent into numbers. What does it actually mean to collect data for scent? One is like you have to make or acquire an actual physical smell, and then you have to turn that physical smell into ones and zeros. The breakthrough came from an unlikely place, the AI built to discover new drugs. They figured out how to teach AI to deal with inputs that aren't shaped like a rectangle. So images are rectangles, they're grids of pixels. Text is a very long, thin rectangle, but molecules are atoms that click together in all kinds of different shapes. They're very much not a grid. And so there was a set of techniques developed to deal with that kind of input called graph neural networks. In 2024, Osmo captured the smell of a real plum, encoding it as data and rebuilding it across the lab. No human involved. We put that prompt of like give me the smell of the New York Stock Exchange into Osmo Studio, and Osmo Studio is able to take really anything like text and image and convert it into scent. What does the New York Stock Exchange smell like? So the first one is woody, floral, earthy. Well, it smells a lot better than the New York Stock Exchange, which is frankly a little bit of a sweatier place. But it does actually smell like the digital quality of it. Yeah, well, let me give you this direction and see what you think of this. That to me smells like the New York Stock Exchange. Did we get it? Yes, that to me smells like the New York Stock Exchange. This feels like a marble lobby where a musky warmth settles on polished stone. I can't stop smelling. At what point, if any, was there a human touch in this process? At the very beginning when you and I actually designed this, we're using a tool, we're using Osmo Studio to create scent, but like that's like using Canva or Photoshop or Figma. Like we were the designers. The brands behind most perfume, shampoo, and detergent don't formulate their own scents. They buy them from suppliers. That's the enterprise business Ozmo is disrupting, the platform layer behind scent. Now, to what extent is Osmo revenue generating? What parts are in research? For the first, I would say 2.5 years. Everything was R&D because what we were staring at had so much scientific risk and after we teleported to scent, we realized we've actually taken a lot of that scientific risk off the table and so we're a hybrid like we're highly technical, but the fastest growing part of our company is in manufacturing and in sales and we. finished fragrance to emerging brands that are growing super fast, but we also sell the software capabilities and access to our data for really large companies that also trade in smell. So we have a few different facets of our business, but they all are involved in making great fragrances for brands. The big suppliers often take months, with many requiring bulk orders. Ozmo, on the other hand, serves smaller brands, a market historically underserved. You could have done this as a B2C business, but you chose B2B specifically. Why? One thing I've noticed in B2C, you can be successful as a brand with an amazing story. And having amazing technology and science behind that can help, but it might not be the single reason why you win. And I wanted to set us up in a business where we're kind of dragged into the future where we keep developing our technology, keep developing our research to achieve even greater ambitions and getting better at the core science of scent and unraveling the mystery of scent to actually be good for the business. Who are your competitors? Who do you look at and say they're also marching towards the same mountain? I don't think there's anybody out there that's as weird as we are. I mean, there's companies that design and manufacture scent for sure. They've been around for hundreds of years. We're doing it very differently, right? So instead of it being a very gate kept process, we actually show our formulas to people. Traditionally, chemists test one molecule at a time. Osmo meanwhile says they use AI to screen billions at once, selecting only the best candidates to develop. The company says its success rates are 10 times higher and that it's 5 times cheaper compared to industry norms. Osmo also says it published more new fragrance ingredient patents than the rest of the industry combined in 2025. Where a traditional patent might cover one or two molecules, Osmo's patents can include up to 250, and they hope that scale means more profitability. What do you consider the moat for something like? Some of it is the data and then being able to be the only company that can predictably produce something with mathematical rigor every single time. We want it just like somebody was tuning a guitar. We want it to be a little bit more pungent or a little bit spikier. We want it to last for longer and Alex can basically say, OK, we're going to add this to this and we're going to be able to do it and we'll be able to do it every single time reproducibly and predictably. While AI expedites molecule discovery, humans are the artists who help craft them into a final scent. So this, this is a mini experimental robot. It can work day and night and can do a lot of samples because AI needs a lot of data, needs a lot of how do you make an apple, how do you make fresh, how do you make more. We have a lot of these, yes, because they're very helpful. Its name is Taylor Swift. That's it. Every robot here has a name. Christophe Laudemil, one of the few master perfumers in the world, leads Osmo's AI human collaboration. We know the AI is going to come and people are working on it, but what happens for us, I think as a perfumer to get into AI and now to steer it as well. It's not all AI, it's not all human. It's this beautiful blend of like creativity being unleashed in new ways that's enabled by the technology that we've created. But fragrance isn't the endgame. Ozma says it's building towards something bigger, a tool that could one day help detect disease. The final frontier is using scent in the way that dogs can use scent to understand wellness or illness. It's insanely hard. Dogs have been trained, who have sharper senses of smell than we do, to detect COVID, cancer, Parkinson's, Alzheimer's. And so every one of those diseases has a signature. For Osmo, fragrance is the commercial work paying for that longer term research. And the reason is we have to fund ourselves in order to do that, and the act of making fragrances actually generates data and understanding, which is strengthening our platform to make even more ambitious bets in the future. For Alex, that motivation is personal. His father died of brain cancer, a condition that some evidence suggests has a unique scent. What we need to uncover is perhaps those are also the same molecules in someone who has brain cancer like my father had. If you have a disease that's developing, there's going to be something that's going to change about the scents that are leaving your body, and nobody knows the details of what that is, but we know that dogs can detect it. And so what OSMO is is an organization, a group of people designed to unravel these mysteries kind of one by one. In a way that is responsible and resilient and designed to last over time.