Osmo

/ Startups / By

Founder: Alex Wiltschko
Founding: 2022
Mission: Giving computers a sense of smell to improve the health and wellbeing of human life
Employees: 30 & 50% Local
Workplace: Hybrid (Mostly In Office)
Stage & Capital Raised: Series A & $68.5M raised
Investors: Lux Capital, GV, Bill & Melinda Gates Foundation
Key Customers: Coming Soon
Glassdoor Rating: N/A
Valuation (estimated): $100M – $500M (assuming they sold ~20% of the company in the Series A fundraise)
^ this is a useless number. There is no tangible valuation until the business is sold or goes public. Don’t forget it!

Osmo is building a first-of-its-kind technology to digitize smell. If you could capture scent in the same type of way that you might take a photograph, what would you do? Would you capture the smell of your pet? Significant other? Your favorite hike? The time of autumn when all the leaves start to fall and they crumple under your footsteps for just an instant? Osmo is tackling a solution to a challenging question: if you could capture and define smell, then play it back, what are the applications? 

Osmo is a machine learning company for chemistry, redefining what smell means. While other startups may have historically focused on servicing the seven deadly sins, Osmo is digitizing one of our trickiest five senses to bring smell to computers.

Founded by Alex Wiltschko in 2022, Osmo’s research was incubated at Google Research. After 5+ years of research, with the help of Josh Wolfe at Lux, Krishna Yeshwant at GV, and Andy Palmer, Osmo was spun out to stand on its own with institutional funding from Google Ventures, Lux Capital, the Bill & Melinda Gates Foundation and others.

Alex had studied as an olfactory neuroscientist, completing his PhD in Neurobiology at Harvard. While studying for his PhD he co-founded Syllable Life Sciences, using AI and computer vision to decipher body language in an effort to accelerate preclinical development and help build better treatments for disease. The company was later acquired, Alex completed his PhD, and then he went off to Google where he led research at the intersection of machine learning and biology. 

Scent is hard to describe. If we say something is blue, you know what that means. But if we describe something as “citrusy”, first you need to know what that means. And it doesn’t necessarily translate across cultures or languages. Smell is fundamentally critical to how we understand the world and we’ve been given this big chemical sensor on our faces we don’t totally understand! 

Digitizing smell is a once-in-a-generation opportunity. With computers we’ve been able to digitize vision and audio, but we haven’t been able to do it with scent quite yet. The progressive power of machine learning now makes unlocking digital scent possible.

The algorithm Osmo is building already predicts what something might smell like better than humans based on understanding its chemical structure. You can’t teach a computer to smell how a human smells, but you can teach a computer to understand the relationship between smell and chemical compounds. Like large language models have turned letters, words, and sentences into numerical probabilities, the graph neural network that Osmo is building helps pair molecules with “smell labels”. Making their own data repository, they have trained models on olfaction by identifying the taxonomy that describes something like ketchup – acidic, tomatoey, burnt – and on the back end mapping to the molecules that constitute the substance.

Osmo’s first commercial application will be in fragrance. It’s a large, $60B global market, with rapid international growth (src). Fragrance companies are effectively “bespoke chemical manufacturing companies” building scents & flavors. Many of the compounds are expensive to make, hard to resource, toxic, and not exactly biodegradable. Sustainable replacements aren’t being discovered fast enough to take these existing products off the market. The discovery process is manual too. 

Fragrance producers will search through the chemical space to find a “hit” and then experiment step by step around adjacencies. By statistically mapping odors Osmo can search for a scent at the molecular level, giving the “olfactory precept” that a researcher might be looking for, and then patent & productize the compound for licensing.

They’re taking the same computational approach used by biotech companies but, instead of looking for an end pharmaceutical use case, they’re looking for “woody” or ”mossy” or “citrusy” outcomes. There are endless uses for defining what it means to smell through a computer; you just need a lot of data to start reading and writing scent. 

Today the team represents 30 software engineers and chemists, planning to grow to approximately 50 team members by the end of 2024. They are working in labs between Cambridge and NYC, planning to grow in both locations. New York is home to the F&F (flavors & fragrance) industry and Boston is home to some of the world’s best computational chemists & machine learning researchers at world class universities like MIT & Harvard. 

Another interesting segment of their business, the source of the Gates Foundation grants, is their work in insect repellency. Human olfaction and mosquito olfaction tend to be similar and Osmo is leveraging its current models to search and predict molecules that might be more effective than “DEET”, the active ingredient in many repellants. 

Osmo believes they are at the cutting edge of olfactory neuroscience research worldwide helping to build cheaper, more environmentally friendly, easier to synthesize molecules for a variety of use cases (starting in fragrance & repellency). In 2024 they are focused on developing their GTM and partnerships strategy, working to get Osmo-developed molecules out into the world, and continuing to build around “reading and writing scent”. What’s the coolest part about working at Osmo on this technology? Getting to be the first person to smell something new.

Operators to Know (Locally):

My investigative powers continue to need work so apologies to the Osmo team I know I missed many up & coming operators internally

Key Roles To Be Hired:

If I were interviewing here are some questions I’d ask:

  • What are the key opportunities for commercialization and what current obstacles stand in the way?
  • Could you share some details around the team’s 2024 initiatives and goals? 
  • What is the long term vision for Osmo? What are the key roles you’ll be looking to add in 2024?

We’re optimizing for readability here so to learn more about Osmo you’ll have to D.Y.O.R. I’m excited to watch this team bring smell into the digital age. All humankind applaud your efforts. See you around town!

Subscribe
Subscribe