From Molecules to Memories: The Era of Scent Teleportation Begins

Can you imagine sending that beautiful scent you smelled on a hike as easily as a song or picture to your friend or family?

Well, now you can because digitization has made it possible. That’s right. Much like all the other parts of your lives, scent is even getting digitized!

Digitizing smell, also known as digital olfaction, is the process of capturing, transmitting, and recreating scents using sensors and software.

By copying how the human brain identifies and differentiates odors, digital olfaction can be used to create digital scent media, develop new products, detect diseases, predict molecules, and grow food.

Compared to pictures, smell has far bigger potential to influence humans. After all, olfactory stimulation triggers more brain activity than visual stimulation.

Despite this, the majority of science advances have been around sight and sound. Today, smart devices like phones and laptops offer nearly everyone impressive built-in capabilities that are primarily centered around sound and light. Scent has yet to get the same technological capabilities and advances because, for one, it is personal.

The olfactory region in the brain is directly connected to the hippocampus, which is linked to memory and cognition, and the area responsible for processing emotion (amygdala). As a result, aroma has the powerful ability to trigger a specific memory.

But most importantly, it’s extremely complex. There are about 1000 genes for scent receptors, which encode an equivalent number of olfactory receptor types.

Interestingly, there are three known types of photoreceptors in the eye, while olfactory receptors are distributed across 6 million specialized sensory nerve cells residing in the upper part of the nose.

It is actually this complexity of smell that allows our noses to detect Parkinson’s disease. Meanwhile, animals like dogs and bees have such a profound sense of smell that they can even sniff cancer and Alzheimer’s. However, the same complexity is the reason why we have such a poor understanding of the olfactory system.

Things, however, are changing as scents get digitized.

Charting the Invisible: Building a Digital Map of Smell

Trying to understand smell has been going on for a long time, but so far, it has been rather limited.

However, technological advancements have led to a growing focus, resulting in the creation of the Digital Olfaction Society (DOS). It aims to create devices to record smells, turn them into digital data, transmit them, and finally recreate them in different parts of the world.

DOS is currently leading a global project to capture and archive the scents of diverse locations for cultural preservation. For this, the group is inviting experts from across the globe to digitize the scents from their local areas, which will be reconstituted by a dedicated team in Tokyo.

A couple of years ago, researchers from Japan’s Kyushu University, in collaboration with the University of Tokyo, also developed an olfactory sensor called ‘artificial nose,’ which could identify individuals by analyzing the compounds in their breath. Built with a 16-channel sensor array, the ‘artificial nose’ was able to biometrically authenticate with an average accuracy of over 97%.

Now, Osmo is successfully teleporting scent from one part of the world to another. A major player in this field, Osmo, is using Google Cloud AI technology to digitize smell in order to “give everyone a chance at a better life.”

Osmo envisions a new era where computers will be able to generate smells much like they generate images and sounds.

In an interview with CNBC a couple of months ago, Osmo co-founder and CEO Alex Wiltschko shared his obsession with smell, saying:

“It’s been my passion to try to understand smell. It’s this very powerful emotional sense, yet we know so little about it.”

Wiltschko, who has studied neuroscience at the University of Michigan and olfactory neuroscience specifically at Harvard, started Osmo as a research project at Google (GOOGL +1.12%) before launching it as a startup in 2022 with $60 mln support from Google Ventures and Lux Capital. Late last year, the Bill & Melinda Gates Foundation also awarded Osmo with a $3.5 million grant, adding to its $5 million equity investment.

At Google Research, Wiltschko worked on validating the digital olfaction research on the Brain team.

There, the focus was on using Graph Neural Networks (GNN) to achieve a breakthrough in predicting the smell of a molecule from its structure. Further research was conducted on designing new molecules that hadn’t been smelled before and then predicting them with high precision. The team also created molecules that repel mosquitoes but smelled very potent in human trials.

In 2022, Wiltschko, along with Richard C. Gerkin, Google Research, wrote about using molecular maps to understand odor.

The team proposed sensory maps to help measure a smell, which is produced by molecules that drift through the air, then enter our noses and get attached to sensory receptors there. But with billions of molecules capable of producing a smell, it’s hard to find out exactly which molecules produce which smells.

While we have maps for colors through the color wheel and use frequency to map sounds, there aren’t any useful maps for smell due to the complexity involved, as we stated above.

A few years ago, before this blog, the team developed the GNN model that explored thousands of examples of distinct molecules paired with smell labels to learn the relationship between a molecule’s structure and the probability of having a smell label.

Each molecule was represented as a vector expressing that molecule in its odor. Based on that, a “Principal Odor Map” (POM) was introduced to identify the vector representation of every odorous molecule.

Click here to learn about a new type of olfactory chip  that can simulate natural sense of smell.

A New Reality: Scent Teleportation is Finally Here

Last year, Wiltschko, Osmo CEO, talked about using Google Cloud’s AI technology to address the challenge of mapping our sense of smell to improve human well-being.

Over the past couple of years, artificial technology has grown substantially in popularity and usage. According to a McKinsey survey, AI adoption has increased to a whopping 72% in 2024 as it is used increasingly in more parts of the business. AI is actually projected by PwC to contribute $15.7 trillion to the global economy by 2030.

AI has simply left no sector untouched, but unlike AI chatbots, which can be trained on data from “the entire internet,” a similar digital library of information on scents hasn’t been available.

A breakthrough in machine olfaction came through the DREAM Olfaction Prediction Challenge hosted nearly a decade ago for teams worldwide to share their ML models.

Since then, this dataset has grown, and now, Osmo is using AI to catalog the exceptionally broad world of scents. It is estimated that as many as 40 billion molecules are likely to have an odor, but only 100 million of them are known.

The startup uses Google Cloud AI solutions to process this immense amount of data. One of the solutions includes Vertex AI, which is used to train AI.

“It was superhuman in its ability to predict what things smelled like.”

– Wiltschko

Once the machine learning models are built, trained, and deployed at scale, they are run across a large molecule library using Dataflow and have outcomes in BigQuery and Bigtable.

While the end goal is to enable computers to do everything that our noses can do—eventually help medical professionals detect diseases—in the short term, Osmo wants to make sustainable and safer aroma molecules for fragrances like perfume, insect repellant, and laundry detergent.

The fragrance in these products is usually “designed by a very small number of secretive companies,” and Wiltschko believes his company “can do better… by building better and safer ingredients that aren’t toxic … and don’t irritate your skin or your eyes.”

With that, Osmo has successfully teleported the scent of a plum. As the team explains it, this means Osmo has taught computers how to read, map, and write a scent from its original form and then recreate the smell or its essence in another location.

“This development has massive implications across industries from fragrance, health, security, agriculture, and more,” noted Osmo in its video update.

Introduced earlier this year, what’s described as “Osmo’s most ambitious project,” Scent Teleportation has been successfully achieved in the lab, announced Osmo, detailing the process.

The first successful attempt involved a simple slice of coconut but involved a novel approach, which requires no human intervention except at the input and output levels. This is the result of weekly experiments conducted using modern software and AI. The team noted:

“(Since first announcing scent teleportation), we have made an astonishing amount of headway. Every month, we make Scent Teleportation a little bit quicker, a little more automated, and challenge it to transmit more complex smells.”

The Science Behind Scent Teleportation: How Does it Work?

The way the scent teleportation technology works is that first, the selected scent is introduced to a machine called the Gas Chromatography-Mass Spectrometry (GCMS), which is a technique used to identify and analyze substances in a sample.

While a liquid can be injected directly, headspace analysis is used for a solid sample, like a plum. This analysis involves trapping the scent in the air around the object and then absorbing it through a tube.

The GCMS system recognizes the raw data and then sends it to the cloud, where it is displayed as a coordinate on its POM. This sophisticated AI-driven tool maps odors into a multidimensional space to help predict what a specific molecular combo smells like and enables more accurate recreations.

The formula is transferred to the Formulation Robot, which blends various scents to produce the sample. These tools are constantly put to the test to help refine processes.

In the last step, the team compared the original sample and the copy with the goal of accomplishing a high degree of similarity. The team aims to recreate scents that align with their original odor families, and with every trial, they were able to better capture every nuance inherent to odors.

“We are doing something that no one has managed before.”

– Osmo

The thing is, some molecules barely register on sensors, but despite their subtleness, they have a significant impact on the overall scent.

Also, just moving one tiny thing around in a molecule can potentially change its scent from roses to rotten eggs. Then there’s the problem of some molecules involved in everyday scents remaining unidentified.

With its data collection process, which is currently in progress, Osmo aims to solve some of these problems and find new ways to recreate molecules.

The team has already created the “biggest AI-compatible scent data bank in the world,” which is required to train algorithms and refine their scent senses. However, there’s still the challenge of making all this data usable, which needs human guidance—a hurdle the Osmo team hopes to overcome soon.

In the next step, Osmo will open select demos to the public so that they can directly experience the scent teleportation and review the startup’s technology and how effective it is in reproducing the smell.

With this, the startup gets all that much closer to bringing the world together in one of the most essential ways, where a beautiful scent that you smell can be easily shared with someone else. And then, eventually, Osmo intends to have disease detected with scent, which is “not going to happen this year or anytime soon, but we’re on our way,” said Wiltschko.

Companies Helping Advancing the Nascent Tech

Now, let’s take a look at companies that are working on creating this new virtual experience where one can share smells of perfumes and coffees over the internet, much like how their pictures are posted.

Currently, there are a few innovative startups that are pioneering new approaches to digital smell technology.

Aryballe is one that combines biochemical sensors, ML, and advanced optics to collect, display, and analyze odor data. Its hardware solutions are based on Aryballe’s silicon photonics-based platform, which detects, records, and recognizes odor data. Aryballe also offers a solid-state odor sensor called Core Sensor Module (CSM) that is sensitive to thousands of odors and can be integrated into customers’  fluidic, mechanical, and electrical architecture.

Aromyx is another consumer-facing biotech startup that measures taste and smell using synthetic biology and data science. Recreating the human nose biology provides insights to their customers. Businesses can use Aromyx’s software and AI programs to deploy their data into its comprehensive Cloud for comparison, modeling, simulation, and interpretation.

Monell Chemical Senses Center, meanwhile, is an independent, non-profit scientific research institute that is developing large-scale ML algorithms to predict odor quality based on molecular structure and ultimately build devices in collaboration with commercial partners to detect, generate, and share scents. Its Project AROMA is currently evaluating the best ways to design a universal system of classifying odors.

Given that digital olfaction is a highly specialized and nascent field, there aren’t really any public companies that are directly involved in its development. However, there are some that are making advances in related technologies.

1. Thermo Fisher Scientific (TMO -0.87%)

The $211 bln market cap Thermo Fisher Scientific, whose shares are up 3.95% YTD as they trade at $551.74, can contribute to digital olfaction through its GCMS system. Its solutions are designed to fit any laboratory workflow.

The company has an EPS (TTM) of 15.95, a P/E (TTM) of 34.60, and a dividend yield of 0.28%. For Q3 of 2024, it reported revenue of $10.60 billion.

2. International Flavors & Fragrances (IFF -1.16%)

This $23.59 billion market cap company primarily works on creating fragrances and flavors but has also invested in scent innovation technologies like AI-driven fragrance creation and sensory perception.

The company created its in-house AI tool many years ago and uses it, along with consumer perception data and neuroscience, to design scents. “We started building an extensive understanding of the connection between smelling and scents and emotions,” said Judith Gross, vice president of communication and branding, scent division, at IFF, adding that the database has shifted from simple emotions to colors and textures.

Meanwhile, its Sensory Perception technology is a microencapsulation delivery system for textiles that allows fragrance to be released from clothing and other textiles over a period of time.

As of writing, IFF shares are trading at $92.56, up 14% YTD, while having an EPS (TTM) of -9.10, a P/E (TTM) of -10.15, and a dividend yield of 1.73%. For 3Q24, it reported net sales of $2.93 billion and $702 million in cash flows from operations.

3. Alphabet (GOOGL +1.12%)

As we shared, Google’s AI innovation has played a key role in Osmo’s digital olfaction research. Given the vast potential of AI, the technology continues to be the focus at the tech giant, which noted during the 3Q24 earnings call:

“We are uniquely positioned to lead in the era of AI because of our differentiated full-stack approach to AI innovation, and we’re now seeing this operate at scale.”

AI innovation at Google involves a robust AI infrastructure that includes data centers and chips, research teams advancing AI research and building models, and a broad global reach through its products.

For this quarter, Google’s parent company reported revenue of $88.27 billion, which grew 15% YoY. Its cloud revenue came out particularly strong at $11.35 billion, an increase of almost 35% from a year ago, which was driven by AI offerings.

Conclusion

Up until now, sharing smell has been limited to sharing our memory or experience of it. However, the digitizing smell will allow for the actual recreation of the smell and, as such, have the potential to unlock countless new possibilities across industries, including product designs, extended reality, quality control, and healthcare.

So, from creating more sustainable aroma molecules for everyday products to predicting molecules that repel insects to prevent mosquito-borne diseases like malaria, smell digitization can help create memorable experiences and improve human health, creating a better future for us all!

Click here to learn all about investing in artificial intelligence.