Welcome.AI
    Skip to content
    Machine Learning

    Osmo's AI Outperforms Human Perfumers in Scent Detection

    Osmo is setting a new standard in fragrance and scent technology with its AI that rivals human olfactory abilities. As the first to develop a foundation model for smell, the startup is poised to transform industries beyond fragrance, including medical diagnostics and security.

    technologyreview.com•November 18, 2026•3 min read

    Key Facts

    • Osmo's AI model outperformed in-house perfumers at a major consumer goods firm, indicating strong competitive edge.
    • The dataset of 10M labels and 800M tokens reveals a significant investment in unique data assets.
    • Osmo's $8.5M funding from the Gates Foundation highlights strong financial backing for innovative applications.
    • The ability to create custom scents on-demand suggests a strategic shift towards personalized consumer experiences.
    • Critics point to reliance on human input, revealing potential vulnerabilities in fully automating olfactory tasks.

    Summary

    Osmo, an AI startup specializing in olfactory technology, asserts that its artificial intelligence can match human capabilities in detecting and distinguishing scents. This development holds significant implications for various industries, particularly fragrance, medical diagnostics, and security. Founded by Alex Wiltschko in 2023, Osmo has developed what it claims is the world’s first foundation model for smell, leveraging a dataset of over 10 million human labels and 800 million scent tokens to train its algorithms.

    The company has already made strides in the fragrance sector, where its technology is being used to assist brands in designing new scents. Notably, Osmo's AI model recently outperformed in-house perfumers at a major consumer goods company, demonstrating its potential to disrupt traditional practices in scent creation. This success could signal a shift in how fragrance companies approach product development, moving towards data-driven methodologies that integrate advanced AI capabilities.

    Osmo's approach to scent modeling involves a comprehensive analysis of odor molecules, utilizing spectrometers and chemical sensors to catalog a wide array of smells, from everyday items to complex natural scents. The startup’s unique methodology parallels advancements in other AI fields, where large datasets are employed to tackle challenging problems. However, the fragrance industry’s secretive nature and the high cost of necessary chemical sensors have historically hindered progress in olfactory science.

    The implications of Osmo's technology extend beyond fragrance. Wiltschko envisions applications in medical diagnostics and threat detection, such as identifying diseases or detecting explosives at airports. These potential uses highlight a growing interest in leveraging AI for complex sensory tasks, which could transform industries reliant on human olfaction. However, skepticism remains regarding the technology's ability to replicate the nuanced context of human smell perception. Critics argue that while Osmo’s model may excel in controlled environments, it may struggle to account for the complexities of real-world scenarios where context significantly influences scent interpretation.

    Despite these challenges, Osmo has attracted attention and investment, including a $5 million initial funding round from the Gates Foundation, which later provided an additional $3.5 million grant for research into disease-repelling molecules. The company's collaborations with major players like Procter & Gamble and StockX further validate its market potential. The ability to detect counterfeit products through scent analysis represents an innovative application that could enhance product authenticity in retail.

    As Osmo continues to refine its technology, it faces the challenge of improving sensor capabilities and expanding its dataset to enhance the model's accuracy and reliability. The potential for medical diagnostics, in particular, represents a lucrative opportunity, albeit one fraught with complexity and high stakes. The historical difficulties faced by companies attempting to penetrate this market suggest that Osmo must proceed cautiously, building a robust foundation before tackling more ambitious goals.

    Looking ahead, the evolution of Osmo's technology could redefine the boundaries of artificial intelligence in sensory applications. If successful, it may not only revolutionize the fragrance industry but also pave the way for AI systems capable of performing intricate sensory tasks across various sectors. The journey from the foothills of olfactory science to the peaks of medical and security applications will require sustained innovation and a commitment to overcoming the inherent challenges of replicating human sensory perception.

    Entities Mentioned

    Companies

    Osmo
    Procter & Gamble
    StockX

    Technologies

    AI
    scent tokens
    smell-printing robot

    People

    Alex Wiltschko
    Avery Gilbert
    Michael Schmuker

    Organizations

    Gates Foundation
    Google Brain
    University of Hertfordshire

    Key Concepts

    AI olfactory intelligence
    fragrance industry applications
    medical diagnostics potential
    threat detection
    data collection challenges
    human sensory perception
    context in smell
    scent creation technology

    Definitions

    AI olfactory intelligence
    The ability of an AI model to detect and discriminate between smells at a level comparable to that of an average human.
    scent tokens
    Descriptors that define the exact molecules and concentrations that create specific odors.
    smell-printing robot
    A robotic system that mixes ingredients according to precise formulas to create scent samples.
    fragrance industry
    An industry focused on the creation and manufacturing of scents for various applications, including personal care and household products.
    context in smell
    The idea that the perception of a smell can vary significantly based on the surrounding circumstances and associations.

    Use Cases

    • →Designing and manufacturing new fragrances
    • →Threat detection at airports
    • →Medical diagnostics for diseases
    • →Creating retail fragrances
    • →Spotting counterfeit products using smell
    • →Assessing molecules that repel disease-carrying insects

    Frequently Asked Questions

    What is Osmo's main technology?

    Osmo's main technology is an AI model designed to replicate human olfactory intelligence, allowing it to detect and create scents similarly to a human perfumer.

    How does Osmo collect data for its AI model?

    Osmo collects data through a combination of human sniffers who describe odors and chemical analysis, creating a comprehensive dataset of scent tokens and human labels.

    What industries can benefit from Osmo's technology?

    Osmo's technology can benefit various industries, including fragrance, food production, medical diagnostics, and even security through threat detection.

    What challenges does Osmo face in its development?

    Osmo faces challenges such as the high cost of chemical sensors, the secretive nature of the fragrance industry, and the complexity of human olfactory perception.

    What are the future applications of Osmo's technology?

    Future applications may include detecting spoilage in food, diagnosing diseases through smell, and enhancing AI's sensory capabilities in various domains.

    Where AI Leaders Stay Informed

    The latest AI intelligence, case studies, and research — delivered to your inbox every week.

    Free to read. Unsubscribe anytime.