Osmo Claims Its AI Nose Rivals Human Olfaction—But Can It Smell

Osmo Claims Its AI Nose Rivals Human Olfaction—But Can It Smell Beyond Perfume?


Osmo's smell model is already being used in the fragrance industry. But are medical diagnostics and threat detection a sniff too far?


By Amit Katwala | November 18, 2026




A Startup Betting on Digital Scent


Osmo, a startup building artificial-intelligence models for smell, says its "AI nose" can identify and describe odors as accurately as a human—a claim that, if validated, would mark a significant step forward for digital olfaction.


The company's smell model is already finding commercial traction in the fragrance industry, where it can assist perfumers and product developers in characterizing and reproducing scents. But Osmo has broader ambitions: it envisions applications in medical diagnostics, where subtle odor signatures can signal disease, and in threat detection, where the ability to identify hazardous substances by smell could be lifesaving.


Why Smell Is AI's Final Frontier


Smell has long lagged behind vision and hearing in the race to digitize human senses. Computer vision and speech recognition have matured into mainstream technologies, but olfaction remains notoriously difficult to model. Unlike light or sound, scent molecules don't map neatly onto a simple physical spectrum, and human perception of smell is highly subjective and context-dependent.


Osmo's approach—training machine learning models on large datasets of scent molecules and their corresponding human descriptions—aims to bridge that gap. If the company's claims hold up, it could open the door to a new class of applications that rely on machines interpreting the chemical world the way humans do.


The Hard Part: Proving It Works Beyond the Lab


The fragrance industry offers a natural first market, with well-defined use cases and tolerant of incremental accuracy. But medical diagnostics and threat detection demand far higher reliability. A model that confuses a benign compound with a hazardous one—or misses an early disease biomarker—could have serious consequences.


As of 2026, regulators and industry observers are watching digital-olfaction claims closely. Osmo's bold assertion that its AI nose matches human capability will need independent validation before it can move from perfume labs into clinics and security checkpoints.


What to Watch


The coming year will be telling. Key questions include how Osmo's model performs outside controlled settings, whether peer-reviewed studies back its accuracy claims, and whether partners in healthcare and safety adopt the technology at scale. For now, Osmo has made its case in fragrance—and sparked a bigger conversation about how far machine smell can go.




Originally reported by MIT Technology Review.

via MIT Tech Review AI

Related