AI Personalized Scent Analysis Engine ScentMind Launches: Predicting Human Olfactory Responses From Molecular Structures

AI Personalized Scent Analysis Engine ScentMind Launches: Predicting Human Olfactory Responses From Molecular Structures

ScentMind, the AI scent analysis engine jointly developed by Japan's Ajinomoto and Google DeepMind, officially launched on June 12. The system achieves, for the first time, direct prediction of human olfactory perception from the three-dimensional structure of scent molecules, with prediction accuracy reaching 89% of professional perfumer level.

ScentMind was trained on structure-odor correlations of over 50,000 known scent molecules and can predict the olfactory descriptors a novel molecule is likely to produce, including scores across 12 dimensions such as floral, fruity, woody, and musky. In blind tests, the engine's predictions showed 89.3% agreement with human olfactory evaluations.

The technology is already being applied in perfume design and food flavoring. France's Chanel has used ScentMind to accelerate its new perfume development process, cutting formulation screening time from six months to three weeks.