AI-based Software for Olfactory Prediction

Imagine you could get the tedious lab work of odorant testing done by a software to free up the time of your experts for creative activities and innovation. This would shorten time-to-market in scent and materials development and offer your customers a greater variety of products. Together with you, we want to work on this. 

Customized Scent Prediction Models

By using AI-based models, we want to enable digital scent prediction & materials design without physical laboratory tests.

Customized Scent Prediction Models
© Fraunhofer IIS/Bianca Möller

Our customized AI-based software for scent prediction & materials design based on the relationship between scent and molecular structures will help you to:

  • Avoid redundant laboratory experiments and accelerate development processes
    How: Digital prediction of olfactory properties
  • Identify malodor and its source
    How: Detection of the source of smell within the molecular structure
  • Take advantage of in-house data, customary odor-descriptors and other individual modifications
    How: Adjustment of the software to your specific requirements

Parfum/Aroma Ingredient Discovery

Materials Design


Linking scent perception to the chemistry of odorants is elaborate work by human experts until today. This process is associated with numerous laboratory experiments requiring substantial time and material inputs.

The potential of AI-based scent prediction:

Materials modeling methods often neglect scent while focusing on other chemical properties. This leads to negative surprises at the synthesis stage, the necessity to start over the process from the beginning, higher costs and longer product cycles.

The potential of AI-based scent prediction:

  • Check the scent of odorants digitally just by using software
  • Let your fragrance experts concentrate on their craft
    without the necessity of conducting monotonous lab work
  • The potential of the link between scent and molecular structures has not been exploited in parfume and aroma creation yet
  • Conduct digital odor checks before launching production
  • Include odor in the conceptual phase of materials design to cut unnecessary costs and increase resource efficiency
  • The potential of the link between odor and molecular
    structures has not been exploited in materials design yet

Your Advantages at a Glance

  • We provide necessary IT infrastructure and support the aggregation of your in-house data.
  • We define information sources and odor-descriptors in close cooperation with you.
  • We develop customized models for digital scent prediction & design of fragrances and flavors.
  • We are an independent research institute with flexible network structures and in-house capacities.

Why we are the perfect partner

  • State-of-the-art AI performance
    by applying proprietary machine learning methods for scent prediction
  • Experience in application-specific development of AI solutions
    through many successful development projects
  • Many years of experience as a partner in large industrial projects
    from the development of the first idea up to consumer goods

Our research in AI-based software for scent prediction within the Campus of the Senses initiative is funded by the Bavarian Ministry of Economic Affairs, Regional Development and Energy.