Many of our customers are using the Citrine Platform to remove so-called “Forever chemicals”, PFAS, from a variety of products. Others are replacing petroleum-based ingredients with naturally-derived alternatives. Still more are assessing their inventory of raw materials to ensure that they can rapidly transition, should a supply shock occur. In each case, Citrine’s AI is able to help because it is chemically-aware and can be trained to deeply understand a domain by our customer’s experts.
Chemically-aware Platform
For over a decade, Citrine has been helping companies to use AI in data scarce, knowledge rich environments to solve some of their toughest product challenges.
Automated Enhanced Chemical Data
Chemical structure and formulas can be used to automatically generate a set of ingredient features, such as molecular weight and number of hydrogen bonds, which are then used by the AI to make predictions about the final properties of a product.
Molecular structure entered generates many different chemical “features” that the AI then uses in predictions.
Raw Material Type Labeling
Product experts can label types of raw materials and input their properties. For instance, surfactants or grease proofing agents can be grouped, enabling the AI to learn not just how a particular additive affects final properties, but also how an ingredient from this group is working.
Capturing Materials Histories
An experimental result is not useful unless you know all of the raw materials, conditions, and processing parameters that occurred to make the test sample. The Citrine Platform captures these as a Material History and enforces rigorous record keeping and retention of high-quality data. The data model used to store the data is graphical – this means that unlike a standard SQL database – as your project progresses and you decide to measure different parameters, these are easy to add. The material history structure enables experiments to be shared across colleagues and teams, so they can learn from another and avoid conducting redundant tests.
No-code Platform Anyone Can Use
The Citrine Platform has been built to be used by product experts, not data scientists. Product developers of all backgrounds and experience are able to easily use the platform to find data and train Citrine’s AI on their products. They can tell the platform which raw materials or types of raw materials are affecting which properties in the final product and this is then used by the AI. This is vital not just to ensure a smooth adoption of the platform, but also to capture the expert knowledge of experienced team members. This then accelerates the work of the whole team and ensures continuity when veteran team members retire.
Summary
Together, this set of information enables the Citrine Platform to understand what part a particular raw material is performing in your formulation and suggest sensible alternatives. Our customers have successfully reformulated without problem ingredients in 20% of the time it typically takes.
Want to learn more about how the Citrine Platform could work for your team? Book a demo today!