SUPPLY CHAIN MANAGEMENT
Having a plan B
AI is good at spinning up many different options and ranking them. A speciality chemical company used the Citrine platform to model the technical properties of many different formulations, using different chemical approaches to achieve the same goal. The formulations that successfully met the technical specification when tested were then ranked on the cost of ingredients and a final formulation chosen.
However, this left the company with a library of other successful formulations, that met the technical specs available, albeit for a slightly higher cost. This reserve of alternative formulations offers flexibility and security, enabling the company to adapt swiftly, should the need arise, to change their approach due to supply chain disruptions or other unforeseen circumstances. By having these backup formulations readily available, the company can ensure continuous production and maintain their competitive edge in the market.
In another example, a large polymer producer with business units in the EU and US took a different approach. They had two ingredient catalogues from two different business units. They wanted to find out if there was a way to use AI to optimise the use of these catalogues to make better, high-performing materials, cheaper materials and to understand if any ingredients were irreplaceable. Supply chain disruption risk could then be mitigated for the irreplaceable ingredients.
Citrine and this customer worked together to get the data from both catalogues onboarded into the Citrine platform. The customer then built an AI model to model the target technical properties of their polymer, combining the two catalogues. It was thus able to generate formulations optimising the key target properties while maintaining low costs( Figure 1).
This resulted in a polymer formulation using cheaper ingredients, a bank of recipes that could be used in the future, and a better understanding of which raw materials were essential to the performance of the polymer and which could be easily replaced. The whole exercise took just under three months, where before it would probably have taken years, given the complexity of the ingredient catalogues.
Local sourcing
AI helps companies create consistent quality, using local ingredients. One way to reduce the likelihood of supply disruptions is to turn to local suppliers. However, this can pose challenges in terms of maintaining quality and consistency.
A global personal care company, known for a particular well established, household name skin care product, wanted to reduce its ingredient costs and believed it could do so using local ingredients. However, it also had to protect its brand integrity at all costs.
To address this, the company employed the Citrine platform to model the rheological properties of a suite of skin care products. Similarly to the previous example, all of the data was uploaded, and all of domain knowledge the company had was integrated into AI models. The customer also took advantage of the chemical featurisation done within the platform( Figure 2).
The AI model was utilised to generate adjusted formulations tailored for different regional manufacturing hubs, using locally available ingredients. This approach had the added benefit of reducing raw material costs by 20 %.
The increase in profit margins enabled the company to enter previously inaccessible markets, offering products at a lower price point while remaining profitable. This strategy not only safeguarded the brand’ s quality and consistency but also expanded its global market reach. It was all done in eight months as opposed to the years it would have taken without AI
Speed
AI-driven experimentation reduces the number of experiments needed to hit target properties by 50-80 % compared to traditional approaches. Whether it is a supply chain disruption, a shift in customer sentiment or new regulations, there are numerous scenarios where a substitute ingredient may be required, sometimes very rapidly. The Citrine platform features patented technology that can‘ fingerprint’ ingredients and their roles in a formulation, allowing it to suggest alternative formulations without those specific ingredients.
A speciality chemical company was producing high-performance polymers and saw growing consumer demand for greener, more sustainable plastics. This company had plenty of suitable ingredients available but was far from certain that these would enable its products to achieve the performance levels required.
Screening all of the possible ingredients using traditional methods would have taken years. Instead, Citrine helped the company to programmatically combine all the monomers it had to generate 2,500 new, hypothetical polymers in a workflow that took five months.
The company then incorporated its domain knowledge into the model to predict the thermal properties of these polymers. The Citrine Platform was able to generate a list of the ten most suitable novel polymers( Figure 3). Using traditional experimentation, the customer validated the results. It was also able to take the ML algorithm it had developed for the project and scale the learnings to other projects.
Summary
AI is proving to be a game-changer for speciality chemical companies, being successfully applied across a wide array of applications. By accelerating development and offering numerous alternative options, AI empowers companies to be agile and adapt to change without compromising on product optimisation. ●
J j
Hannah Melia
HEAD OF MARKETING
CITRINE INFORMATICS h. melia @ citrine. io www. citrine. io
MAY / JUN 2025 SPECCHEMONLINE. COM
65