Our CEO and co-founder Noel Hollingsworth wrote a feature for COSSMA on the real bottleneck holding back AI in cosmetics formulation: not the algorithms, but the data underneath them.
His argument: most mature cosmetics R&D organizations are sitting on "dark data," years of formulation results scattered across spreadsheets, shared folders, and the heads of scientists who eventually retire. Predictive models need a starting point of clean, structured history, and that starting point rarely exists. Noel lays out four requirements that separate data an AI model can actually learn from, from data it can't: completeness (including the failures, not just the wins), consistency (the same attribute recorded the same way, every time, everywhere), context (a result without its conditions is close to meaningless), and linkage (formulation, scale-up, QC, and regulatory data need to connect, not live in separate systems).
The piece also covers how Clariant rebuilt its data foundation: replacing a legacy electronic lab notebook with Uncountable's integrated ELN and LIMS, onboarding more than 800 scientists across global sites in under ten months, and reaching over 1,000 users across 35 facilities by the end of 2024. As Cesar Chocou, Head of IT at Clariant, put it: "AI can make us more productive. But the real difference is when we can use our own data to accelerate our proprietary know-how."
.png)
.png)
.png)
