Proven aggregates data from 8 million consumer reviews, 100,000 skincare products, 20,000 ingredients and 4,000 academic journals to decipher what types of ingredients work well for specific individuals.
Every year, consumers spend billions on skincare products that promise to reduce lines instantly, fade brown spots, improve firmness and elasticity and more. Yet despite the wealth of products available today, consumers continue to be frustrated by manufacturers’ inflated claims of smooth, silky, younger-looking skin.
Now beauty start-up Proven — launched in November by Harvard Business School grad Ming Zhao and computational physicist Amy Yuan — is aiming to curb all that frustration, rejecting the traditional one-size-fits-all miracle remedy and instead relying on artificial intelligence to develop data-driven skincare routines that are completely personalized and sent straight to your doorstep.
Proven’s formulations are made in-house and based on an online questionnaire that asks about skin concerns, lifestyle and environment. Written to mimic a dermatologist visit, the skin quiz asks about visible genetic background, such as ethnicity and skin tone, to help determine which ingredients to use.
Using machine learning and natural-language processing, Proven’s AI engine pores through the data to decipher what ingredients will work best for specific individuals and environments. Their Skin Genome Project, which aggregates data from 8 million consumer reviews, 100,000 skincare products, 20,000 ingredients and 4,000 academic journals, won a 2018 MIT AI Idol award.
Zhao and Yuan then work with dermatology experts to fill in knowledge gaps as they figure out the actual product formulations based on ingredients suggested by the data. For example, Zhao said, certain acids work particularly well for hyperpigmentation in African-American and Asian skin that has high melanin.