100,000 years in 10 months: Debut’s AI tool advances biotech beauty ingredient development

Biotech firm Debut is researching ingredients for personal care products using its latest generative AI platform, BeautyORB. The tool has already formulated ingredients to address inflammaging and epidermal barrier repair issues and is set to calculate two biotech ingredients annually.

The BeautyORB platform uses over 10,000 molecules to understand the effect of different molecular structures on skin cells. Data engineers train the platform to recognize molecular patterns, which enables Debut to find the right molecules to activate specific gene responses. This genomics-based approach can be adaptable depending on the company’s desired ingredient formulation.

Previous research into ingredient information required researchers to screen thousands of molecules in a lab and make improving modifications to each molecule individually — which Debut calls a costly and time-consuming process. 

Personal Care Insights speaks to Joshua Britton, founder and CEO of Debut, about BeautyORB’s impact on traditional ingredient formulation in the personal care industry, challenges in AI-generated ingredients, and genomics-based AI. 

How does this AI platform stand out from competitors finding ingredients for skin care?

Traditional ingredients can take 100,000 years to find but with AI can take 10 months.Britton: Currently, ingredient providers screen large libraries of natural extracts to see if something in the extract does the job — roughly accounting for 0.0002% of potential ingredients. Debut screens a potentially limitless quantity of ingredients, upward of 50 billion, to discover the next big ingredient at incredible speed. What would take 100,000 years using traditional approaches can be done in 10 months or so computationally. 

The second reason our AI platform stands out is our vertical nature. The fact that Debut is vertically integrated allows our novel ingredients to pass through the right pre-clinical, clinical, and formulation screens to ensure they come to fruition and launch on the market.

For us, we can just walk next door to our biomanufacturing teams and say, “please make this for us, the clinicals are amazing and this is a top-performing ingredient no one else has.” Our engineering team can deliver very quickly, in 12 to 18 months. 

What makes the platform a “genomics-based” AI tool? 

Britton: We look at all the genes regulated in the genome within a specific cell. Much research examines a small specific gene panel and what happens with those particular genes. For example, they might look at anti-inflammatories and then study ten genes they know are involved in the anti-inflammatory pathway.

We take a reverse approach: we want to let the data tell us what we are looking for. So, we have technology that allows us to not only see what all 30,000 genes are doing but also map those to every known pathway.

We get immediate feedback that says this is anti-inflammatory — here are the 50 genes, for example, that are in that pathway, and here is what is happening with each of those genes and their downstream effects.

What are the challenges in creating an AI platform for the beauty market? 

One challenge of AI created ingredients is having a reliable data set. Britton: The key challenges are generating robust, reproducible data. Otherwise, it’s a case of bad data in and bad data out. Many publicly available data on the pharma and beauty industry are non-reproducible, so models based on that data are inaccurate. 

Debut went back to the drawing board to generate our own data sets, assays, and validation steps. We wanted to ensure that we had the most reliable data set in the beauty industry.

What partnerships might Debut enter into with the AI tool?

Britton: We look to work with large beauty companies on specific programs and have started to examine working with pharmaceutical companies focused on over-the-counter and traditional disease states.