Michael Polansky’s biotech venture Outer Bio is training predictive AI models on data collected from living human skin, using a proprietary platform called Yuna that can keep full-thickness skin tissue observable for four weeks, a duration the company says sets it apart from the limitations of artificial skin and lab-grown organoids that have long constrained dermatological research. The Outer Bio skin AI programme, founded in 2020 and based in Massachusetts, remained deliberately out of public view until Lady Gaga, whose legal name is Stefani Germanotta, announced it on X on 21 August.
What Outer Bio’s skin AI platform actually does
The core technical challenge in studying human skin is one of longevity. Living skin removed from the body deteriorates quickly, and while artificial substitutes exist, they do not replicate the biological complexity of actual human tissue. Yuna, developed in 2022, addresses this by combining a 3D-printed support structure with a nutrient-delivery and waste-removal system that keeps donated skin samples viable across a month-long experimental window. That window matters: it allows the team to remove samples at multiple points, building longitudinal datasets rather than single-point snapshots.
Outer Bio sources its skin from a network of hospitals, biobanks, and medical research centres, working with tissue that is typically left over from surgery or elective medical procedures. A single donated piece can be processed in-house into dozens of individual samples, each prepped for experimentation and placed into custom-made well plates. The company’s stated research areas include wound healing, pigmentation, chronic inflammation, ageing, and skin barrier function.
The data generated from those experiments then feeds AI models designed to predict which compounds could have bioactive or toxic effects on human skin. Results from each experiment cycle back into those models to improve their predictive accuracy over time. Polansky has been explicit about where the company’s centre of gravity lies. “We think of ourselves as a data company first and foremost,” he told Business Insider. “We’re not generative AI. We’re predictive AI.”
That distinction is worth holding onto. Generative models produce outputs (text, images, protein sequences) from learned patterns. Predictive models, by contrast, are trained to classify or score known inputs: here, to assess the likely bioactivity or toxicity of candidate compounds before they are ever tested on tissue. The value of the approach depends entirely on the quality and volume of the underlying biological data, which is why Polansky describes data collection as the company’s “north star.”
Commercial ambitions and the Haus Labs partnership
Outer Bio’s model is not purely a research operation. The company’s stated goal is to discover proprietary bioactive ingredients that third-party manufacturers can incorporate into finished products, from pharmaceutical formulations to consumer beauty lines. “If you imagine products in Ulta or CVS, they have all sorts of active ingredients in them,” Polansky said. “We would have proprietary active ingredients in products that were sold by other people.”
The earliest commercial partnership is with Haus Labs, the cosmetics brand associated with Lady Gaga, who sits on Outer Bio’s board of directors. Polansky also sits on the Haus Labs board, making the arrangement a genuinely bilateral relationship. He has been careful, however, to frame it as one among several. “We are working with other beauty companies that do totally different things with us than what Haus Labs does,” he said. “Haus Labs’ ambitions around innovation are very unique to them.”
The deliberate silence around Outer Bio until now was a strategic decision rather than an oversight. “We made the decision internally that we weren’t going to talk about what we were doing until the data we generated was publication worthy,” Polansky said. That language (“publication worthy”) signals an intention to submit findings to peer-reviewed journals, which would bring the company’s claims under external scientific scrutiny for the first time.
On the broader rationale, Polansky pointed to what he sees as a pace problem in life sciences. “We’re moving into an era where the amount of innovation that’s needed in biology and life sciences is going too slowly,” he said. “I think that the only way that goes faster is that we have better data.” The company’s next stated milestone is reaching the threshold of peer-reviewable data quality, the bar Polansky has publicly set for the next phase of disclosure.


