Michael Polansky Is Training an AI Model on Living Skin Tissue
TechCrunch first reported that Michael Polansky's startup, Outer Biosciences, is training an AI model on living human skin tissue. The company has spent years developing methods to keep living human tissue viable outside the body for over a month, creating a biological substrate that can serve as input for machine learning training. TechCrunch
The intersection of biology and AI is not new, but using living tissue as a training data source sits at an unusual junction. Most efforts in computational biology feed AI models on genomic sequences, protein structures, or digital pathology images. Outer Biosciences is working with the tissue itself, kept alive and functional, which raises both the fidelity and the complexity of whatever signal the model is learning from.
Polansky's path to this point runs through finance, venture capital, and philanthropy rather than a wet-lab career (that is, hands-on laboratory work). He studied applied mathematics and computer science at Harvard, graduating in 2006, then spent three years at Bridgewater Associates before joining Founders Fund as a principal during the firm's early period, when it was run by Peter Thiel, Sean Parker, Luke Nosek, and Ken Howery. TechCrunch
He later ran Sean Parker's family office and helped establish the Parker Institute for Cancer Immunotherapy, where he remains executive director. That institution brings together cancer immunotherapy researchers across multiple centers, giving Polansky sustained exposure to the operational realities of translational medicine — the process of moving discoveries from the lab toward clinical use. The Parker Institute role provides a plausible bridge between his quantitative, finance-oriented background and the life-science domain where Outer Biosciences now operates.
Polansky's personal life has drawn public attention partly through his relationship with Stefani Germanotta, known professionally as Lady Gaga. The two met in late 2019 at one of Sean Parker's birthday parties, at the urging of Germanotta's mother, Cynthia Germanotta. Lady Gaga's most recent world tour began in July of the previous year and ended in April, managed as a large operation across three 747s. She also built Haus Labs, a cosmetics brand based in El Segundo, California, with roughly 70 employees.
The connection to Haus Labs is not incidental to Outer Biosciences' choice of tissue. Skin is the body's largest organ and the cosmetic industry's primary substrate. A model trained on living skin tissue could in principle generate insights relevant to product formulation, irritation testing, and aging research, domains where Haus Labs and similar brands operate. Whether Outer Biosciences is explicitly targeting cosmetic applications is not yet clarified in subsequent reporting, but the proximity of a cosmetics brand within Polansky's immediate orbit provides a natural commercial vector for whatever the model produces.
The technical details of how the AI model is trained, what architecture it uses, and what specific signals it extracts from the living tissue are not yet public. What is known is that Outer Biosciences has invested multi-year effort into the tissue viability problem itself, keeping living human tissue functional outside the body for more than thirty days. That is a nontrivial bioengineering achievement. Standard tissue culture protocols vary by tissue type, but maintaining viability beyond a month typically requires perfusion systems (mechanical circulation of fluids through tissue), controlled nutrient delivery, and careful management of oxygenation and waste removal.
The broader context here is that the AI-biology convergence has been accelerating on multiple fronts. AlphaFold solved protein structure prediction. Large language models have been adapted for genomic sequence analysis. Diffusion models generate novel protein scaffolds. What Outer Biosciences appears to be doing is different in kind: not using AI to model biological data, but using living biological material as a direct input to the training process. Whether the signal from living tissue is sufficiently structured and reproducible to improve model performance over conventional digital-data approaches is an open empirical question, and one that will likely determine whether the approach scales beyond the laboratory.
In my view, the most interesting aspect is not the AI model itself but the tissue platform. Keeping living human skin viable for over a month creates possibilities that extend well beyond machine learning: pharmacological testing without animal models, personalized medicine assays, and wound-healing research all benefit from long-duration tissue kept alive outside the body. If Outer Biosciences has genuinely solved the viability problem in a reproducible, scalable way, that platform may matter more than any single model trained on it.
Polansky's background is unusual for a founder pursuing this type of work. He is not a bench scientist. He is a mathematician and operator who has spent years at the intersection of capital, philanthropy, and life sciences. The Parker Institute experience in particular suggests he understands how to coordinate distributed research efforts and navigate the institutional complexities of translational medicine. Whether that skill set translates to building a commercial tissue-platform and AI company is a separate question, but the combination of quantitative rigor and institutional fluency is not a common founder profile in this space.
For now, the facts are these: a founder with deep ties to both the technology and life-science ecosystems is building a company that pairs long-duration living-tissue culture with AI training. The specifics of the model, the funding, and the go-to-market strategy are not yet reported. What is visible is the ambition: to bring living biology into the AI training pipeline in a way that few others have attempted.
