Polyphron's $20M seed bet: lab-grown heart tissue as AI's drug-testing ground

Polyphron, the New York company that grows living human heart tissue in an automated foundry and uses it to teach AI what a drug will do next, has raised $20M in seed funding.

Quiet Capital led the round, with participation from Haystack, Gradient, and Compound, plus a group of angel investors, according to cofounder and CEO Matthew Osman.

The premise is straightforward: AI can now generate more drug candidates than scientists can test. Animal studies and clinical trials move slowly and carry high costs, while lab shortcuts like cell assays often miss how real human tissue behaves. Polyphron manufactures donor-defined human tissues in its foundry, perturbs them in controlled ways, measures how they change over time, and uses those longitudinal trajectories to train models of tissue behavior.

Cardiac tissue is already in production across multiple donor lines. Liver is being onboarded next, extending the platform into drug metabolism and liver toxicity, with additional tissues to follow. Today that gives biotech and pharma companies a way to test drug hypotheses directly against living human tissue, while also supplying experimental datasets and benchmarks to AI labs.

The company argues its differentiator is the closed loop. "Most approaches have only one side of that loop: either models trained largely on existing datasets with limited ability to generate new ground truth, or experimental systems that produce data without turning every experiment into a reusable predictive model," Osman said. Each new donor, tissue, and perturbation is meant to improve the simulator.

Osman traces the company's origin to a health scare. In late 2023, an MRI turned up a mass on his pancreas. It proved benign, but the months of not knowing made him think about how much of what happens inside the body medicine still cannot predict. He began comparing notes with CSO Fabio Boniolo, who had spent years on tissue engineering at Dana-Farber and the Broad Institute.

The team also includes cofounder and Chief AI Scientist Vinh Q. Tran, a post-training and self-improvement researcher for Gemini at Google DeepMind, and COO George Pilitsis, who scaled a data product at Ginkgo Bioworks from nothing to millions in under two years.

Over the next six months, Osman said the company is focused on scaling the core infrastructure behind the platform: expanding the training dataset across cardiac and liver tissue, building the automation required to run experiments with greater throughput and consistency, and growing the donor panel to more than 20 lines. Donor diversity matters, he said, because it moves the models beyond a single biological response toward variation across a broader patient population. He added that the round funds the next phase of the platform without a return to market on a compressed timeline.