The 'ChatGPT cured my dog' story is now a Y Combinator-backed startup

Paul Conyngham, the Australian entrepreneur who went viral earlier this year for claiming he used ChatGPT and other AI tools to design a personalized mRNA cancer vaccine for his dog Rosie, has now launched a startup called Gamgee. The company promises personalized mRNA cancer vaccines for dogs, with ambitions stretching far beyond pets to humans and a range of diseases. Y Combinator is backing the venture, and Gamgee says it’s recruiting for its first clinical trials in Australia while accepting cases globally through a waitlist.

Rosie, a Staffordshire bull terrier-Shar Pei mix, is the emotional anchor of Gamgee’s pitch. She appears across the startup’s website, social media, and an AI-generated anime launch video. Conyngham, who explicitly states on the site that he is “not a biologist,” credits the intervention with adding years to Rosie’s life and claims several of her tumors shrank. Gamgee’s veterinary-facing materials tout a roughly 75 percent tumor reduction in its “first documented case” — which is, of course, Rosie — while promising a finished vaccine in about four weeks and less than 20 minutes of a veterinarian’s time.

But the science is murkier than the narrative suggests. My reporting for The Verge earlier this year found that the exact role AI played was unclear, and researchers said the broad claims gave the technology too much credit while minimizing the human contributions. More critically, Rosie received the mRNA vaccine alongside another therapy, making it impossible to attribute her improvement to the vaccine alone. That fundamental confound — no control, no isolation of variables — remains the startup’s weakest structural beam.

Gamgee says it’s collaborating with the University of Queensland and institutes at the University of New South Wales. Costs aren’t disclosed. The leap from a single, confounded canine case to a platform promising personalized treatments for humans is enormous, and the startup’s own materials concede the timeline estimates are based on one case and may vary. The question isn’t whether AI can assist in drug discovery — it demonstrably can. The question is whether a compelling story and a sympathetic dog can carry a company through the rigorous evidence threshold that veterinary medicine, let alone human medicine, demands. Y Combinator’s check suggests someone thinks the answer is yes.