In March, a Sydney tech entrepreneur with no biology training became a global story. Paul Conyngham had used AI tools to help design a personalized mRNA cancer vaccine for his rescue dog Rosie, who had aggressive mast cell cancer and a prognosis measured in months. Several of her tumors shrank. OpenAI chief executive Sam Altman called it an amazing story to his millions of followers.
Five months later, the company built on her case launched with a $4 million seed round led by Founders Fund. The announcement carries a detail that most of the viral coverage never reached: Rosie's cancer came back.
What Actually Happened in Rosie's Treatment
The version that circulated online was compressed a great deal.
Rosie was an eight-year-old Staffordshire bull terrier cross diagnosed with mast cell cancer, the most common canine skin cancer. Surgery and chemotherapy slowed the disease without stopping it.
Conyngham, who has 17 years of machine learning experience, paid about $3,000 to have Rosie's tumor DNA and healthy DNA sequenced at the University of New South Wales. He used AI to sift through thousands of mutations for neoantigens—mutated proteins that appear only on cancer cells—and used protein-structure prediction to guess which would be visible to her immune system.
He did not make a vaccine himself. Researchers at UNSW reviewed the data and built the mRNA construct in a university lab, with Pall Thordarson, director of the UNSW RNA Institute, leading that work. According to Reason, Conyngham used ChatGPT and AlphaFold to identify mutated tumor proteins and then Grok to design the final construct, and obtaining Australian ethics approval took him three months and a 100-page document.
Rosie received her first injection in December 2025 at a veterinary facility in Queensland, with a booster the following month. The Y Combinator listing for the resulting company describes it as what is believed to be the first fully computationally designed mRNA cancer vaccine administered to a dog, and notes that several tumors shrank significantly.
Conyngham has consistently declined to call it a cure.
Remission, Then Relapse
The company's launch announcement states it directly: Rosie entered remission, but her cancer returned following a later surgery. The release is titled around her legacy and writes about her in the past tense, saying her story may have ended while the work it produced continues. The company has not publicly stated what ultimately happened to her or when.
That ambiguity matters less than what the case can and cannot show. There was no control group, no comparison animal, and no way to separate the vaccine's effect from surgery, chemotherapy, immunotherapy, or the natural variability of an individual tumor.
Key clinical details, including the grade of Rosie's tumor, have not been made public. Grade is one of the strongest predictors of outcome in canine mast cell disease, and without it, the response is difficult to interpret.
As recently as June, Australian broadcaster ABC reported on the case as a success story while scientists involved argued that regulatory hurdles were slowing similar work in humans.
From One Dog to a Venture-Backed Platform
Gamgee, founded by Conyngham, joined Y Combinator's Summer 2026 batch and launched publicly on Aug. 17. Alongside Founders Fund, the round drew investment from Celine Halioua of canine longevity company Loyal, Steve Eidelman of Modern Animal, Dropbox co-founder Arash Ferdowsi, and Gokul Rajaram.
The model industrializes what Conyngham did by hand. The company sequences each dog's tumor and matched healthy tissue, identifies tumor-specific neoantigens, and designs a personalized mRNA vaccine intended to train the immune system to attack those targets.
Clinical studies are running in Australia in collaboration with UNSW, the University of Queensland, and the Garvan Institute of Medical Research. Enrollment is currently limited to eastern Australia; participating dogs receive the investigational vaccine at no cost, and eligibility is determined by the study protocol and participating veterinary oncologists.
The company says it is working with regulators on a pathway through the U.S. Department of Agriculture for a patient-specific veterinary biologic. That is the correct agency: animal vaccines fall under the USDA's Center for Veterinary, not the FDA.
Why the Caution from Oncologists Is Not Pedantry
A cancer researcher writing in The Conversation in March described the treatment as a one-off, made just for Rosie, and urged readers not to mistake a single animal's response for evidence that the approach works.
The underlying science is not fringe. Personalized neoantigen mRNA vaccines have been in human development for over a decade, and Moderna and Merck's melanoma candidate has produced randomized data. Rosie's case did not invent that field. What it demonstrated, as industry observers noted, is that the design steps have become accessible enough for a determined outsider working with academic collaborators.
Whether the approach helps dogs in general is what the Australian studies are meant to answer, and they have not yet answered it. Pet owners facing a cancer diagnosis should work through a board-certified veterinary oncologist rather than treating a viral case as a treatment plan. Nothing here is approved, and no outcome data across multiple animals have been published.
Key Questions Answered
What happened to Rosie? Her tumors shrank after the personalized mRNA vaccine, and she entered remission. The company says her cancer returned following a later surgery and writes about her in the past tense.
Did AI create the vaccine? AI tools helped plan the work, filter mutations, and model protein structures, and reporting indicates Grok was used for the final construct. The physical vaccine was made by researchers at UNSW.
Does Rosie's case prove the approach works? No. It is one animal with no control group, alongside surgery, chemotherapy, and immunotherapy. Her tumor grade was never made public, which matters for interpreting the response.
What is Gamgee? The startup Conyngham founded. It launched Aug. 17, 2026, with a $4 million seed round led by Founders Fund, and develops personalized mRNA cancer vaccines for dogs.
Can owners get this for their dog? Clinical studies are running in Australia, with enrollment currently limited to eastern Australia. The company says it is pursuing a USDA pathway for a patient-specific veterinary biologic.
Is the underlying science legitimate? Personalized neoantigen mRNA vaccines have been in human clinical development for years, with randomized data in melanoma. Whether this specific canine approach works is unproven.
