In late 2024, Sydney tech entrepreneur Paul Conyngham was told his rescue dog, Rosie, had months to live. She was battling a tennis ball-sized mast cell tumor that had resisted surgery and chemotherapy. For most, this is where the story ends in heartbreak. But Conyngham, a data analyst with 17 years of experience in machine learning, and zero background in biology, decided to treat the cancer not as a medical death sentence, but as a data problem.
What followed is being called a "world first": a personalized mRNA cancer vaccine designed by a non-scientist using a stack of off-the-shelf AI tools.
The core of the "Rosie Protocol" was turning biological tissue into actionable code. Conyngham paid to have Rosie’s healthy DNA and her tumor DNA sequenced at the University of New South Wales (UNSW). This resulted in a massive 300GB raw sequencing file.
In traditional medicine, interpreting this data to find "neoantigens", which are the specific mutations the immune system can target, takes a team of bioinformaticians months. Conyngham did it by leveraging a specific AI tech stack:
The result? That 300GB of data was compressed into a half-page mRNA formula.
Conyngham took his AI-generated "recipe" to Professor Páll Thordarson at the UNSW RNA Institute. While the AI provided the blueprint, the human experts provided the "foundry." They synthesized the mRNA and packaged it into lipid nanoparticles, the same tech used in COVID-19 vaccines.
By December 2025, Rosie received her first injection. By January 2026, the primary tumor had shrunk by nearly 50%. Within weeks, a dog that could barely walk was jumping fences to chase rabbits.
The most provocative part of Rosie’s story isn't just the science, it’s the bureaucracy. Conyngham spent months fighting for "compassionate use" and navigating an ethics approval process that moved at a snail’s pace compared to his AI. While he secured approval for Rosie by mid-2025, the broader "ethics framework" for this kind of citizen led medicine is still catching up.
As we move through 2026, we are seeing a massive Regulatory Gap. The technology to sequence, model, and print a personalized cure now exists in the hands of talented individuals. However, our legal systems are still built for the era of mass-produced, one-size-fits-all blockbusters.
Paul Conyngham is now looking to turn this workflow into a scalable model. This isn't just about saving one dog; it’s about the democratization of medicine. If a data engineer can design a vaccine for a dog in his home office, how long until we see "Print-on-Demand" cancer clinics for humans?
The Rosie Protocol proves that the future of medicine might not be found in a lab, but in a sequence of prompts.
Sources:
1. UNSW Newsroom: "Meet the man who designed a cancer vaccine for his dog" (Published March 17, 2026). This is the definitive source for the collaboration between Conyngham and the University of New South Wales.
2. The Jerusalem Post: "ChatGPT helps owner craft cancer vaccine for dog Rosie" (Published March 16, 2026). This report provides details on the 50-75% tumor reduction and the timeline of the treatment.
3. Bioplatforms Australia: "AI and genomics combine to create personalised cancer vaccine for dog" (Published March 19, 2026). This source details the specific facilities used for the sequencing and analysis.
4. NBC News: Interview with Gadi Schwartz (March 2026).
5. CBS News 24/7: Joint interview with Paul Conyngham and Professor Páll Thordarson (March 18, 2026).