This Guy Used ChatGPT and AlphaFold to Create a Custom mRNA Cancer Vaccine for His Dying Dog

We all talk or hear about how artificial intelligence (AI) can change human life for the better, and how it is here to help us be more productive, efficient, and work smarter. However, we only see and hear those things on paper and hardly see that to be true in real life, but that changed when Paul Conyngham, a Sydney tech entrepreneur, used ChatGPT and Google DeepMind's AlphaFold to brainstorm and analyze Rosie, his dog's mast cell tumor, to engineer a personalized mRNA cancer vaccine in collaboration with The University of New South Wales (UNSW) scientists Martin Smith and Pall Thordarson.

This pioneering effort shrank one of Rosie's major tumors roughly by half, improving her mobility, and also became the world's first custom cancer vaccine designed for a dog. The remarkable story is now sending ripples through the medical and tech communities, highlighting how generative AI can accelerate personalized, time-sensitive healthcare.

Quick backstory

Rosie, an eight-year-old Staffordshire Bull Terrier-Shar Pei cross, was diagnosed with aggressive mast cell cancer in 2024. The disease progressed despite Paul spending thousands of dollars on conventional treatments like surgery and chemotherapy, and veterinarians estimated she had only one to six months to live. Unwilling to give up, Conyngham, who has 17 years of experience in machine learning and data analysis but no formal biology background, turned to ChatGPT. Paul used the generative engine as a research assistant to formulate a complex action plan to tackle the cancer at a genomic level.

Following guidance given by ChatGPT, Conyngham had paid $3,000 to have Rosie's healthy and cancerous DNA sequenced at the UNSW Ramaciotti Centre for Genomics. He compared the two sequences to pinpoint the exact mutations driving the cancer by processing gigabytes of genetic data. He then used AlphaFold, an advanced AI system that predicts 3D protein structures, to model the mutated proteins and identify possible therapeutic targets. Conyngham condensed this complex data into a half-page formula and collaborated with the UNSW RNA Institute, where scientists successfully manufactured the custom mRNA vaccine in under two months.

Key milestones of the AI-Driven vaccine process:

• AI-Assisted Strategy: ChatGPT worked as an intelligent planning layer, helping Conyngham navigate unfamiliar biomedical literature and structure a highly targeted genomic strategy.
• Genomic Sequencing: Researchers accurately identified the specific genetic mutations causing the disease by sequencing both healthy blood cells and tumor tissue.
• Targeting with AlphaFold: Google DeepMind's AlphaFold was crucial for visualizing mutated protein structures and identifying vulnerabilities that the immune system could attack.
• Custom mRNA Synthesis & Approval: The UNSW RNA Institute turned the AI-generated formula into a bespoke mRNA vaccine, while Conyngham spent three months securing strict ethical approval for the veterinary trial.
• Remarkable Results: Administered at the University of Queensland in December 2025, the vaccine shrank a tennis-ball-sized tumor on Rosie's leg by half within weeks, drastically improving her energy and mobility.

Conyngham cautions that this is a life-extending management strategy rather than a definitive cure; another tumor did not respond, and he is already designing a second vaccine. However, the implications are deep. Rosie's case has proved that AI-assisted, personalized mRNA therapies can be developed quickly and cost-effectively.

Key takeaway:

This case has hinted at how AI may compress parts of the personalized-medicine workflow without replacing the institutions, validation steps, and specialist judgment that make such work real. That idea fits a wider trend.

• The U.S. National Cancer Institute says personalized mRNA cancer vaccines are being explored in multiple cancer studies, but the FDA has yet to approve any.
• NHS England likewise describes personalized mRNA cancer vaccines as immunotherapy treatments currently being tested in clinical trials.
• In January 2026, Merck and Moderna reported five-year follow-up data showing that their investigational individualized neoantigen therapy plus Keytruda reduced the risk of recurrence or death by 49% in high-risk melanoma patients compared with Keytruda alone.

That is why Rosie's story should be treated neither as science fiction nor as settled science. It is best understood as an early, emotionally powerful example of AI-assisted biomedical problem-solving that promises deep collaboration and is still far from proof that chatbots can cure cancer.

For now, the most factual conclusion is also the most human one that AI may have helped buy Rosie more time and a better quality of life, while showing researchers just how much faster personalized medicine workflows may become.