An engineer just used ChatGPT and AlphaFold to make a cancer vaccine for his dog

When vets told Sydney tech entrepreneur Paul Conyngham that his dog Rosie had only months to live after surgery and chemotherapy failed, he refused to accept the death sentence. Instead, he fired up ChatGPT.

Conyngham has no medical training, but he is an electrical and computing engineer, co-founder of Core Intelligence Technologies, and former director at Australia's Data Science and AI Association. Desperate to save his pet, he turned to artificial intelligence for answers.

A $3,000 sequence and an assist from AlphaFold

ChatGPT suggested looking into immunotherapy and directed him to the Ramaciotti Centre for Genomics at the University of New South Wales (UNSW). Conyngham paid UNSW $3,000 to get Rosie’s tumor sequenced, then dove headfirst into the raw genomic data.

He used ChatGPT to draft an analysis plan and deployed Google DeepMind’s AlphaFold to identify mutated proteins that could serve as viable drug targets. Conyngham actually found an existing immunotherapy candidate that fit the profile, but the pharmaceutical manufacturer flatly refused to provide it.

From rejected plea to a custom mRNA jab

Rather than giving up, Conyngham took his findings to Pall Thordarson, director of UNSW’s RNA Institute. The nanomedicine pioneer was stunned by the quality of the data and agreed to step in.

Using Conyngham’s targets, Thordarson’s team formulated a bespoke mRNA cancer vaccine in less than two months. It marks the first time a personalized cancer vaccine has been designed specifically for a canine. After three months of navigating regulatory approvals and a 10-hour drive for the first dose, Rosie received her shot in December and a booster in February.

The response was dramatic: most of Rosie's tumors shrank significantly. Six weeks after treatment, Rosie was jumping fences to chase rabbits at the park. Both Conyngham and Thordarson caution that this is not an outright cure since some tumors did not respond, but it vastly improved her quality of life. As OthersideAI CEO Matt Shumer posted on X, the world is going to get weird very quickly, and stories sounding increasingly insane will soon become common.

Why it matters

We think Rosie’s survival story is a genuine watershed moment showing how consumer AI can break down institutional medical gatekeeping.

Historically, developing a personalized mRNA therapeutic was the exclusive domain of deep-pocketed biotech giants, requiring millions of dollars and multi-year trials. Conyngham compressed the earliest and most complex phases of target identification into weeks using a $3,000 sequencing run, ChatGPT, and AlphaFold.

This doesn't mean patients should start DIY-ing oncology treatments in their living rooms. Conyngham is a skilled engineer capable of debugging complex datasets, and the actual chemical synthesis still demanded top-tier university facilities. Relying on LLMs without medical oversight remains extremely dangerous.

Still, the broader industry implications are hard to ignore. It proves that combining AI with modern mRNA platforms can democratize personalized medicine. If a software founder and a university lab can spin up a targeted cancer vaccine for a pet in two months, the timeline for human precision medicine is about to move much faster than anyone anticipated.