In early 2026, a man in Sydney named Christian Conyngham was told that his dog Rosie had a tumor that wouldn't respond to the available treatments. He happened to be an AI consultant. He happened to know how mRNA neoantigen vaccines work in human oncology trials. And he happened to know — most people don't — that the underlying recipe had become open enough that, in principle, anyone with access to a lab could try it.

So he did. He used ChatGPT to talk through the design. He used AlphaFold to look at how the predicted neoantigen peptides would be presented by Rosie's immune system. He took the design to the UNSW RNA Institute and the University of Queensland, where researchers were willing to make the construct. He worked with a willing veterinary team to administer it.

The result

Rosie's tumor shrank by approximately 75 percent. The story spread. Sam Altman publicly said “this should be a company.” Christian himself said he intended to scale it.

What made this remarkable wasn't the science. The science had quietly become possible: open-source structure prediction, open-source neoantigen pipelines, mRNA-LNP manufacturing capacity at university institutes, and veterinary regulators with a conditional-licensure pathway for autologous biologics. What was new was that one family, with the right knowledge, could put the pieces together for one dog.

What we took from it

Most families don't have an AI consultant in the house. Most families have a vet appointment, a PDF, and a weekend of reading low-quality forum posts trying to figure out what to do next. That asymmetry is the gap we're trying to close.

SciRouter Oncology is what happens when you take Rosie's story seriously and turn it into a workspace any family and any vet can use, without anyone needing to be an AI consultant. The same open-source models that helped Rosie are routed underneath the portal — protein structure, neoantigen prediction, drug-mutation matching, RNA-seq, imaging — so a vet sees the same kinds of analyses on their patient that the Sydney team ran on Rosie.

What this is and isn't

This isn't a promise that every dog gets Rosie's outcome. Cancer is hard. Different cancers respond differently. Some don't respond at all. The clinical tools we surface — including the personalized mRNA neoantigen pathway, branded internally as SciRouter mRx — are built under USDA-CVB conditional licensure or university IACUC trial wrappers. They're not magic. They're a real shot when the standard of care has run out, designed and administered by people who know what they're doing.

This is also a workspace for the dogs whose treatment plan is well-understood and just needs to be coordinated well between the family, the GP vet, and the specialty oncologist. Most of our patients fall into this category. The personalized mRNA option is the long tail; the daily quality-of-life support is the spine.

Why we're thanking Rosie

We don't know Rosie. We've never met Christian Conyngham. We have nothing to do with the Sydney lab or the construct that helped Rosie. What we have is a debt of inspiration: their story made it concrete that the modern stack works in canine oncology when someone bothers to assemble it. We're bothering.

If Rosie's story made you want to do something for your own dog, the next step is short: tell us about your dog, share your record with your vet, log how she's feeling each day, and we'll route the rest.

Get early access for your dog How the pipeline works

Sources & further reading

SciRouter Oncology is independent of the Conyngham family, the UNSW RNA Institute, and the University of Queensland. We're telling Rosie's story with respect and gratitude, not as an endorsement on their part. If anyone connected with Rosie's care would like changes to this page, write to stories@oncology.scirouter.ai and we'll act on it the same day.

Rosie's story.

The result

What we took from it

What this is and isn't

Why we're thanking Rosie

Sources & further reading