Three months after Rosie the dog passed away, her medical records developed into an AI medical company.

Yesterday, the news of "successful cancer vaccine" immediately caused a stir on Wall Street.

Moderna's stock price soared 177% in a single day, hitting one of the most dramatic one-day gains in recent years.

Why are capital players so excited?

Because this new cancer treatment method sounds extremely "sci-fi": first read the gene mutations in the patient's tumor, then customize an mRNA vaccine for the patient.

When the entire industry is still immersed in this carnival, another "miniature version" mRNA cancer vaccine story has also reached a new milestone.

A few months ago, Paul S. Conyngham, an AI entrepreneur with no medical background, tried to use AI tools such as ChatGPT, Gemini and Grok to "hand-make" a personalized mRNA vaccine for his cancer-stricken dog Rosie, and this story quickly went viral on the Internet.

Just today, Paul announced that his newly founded company Gamgee has secured $4 million in seed funding, led by Founders Fund.

What this company aims to do is to turn the solution previously tested for Rosie into a replicable process, to provide personalized mRNA cancer vaccines for more dogs suffering from cancer.

Unfortunately, Rosie did not live long enough to see the next chapter of this story.

The first sentence Paul wrote when announcing the financing is: "I lost my dog because we failed to produce her next mRNA cancer vaccine fast enough."

Rosie's Story

In March this year, Rosie's story went viral on English social media: a tech entrepreneur with no biology training background used AI tools to design a personalized mRNA cancer vaccine for his terminally ill dog, and several of the dog's tumors shrank.

OpenAI CEO Sam Altman shared this story to his millions of followers.

Rosie is a Staffordshire Bull Terrier mix that suffers from mast cell tumor, the most common skin cancer in dogs. Surgery and chemotherapy delayed the progression of the disease but failed to stop it, and immunotherapy also failed. The prognosis given by the veterinarian was only a matter of "a few months".

Conyngham has 17 years of working experience in machine learning. According to Medical Daily, he spent about $3000 out of his own pocket to conduct paired sequencing of Rosie's tumor DNA and healthy DNA at the University of New South Wales (UNSW), then used AI to screen neoantigens from thousands of mutations, and used protein structure prediction to determine which ones are likely to be "seen" by the immune system. According to a report from Reason, he used ChatGPT and AlphaFold to complete the identification of mutant proteins, and Grok for the final construct design.

He did not produce the vaccine himself. Researchers at the UNSW RNA Institute reviewed the data and produced the mRNA construct in the lab, led by the institute's director Pall Thordarson. To obtain ethical approval in Australia, Conyngham spent three months writing a 100-page document.

3D rendering of Rosie's healthy c-KIT protein rendered using Alphafold 2

Rosie received her first injection at a veterinary facility in Queensland, Australia in December 2025, followed by a booster shot the next month. Several of her tumors shrank significantly and her quality of life improved. The company page of Y Combinator states that this is considered the first mRNA cancer vaccine completely computationally designed and actually administered to a dog.

Conyngham has always refused to call it a "cure".

The ending is written in the text released by Gamgee: Rosie once went into remission, but the cancer recurred after a subsequent surgery.

The more complete version comes from Conyngham's self-narrative on X in June: that surgery pushed the cancer out of remission. Four weeks later, the tumors spread rapidly to her hindquarters, and she was in so much pain that she was no longer the dog he knew. He made the decision to euthanize her, calling it the hardest decision of his life. He wrote that in the truest sense, Rosie was a pioneer.

Gamgee Aims to Turn Handcraft Work into Assembly Line

Gamgee is a member of Y Combinator's 2026 summer batch. This $4 million seed round of financing, in addition to Founders Fund, also includes Celine Halioua, founder and CEO of canine longevity company Loyal, Steve Eidelman, founder of Modern Animal, Arash Ferdowsi, co-founder of Dropbox, and Gokul Rajaram, creator of Google AdSense.

Amin Mirzadegan, partner at Founders Fund, gave the investment reason in the press release focusing on people rather than technology: Paul crossed several disciplines for a dog, taught himself a whole new field, and kept pushing forward where most people would stop.

What the company plans to do is to industrialize the process that Conyngham manually ran through once: sequence the tumor and paired healthy tissue of each dog, identify the unique mutations of the cancer, recognize tumor-specific neoantigens, design a personalized mRNA vaccine that teaches the immune system to accurately target these targets, and manage the entire chain from veterinary consultation, sequencing, design, production, treatment to follow-up.

Clinical research is carried out in Australia, with partners including UNSW, the University of Queensland and the Garvan Institute of Medical Research. Currently, enrollment is limited to eastern Australia, and participating dogs receive the investigational vaccine for free. Eligibility for enrollment is determined by the research protocol and participating veterinary oncologists. The company says it is in communication with regulators to follow the U.S. Department of Agriculture (USDA) "patient-specific veterinary biological product" pathway.

For this round of financing of Gamgee, the company stated that the funds will be used to launch a clinical trial to verify the personalized treatment solution developed for Rosie. Conyngham also previously admitted on X that the company is recruiting veterinary oncologists and trial partners, and seeking funding to conduct a more rigorous study — which is equivalent to acknowledging that the evidence supporting its core claims is not yet in the form that regulators, veterinarians or cautious pet owners will ultimately require.

Why Dogs?

The choice of dogs is not just because Conyngham has one at home.

Comparative oncology is a field with decades of accumulation. Spontaneous tumors in pet dogs share many commonalities with human tumors in biological, genetic and histological characteristics. More importantly, they retain the natural complexity of drug resistance, metastasis and tumor-host immune interaction, which are extremely difficult to replicate in mouse models. The U.S. National Cancer Institute has a dedicated Comparative Oncology Program, and the "Cancer Moonshot" initiative has also driven resource investment in this direction.

From an industrial perspective, the appeal of the canine track also lies in a shorter regulatory path, lower costs and faster trial cycles. There are already players in this field: Torigen develops autologous tumor cell vaccines, ELIAS Animal Health develops autologous cancer vaccines combined with T-cell therapy, and Yale's EGFR/HER2 vaccine follows another more universal route that does not require tumor samples. Gamgee's differentiation lies in the "N-of-1" approach — one exclusive vaccine for each dog, rather than one universal target.

The long-term goal Conyngham set for the company goes beyond pet healthcare itself: to enable personalized mRNA vaccines to be delivered by any veterinary oncologist, and to establish N-of-1 medicine as a better treatment model across species, including humans.

Speed Is the Core Competition for Cancer Vaccines

Putting these two pieces of news together, you will find that they both ultimately point to the same issue: time.

For Merck and Moderna, the positive results obtained in Phase III clinical trials proved the feasibility of "making a separate drug for each patient". However, it took five years from the determination of the algorithm solution to the release of Phase III results, and ten years from the launch of the project to today.

In Rosie's case, it took about two years from diagnosis to the first injection, three months of which were spent on the 100-page document for ethical approval, while the time she could wait was only a few months.

The core contradiction of personalized treatment has never been just "whether it can be designed", but whether the total duration of the entire chain of design, approval, production and administration can be reduced to a range that patients (or sick dogs) can afford to wait. AI only shortens the front part of the chain — target identification and sequence design. The rest of the process has not been compressed to the same extent for now.

Conyngham's line "we failed to produce her next vaccine fast enough" is exactly referring to this matter.

Essentially, the $4 million that Gamgee raised is used to smooth out this entire chain and obtain evidence that is acceptable to regulators and veterinarians. The success of this matter depends on what data the Australian clinical research produces, not on how touching Rosie's story is.

Until then, for people whose dogs have just been diagnosed with cancer, the only responsible advice remains: go to a qualified veterinary oncologist, instead of taking a viral case as a treatment plan.

References:

https://www.reuters.com/legal/litigation/merck-moderna-say-melanoma-skin-cancer-vaccine-meets-goals-large-trial-2026-08-19/

https://www.gamgee.io/story/

https://x.com/paul_conyngham/status/2090502003449590254

This article is from the WeChat official account "jiqizhixin" (ID: almosthuman2014), authored by jiqizhixin focusing on AI, and republished with authorization from 36Kr.