Why Veritope, and how we intend to earn trust
We're starting a company to find the tumor targets a patient's immune system can actually see — and we're going to show our work.
Cancer is, in part, a problem of recognition. A tumor is built from a patient’s own cells, and it survives by being hard for the immune system to tell apart from healthy tissue. Some of what makes a cancer visible to the immune system are specific fragments it displays on its surface — and some immune cells can learn to see them.
The hard, valuable question is not “what might the immune system attack?” It is “what is this tumor truly showing, and what can the immune system truly recognize?” Most candidate answers in the literature are predictions that were never actually seen on a real tumor, or that can’t be told apart from a healthy look-alike. Chasing those costs years.
We think this is, at its core, a data-and-modeling problem — the kind disciplined AI is genuinely built for. That is the bet behind Veritope.
Where we start
We start in companion-animal oncology. Dogs develop cancer with real biological parallels to human disease, and working here lets us build and validate with genuine ground truth, shorter timelines, and families who want every good option for their animal. As approaches prove out, the same methods are designed to translate toward human cancer — carefully, and only when the evidence earns it.
How we’ll earn trust
The failure mode in this field is letting a hypothesis wear the clothes of a result. So we hold ourselves to a simple discipline, and we’ll show it publicly:
- Know. Published, reproducible evidence — the ground we stand on. We cite it and re-check it.
- Believe. Hypotheses we find promising but unproven. We label them as such, and design tests that could prove them wrong.
- Prove. The specific, pre-registered result — measured against strong baselines, on data the model has never seen — that would move a belief into knowledge.
We’ll prove value on public data before spending on new experiments. We’ll publish our data-source inventories, our hypotheses, and our misses — not just our wins.
This is the first of our notes. More to come, in public.