The answer to your cancer was already on the slide. They taught a machine to read it.
Somewhere in a pathology lab right now, a glass slide is sitting under a microscope. A thin ribbon of bladder tissue, stained the standard pink and purple, the same way it has been stained for a century. A pathologist glances at it, writes a few lines, and moves on. The slide goes in a drawer. What nobody in the room can see is that the slide is quietly holding the answer to a life-or-death question: will this patient's treatment work?
Valar Labs built a machine that can see it.
That is the whole bet, and it is a strange one. Most diagnostics companies want to run a new test, take a new biopsy, sequence a new genome - add something. Valar's founders went the other direction. They asked what was already there, ignored, in the slides every cancer patient already has. Their answer, it turns out, is a great deal.
The company sits in Palo Alto with a team of roughly 27. It is small. The problem it picked is not. Bladder cancer is one of the most common cancers in the world, and most cases are caught early, at the non-muscle-invasive stage, where the tumor is treatable but maddeningly unpredictable. Two patients with slides that look identical to the human eye can go on to have completely different fates. One responds to the standard BCG therapy. The other does not, loses months, and watches the cancer return.
"Only one in two patients responds to standard care."
Vesta takes a standard H&E-stained pathology slide - no new biopsy, no special stain - and runs it through two layers of intelligence. First, a computer-vision model reads the high-resolution tissue image and pulls out thousands of histological features, many of them invisible to a human pathologist. Then a clinical AI model translates those features into something a doctor can act on: a prediction of who will benefit from BCG therapy, and a detailed risk score for recurrence and progression.
Risk Stratify Dx. Predicts which non-muscle-invasive bladder cancer patients are unlikely to respond to BCG, and stratifies recurrence and progression risk. The first AI digital-pathology prognostic test in bladder cancer to win FDA Breakthrough Device Designation.
The Vesta platform extended to prostate cancer for risk stratification and treatment-response prediction - the next front in the same approach.
An additional oncology diagnostic in the Valar Labs lineup, broadening the company's reach across solid-tumor cancer care.
Proprietary models trained on thousands of patient specimens paired with real clinical outcomes - built to surface predictive image features no human review can appreciate alone.
"If this information was knowable, every patient and provider would want to know."
They met around 2020 in Andrew Ng's machine-learning group at Stanford - the kind of room where AI ambition and medical seriousness sit at the same table. The combination shows in the company they built: peer-reviewed, clinically validated, and unusually patient about proof.
| Round | Amount | Date | Lead Investors |
|---|---|---|---|
| Seed | $4.15M | 2022 | a16z Bio + Health, Pear VC |
| Series A | $22M | May 2024 | DCVC, a16z, Pear VC |
a16z's read on the deal: Valar is AI solving a "magic quadrant" problem - hard-to-replicate models doing work humans simply cannot do alone. Vineeta Agarwala and James Hardiman joined the board with the Series A.
Avoid months lost on a therapy that was never going to work - and make the treatment decision with data instead of a coin flip.
Get a risk score and response prediction from features in the tissue that no human review surfaces, layered onto care you already deliver.
Run as a CLIA-validated, insurance-billed lab test - in the model of genomic testing, but on slides that already exist, cutting wasted cost of care.
Return to that slide under the microscope. Same pink-and-purple tissue, same drawer it would have gone into. The difference now is what happens before it gets there.
The slide gets scanned. Valar's models read the thousands of features the human eye slides past, and the pathologist's few lines come with something new attached - a number, a prediction, a sense of which way this particular cancer is likely to break. The drawer is still there. But the slide is no longer just a record of what is. It is a forecast of what will be, and a patient walks out of the clinic with a treatment chosen on evidence instead of odds.
Valar Labs did not invent the slide. It just refused to let it stay quiet.
Want to hear it in their words? Search the Pear Healthcare Playbook interview with CEO Anirudh Joshi, and check the Valar Labs site for product walkthroughs and demo videos of the Vesta platform.
Sources: Valar Labs, TechCrunch, a16z, DCVC, Pear VC, Urology Times, BusinessWire, MD+DI. Figures are public as of June 2026; revenue and headcount are approximate.