15 beagles, 4 cancers, 94% accuracy: my favorite AI workflow in health tech

Thirteen days from now, the FDA convenes a panel that has never convened before.

Funding stage: Series A Plus. SpotitEarly announced a $20.3M round in May 2025 on top of a 2022 seed, and is currently running a Regulation A+ public offering at $7.15 per share. Institutional Series A capital plus a live retail raise puts this well past seed.

On September 23, 2026, the FDA's Molecular and Clinical Genetics Panel sits down to vote on Grail's Galleri test, the first multi-cancer early detection test ever to reach an FDA advisory committee. Every founder building in early detection is watching. Shlomi Madar is watching too. He just is not building the same thing.

Madar is the CEO of SpotitEarly, an Englewood, New Jersey company founded in 2020 that screens for four cancers from a single breath sample collected at your kitchen table. Grail reads methylation patterns in cell-free DNA. SpotitEarly reads a mask you breathed into for a few minutes. And the sensor at the front of its pipeline is not a machine. It is a beagle.

This is the part where most people assume it is a gimmick. It is not. Madar has a PhD in cancer research from the Weizmann Institute of Science, and the reason the company landed on dogs is boring and rigorous: nothing built by humans comes close. The entire field of electronic noses, mass spectrometry included, keeps running into the same wall, which is sensitivity threshold. A dog detects at parts per trillion. Madar's favorite way to explain it: fill two Olympic swimming pools, stir half a teaspoon of sugar into one, and a dog will tell you which pool you tried to trick him with.

The problem with a dog is not the nose. It is that a dog cannot fill out a lab report. That is the entire engineering problem SpotitEarly set out to solve. Madar's team spent five and a half years turning an animal's opinion into a reproducible, scalable, auditable data output. You breathe into a surgical mask, seal it, and mail it. At the lab, the mask goes into a sniffing port, where cameras, a microphone, and accelerometer vests capture the dogs' behavioral and biological signals. Each sensor feeds its own trained machine learning model, and multiple dogs sniff each sample more than once, throwing out anything that deviates from an established behavioral pattern.

The company calls the platform LUCID. The Rainbow Study, run with medical centers in Israel including Sourasky, Hadassah, and Rambam, put 1,386 participants through it under double-blind conditions and reported 93.9% sensitivity and 94.3% specificity, published in Scientific Reports. Madar is careful, repeatedly and unprompted, about what the test is: a triage layer that routes people toward the full standard of care, not a replacement for it.

The go-to-market is the founder story underneath the science. Only about 14% of cancers in the US are caught through a recommended screening test, and Madar walks through why: invasiveness, distance, cost, testing fatigue, and scanxiety. SpotitEarly priced its panel at roughly $500, well under Grail's $950, after surveying physicians and patients on price tolerance directly.