Viz.ai

The problem is logistics, not diagnosis

In a stroke, every minute until treatment counts. The problem Viz.ai went after is not diagnostic accuracy but logistics. The scan lands in the radiologist's reading queue, the radiologist reports, someone calls the neurologist, the neurologist finds a team, and meanwhile hours go by. The software scans the CT angiography as soon as it is produced, detects a suspected large vessel occlusion (LVO), and sends an alert straight to the right doctor's phone. It runs alongside the usual pathway, not instead of it.

Eleven FDA clearances: what they are and what they are not

The regulatory database of the U.S. Food and Drug Administration (FDA) lists eleven 510(k) clearances for the company between 2018 and 2025: for suspected stroke, intracranial hemorrhage, aneurysm, subdural hemorrhage and abdominal aortic aneurysm. It also lists a De Novo authorization granted in 2023 for a product that detects hypertrophic cardiomyopathy. The regulatory definition is precise and important: these are triage and notification tools. The software does not diagnose and does not replace a doctor's reading. It shortens the time until the right doctor takes a look.

A Medicare payment for an algorithm

In 2020, the U.S. Medicare agency granted the software a New Technology Add-on Payment. This payment mechanism is usually given to devices and drugs, and almost never to software. It was a business milestone for the whole field, because it established that a dedicated reimbursement can be paid for an algorithm, not only for hardware or a drug.

A British founder and an Israeli founder: what makes it "Israeli"

Viz.ai was founded in 2016 at Stanford by two partners: Dr. Chris Mansi, a UK-trained neurosurgeon who serves as CEO, and Dr. David Golan, an Israeli postdoctoral researcher in machine learning who served as CTO. In April 2022 the company raised $100 million at a valuation of about $1.2 billion. Its headquarters are in San Francisco, so only one of its two founders is Israeli. But its development center is in Israel, and that is the criterion that puts it in this category. The geographic rule looks at where a company was founded or where its development center is, not at the citizenship of each founder.

Saving time, not necessarily improving diagnosis

The idea behind the product is to shorten the time between the scan and the decision, rather than to improve the decision itself. It has become a common pattern in medical AI. It is also easier to prove: you can measure in minutes how much earlier the patient reached the catheterization lab, while proving a better diagnosis takes long and costly studies. Still, saving time is not the same as a better clinical outcome. An alert tool that flags too many non-urgent cases wears down the team's trust. That is one reason clearances in this field are for use as an extra layer alongside the usual pathway, not instead of it.