Here is a fact about magnetic resonance imaging that is either mundane or slightly alarming depending on how much you think about it: two technologists can put the same patient in the same machine on the same afternoon and produce meaningfully different images. MRI is not like taking a photo, where you point and the camera does the rest. It is more like conducting an orchestra where the sheet music is a few dozen physics parameters, the tempo is the patient's heartbeat, and the whole thing has to be re-scored for each person. Cardiac MRI is the hardest version of this - a moving target, timed breath-holds, planes that have to be angled just so. It is, in the technical parlance of the field, a pain.
Vista AI's entire premise is that this pain is a software problem. The company, founded in 2012 as HeartVista by three researchers - William Overall, Juan Santos and Bob Hu - out of Stanford's Magnetic Resonance Systems Research Laboratory, makes software that automates the acquisition: where to position, how to plan the slices, how to tune the sequence. The pitch is that a general technologist, not a rare cardiac specialist, can press a button and get a specialist-grade scan. The product is called Vista AI Scan, it is FDA-cleared, and it runs on the Siemens and GE machines hospitals already own.
"Radiology faces a critical bottleneck: MRI demand vastly exceeds skilled technologist supply, burdening staff and delaying patient care."
— Daniel Hawkins, President & CEOThe market is a subtraction problem
The thesis is almost boringly demographic. The number of people who need MRIs goes up every year. The number of technologists trained to run the complicated ones does not. Subtract one from the other and you get backlogs, burnout, and patients waiting weeks to find out what their heart is doing. You can respond to that gap by training more specialists, which is slow, or by buying more machines, which is expensive and doesn't solve the staffing part. Or you can make the machine you already have easier to operate. Vista AI sells the third option.
What's clever - and slightly unusual for a healthcare AI company - is where they aimed. The crowded, glamorous corner of radiology AI is reading: algorithms that flag a nodule or measure an ejection fraction after the fact. Vista AI went the other direction, to acquisition. The logic is unglamorous but hard to argue with. A brilliant reading algorithm fed a badly acquired scan produces a confident wrong answer. Fix the input, and everything downstream gets better. It is the plumbing, not the penthouse, and plumbing is where the money quietly is.
When your customers write the check
In January 2026 the company - which rebranded from HeartVista to Vista.ai at the RSNA meeting in late 2022, to signal that the heart was just the start - closed a $29.5 million Series B. The interesting thing about the round is who is in it. New investors included Cedars-Sinai, Intermountain Health, the University of Utah's hospital system, Temple University's Fox Chase Cancer Center, and Tampa General. These are not venture funds. They are hospitals. They are, in many cases, the customers.
When a customer becomes an investor, one of two things is true: either the salesperson is very good, or the product actually works and the buyer wants more of it to exist. Health systems are famously slow to part with money and famously allergic to risk. Five of them putting equity into a 36-person software company is a stronger signal than any testimonial. Khosla Ventures, an early backer, and Bold Brain Capital rounded out the round, bringing total funding to roughly $38 million.
"Vista AI uniquely addresses workforce shortages through automation - leading health systems recognize it as essential infrastructure."
— Bruce Armstrong, Partner, Khosla VenturesThe numbers that make a hospital pay attention
The case studies are the sort of thing that make a hospital operations director sit up. Brigham and Women's Hospital used Vista AI's software to eliminate a 28-day cardiac MRI backlog and increase its cardiac MRI slots by roughly half. Radiology Regional, a group in southwest Florida, reported cutting scan time by more than 50%. These are not moonshot numbers; they are throughput numbers, the kind that show up on a spreadsheet and pay for the software several times over. A 28-day backlog is not an abstraction. It is patients waiting a month for an answer about their hearts, and then not waiting.
None of this required a new machine. That is the part worth sitting with. The MRI scanner in the basement of a regional hospital is, in hardware terms, roughly as capable as the one at a flagship academic center. What differs is the expertise available to drive it. Vista AI is a bet that you can encode a lot of that expertise in software and ship it to the basement.
The heart is just the start
The Series B money is pointed at two expansions. The first is anatomical: brain, prostate and spine, each of which will need its own FDA clearance and each of which represents a larger imaging market than cardiac. The second is operational - remote MR imaging, where the expertise runs the scan from somewhere else entirely, so a rural hospital can deliver specialist-grade imaging without a specialist on staff. If it works, it turns MRI from a thing you can only get done well in certain buildings into a thing you can get done well anywhere. That is a genuinely large idea dressed in the modest clothing of workflow software, which is more or less how the good ones usually show up.