Care at the Speed of Ideas

Every year, millions of patients move through hospitals carrying medical records full of information that could change their care.

A physician sees a pattern and knows exactly what should be done about it. A researcher publishes a finding that could catch a disease months earlier. A clinical team identifies a gap in follow-up that puts patients at risk. The ideas are sound. The evidence is there. But turning any one of them into something that actually reaches a patient is a different problem entirely.

Nishith Khandwala spent two years learning just how different.

Nishith and his cofounder, David Eng, were computer science graduate students at Stanford when a cardiologist from the School of Medicine came to them with an idea. He had noticed a pattern in the patients he treated for heart attacks: many had received chest CTs years earlier, for reasons unrelated to their hearts, and the warning signs of cardiovascular risk were already sitting in those scans. But since no one was looking for these risks, nothing was done. His idea was to use AI to read every one of those scans for cardiovascular risk, so care teams could intervene before a patient ever had a heart attack. He knew what should be done. He had the clinical reasoning, the data to support it, and colleagues who agreed. What he needed was a way to put that idea to work inside a hospital.

Nishith and David built the technology to do it. The algorithm worked. The clinical logic was validated. And then came the part no one had warned them about. Getting a new tool into the hands of the care teams who could actually use it meant navigating IT reviews, integration requirements, compliance approvals, and a queue of institutional priorities that moved at its own pace. After two years and the combined efforts of some of Stanford's top cardiologists, the physician's idea had still never reached a single patient.

Then, Nishith’s father had a heart attack.

He recovered, and he is well today. But in the aftermath, the family learned that an earlier scan had already carried the exact warning signs Nishith’s work was built to catch. The information had been there. No one had a system in place to surface it.

That experience is the reason Bunkerhill Health exists. The company built an AI platform, Carebricks, that helps health systems take clinical and operational ideas and turn them into working solutions, fast. Nishith’s view is that most of healthcare's hardest problems stem from a mismatch between how much care people need and how little capacity there is to deliver it. Patients wait months for specialist appointments. Critical findings go without follow-up. Eligible patients miss out on newly approved treatments because overburdened clinical teams simply cannot keep up.

In nearly every case, a physician, a nurse, or an administrator inside the hospital already knows what should happen. The knowledge exists. The capacity to act on it at scale does not.

Other software teams have tried to tackle that gap one workflow at a time, building a bespoke tool for each individual problem. Nishith and David made a different bet. Instead of building a separate product for every clinical or operational need, they built a central brain that could power many of them at once.

"You don't build a separate power plant for every house," Nishith says. "You build a power plant and a grid that can power them all."

That central brain became Carebricks. Rather than arriving at a hospital with a fixed product to pitch, the company asks the clinicians, administrators, and operational leaders within a health system to name their most pressing problems, then partners with them to turn those specific ideas into working AI agents. The goal, as Nishith frames it, is to lower the cost of trying a new idea and raise the pace at which a hospital can try the next one.

The distinction matters. Healthcare AI is filled with tools that each address one narrow slice of the work. Nishith argues that clinical and operational teams should not have to stitch together a hundred vendors to cover a hundred problems. Bunkerhill's case is that a single platform can span both the clinical and the operational without going shallow on either.

That breadth is visible at the University of Texas Medical Branch, which Nishith calls one of the most AI-forward health systems he has worked with. UTMB now runs more than 20 agents through Carebricks, covering both day-to-day hospital operations and frontline clinical care. On the operational side, teams at UTMB used the platform to identify the sickest kidney disease patients waiting for specialist appointments and move them up in the queue, cutting average wait times by more than 50 percent. On the clinical side, results have been just as striking: within the first month of one deployment, a coronary calcium agent running an FDA-cleared algorithm flagged a patient who had no idea he was at imminent risk of a heart attack. UTMB's cardiology team confirmed the finding, performed a triple bypass, and credited the early catch with saving his life.

The company grew its revenue 20x last year and earned seven new FDA clearances, licensing clinical algorithms from academic centers and carrying them through the regulatory process so hospitals can put them to use.

That platform approach is what Felicis backed. We first invested in Bunkerhill's Series A in 2023 and recently participated in the Series B (opens in new tab).

For all the range of what the platform enables, Nishith is clear about where the technology does not belong. He believes AI should make healthcare feel more human, not less. The relationship between a care team and a patient, especially when delivering difficult news, must remain deeply personal. Trust is why patients rely on their doctors and nurses in the first place, and that bond is something software can never replace.

Nishith keeps returning to a single measure of whether the company is working.

"We don't hinge upon any single agent," he says. "We want to shrink the distance between a good idea and the patient who needs it. When a clinician has an idea on Monday and it's helping patients by Friday, that's a completely different healthcare system."

It is the problem he started with as a student. The clinical innovation was never the hard part. Equipping medical staff to deliver it to the patient was. Bunkerhill Health is a bet that the distance between the two can be made small enough to walk across.