Kedar Mate on why AI won’t help us fix healthcare unless clinicians fix how we use it.
The healthcare industry has never come up short on ideas on how to improve. Change the policy. Adjust the incentives. Support better evidence. Redesign workflows. Train more effective workforces. Build better technology. And now, in this new, shining era, we add “all-AI-everything” to the list.
Every generation seems to discover its own silver bullet that will finally transform healthcare. And Kedar Mate, MD, has spent enough time inside almost every corner of the system to be skeptical of them all.
“I think we get so preoccupied with our points of view on the silver bullet that's going to change everything,” he tells Reena Pande in a new episode of Leadership Rounds.
Policymakers think better policy will do it. Finance folks point to incentives. Academics always think it will be more research and stronger evidence that will change practice. Today, it’s the tech purists and engineers who increasingly believe better technology can solve the system.
Kedar's conclusion after decades of healthcare improvement work?
None of them are enough on their own.
“The truth is,” he says, “all of that is naive.”
But that doesn't make him a pessimist about healthcare AI. Far from it. Today, Kedar is co-founder and Chief Medical Officer of Qualified Health, a generative AI company building infrastructure for health systems. Before that, he spent 15 years at the Institute for Healthcare Improvement, ultimately serving as its President and CEO, and holding a position on the faculty at Weill Cornell Medicine, where he still serves today.
His move from one of the world's best-known healthcare quality organizations into a venture-backed AI company looks like a sharp turn at face value, but it’s actually in pursuit of the same mission.
Kedar began his career in global public health, working on HIV and tuberculosis in resource-constrained environments with Partners In Health, where he witnessed clinical care produce extraordinary results.
Then… the team would run out of materials. Or staff. Or some other critical resource.
So Kedar began asking what happened upstream of the clinic.
That question led him into policy work with the World Health Organization. Policy exposed him to systems problems. Systems problems eventually introduced him to the work of Don Berwick and IHI, and the connecting idea was deceptively powerful:
Systems can be changed. That became the work.
At IHI, Kedar spent the better part of two decades applying improvement science to healthcare systems globally. The organization's roots stretch back to an experiment pairing healthcare organizations with companies from industries including manufacturing and engineering to transfer methods that had made other complex industries safer and more reliable.
It was an early expression of an ethos Kedar still carries with him: “all teach, all learn.”
Healthcare doesn't have to invent every answer itself.
And over time, Kedar saw compelling evidence that disciplined improvement methods could make almost any system better.
But another problem kept showing up.
Kedar's experience at IHI convinced him that healthcare systems could improve. The trouble was keeping those improvements in place.
“We could make the situation transiently better for a health system or a population or a clinical environment,” he explains. But too often, those gains slipped through the cracks. Leadership's attention shifted. Another organizational priority appeared. The improvement team moved on. And the system began behaving the way it had before.
That left Kedar focused on two persistent challenges: stickiness and scale.
And that is where technology became interesting for him.
When a smartphone manufacturer pushes an update, millions of devices can change simultaneously.
When Google Maps changes the way it calculates a route, people can begin moving differently without sitting through a retraining program.
Technology can change behavior at enormous scale and make the new behavior persist.
“AI in particular,” Kedar says, has the two attributes he’d been looking for. It was both “scalable and sticky.” That realization eventually helped pull a lifelong healthcare-improvement leader into artificial intelligence.
It would be easy to hear that story as another variation on the familiar “AI will fix healthcare” narrative. This isn’t that. Kedar is unusually explicit about his thoughts on this.
Healthcare has repeatedly mistaken individual levers for complete solutions. Yes, better reimbursement matters. Policy matters. Evidence matters. Education matters. Technology matters. Human relationships matter.
But the extraordinary global scale-up of HIV treatment, Kedar points out, didn't happen because of one breakthrough. It required
The lesson isn't that none of our tools work. It's that they must work together.
That lesson may be most important now, as healthcare pulls AI into the mix. Because AI may be astonishingly capable. But we know that capability and impact are not the same thing.
Most people who have experimented with modern generative AI have experienced what Kedar calls those almost “magical” moments.
Ask a model to write something. Translate it. Transform it. Synthesize something that would have taken a human dramatically longer. The capability can feel almost impossible.
But in healthcare, Kedar argues, capability alone doesn't get technology to the patient, which is the entire point. He offers an analogy that really stuck with us.
AI can be like a powerful chemotherapy. But chemotherapy is useless if you can't get it into the patient's arm. In terms of healthcare AI, the needle and tubing are trust, accountability, and relationships. Without them, extraordinary technology remains extraordinary technology, not care.
“AI has this enormous potential,” Kedar says, “but it won't reach the patient. It won't reach our systems unless we have those trust and knowledge brokers.”
That reframes one of the central questions facing health systems. The challenge isn't merely selecting the best model. It's creating an environment where clinicians and organizations can responsibly trust the system enough to use it.
That brings Kedar and Reena to the big points in AI, especially right now in this moment: trust and governance. The basic technical requirement is straightforward. Tools need to perform the job expected of them accurately. But Kedar emphasizes that that’s truly just table stakes at this point.
The more difficult question is whether it performs correctly inside of the healthcare system. For whatever population it was built for, for whatever workflow it was designed to streamline, for whichever use case it was intended to enhance. Healthcare has already learned that algorithms don't perform identically when moved from one environment into another. Local context matters.
And crucially, AI systems can operate thousands of times in a mature environment. A committee meeting once a month cannot manually oversee that volume. Instead, Kedar describes governance as continuous monitoring: define the lane in which the technology is expected to operate, monitor whether it remains inside that lane, and flag deviations when performance begins to drift.
If an AI tool informs a patient's care and something goes wrong, who is responsible? Kedar's answer is grounded in how these systems are being used today.
At present, he says, Qualified Health is not delivering wholly autonomous clinical care. Clinical outputs remain filtered through human judgment. The technology can augment. It can support. It can provide intelligence. But humans remain involved in delivering care.
“So accountability still rests with the human clinician today,” Kedar says.
That may evolve as technology becomes increasingly sophisticated, but his expectation is that clinicians remain in the loop for the foreseeable future. That’s not just a safety mechanism. It points toward something much bigger about who should build healthcare's AI future.
Reena's work at Oxeon centers on placing clinician executives in positions where they can drive healthcare innovation, so she asks Kedar directly about clinician leadership inside technology companies like Qualified Health.
His answer is emphatic. At his company, doctors and nurses aren't simply reviewing a finished product from the sidelines. They are helping build and test it. That matters because a technically identifiable error and a clinically meaningful error aren't necessarily the same thing. Determining that difference requires clinical judgment. And Kedar sees the stakes as considerably larger than individual product decisions.
“This is an industrial revolution moment,” he says.
If the people closest to the clinical knowledge base aren't involved in shaping the next generation of the health economy, “we'll do it wrong, I'm sure of it.”
But he also makes an important distinction. This isn't an argument that every clinician needs to become CEO. Clinicians belong in product, validation, delivery, implementation, and leadership. And they can be uniquely powerful bridges between technologists and frontline healthcare workers.
As Kedar puts it, “There's no better ambassador for AI to clinicians than a clinician that uses AI.” The future belongs neither to technologists nor clinicians alone. It requires both working in concert.
For all the conversation's implications for the future of healthcare, one of its most useful lessons is about career moves. Reena asks Kedar how he made the decisions that produced such an unusual trajectory; he all but rejects the premise.
“I recently described my path as a series of detours,” he says.
At nearly every point where there was an obvious track (college, medical school, residency, a conventional job), he wandered away from it.
India.
Global health.
Geneva.
WHO.
Systems improvement.
IHI.
And now, tech.
These weren't random departures. They were opportunities to get closer to questions he didn't yet understand, but was passionate about finding answers for.
His best advice? Choose experience over the formal path.
Leaving IHI was perhaps the most consequential version yet.
Kedar could have imagined staying there for the rest of his career. Instead, AI felt like another moment when the interesting question lay off the established track.
“This moment felt like it was too important not to participate in,” he says. Not observe or advise from the sidelines. Participate. Build.
That philosophy has since become part of the advice he gives others. When Georgetown University awarded him an honorary Doctor of Science in 2026 and invited him to deliver its School of Health commencement address, Kedar told graduates that being “ready” is a myth and encouraged them to find their own detours.
So after decades of trying to improve healthcare, does Kedar believe we're finally on the verge of getting it right?
Not quite. Candidly, he is pessimistic about the present.
Patients and families face real challenges. Healthcare remains difficult to access and afford. Costs have risen dramatically. Many of the improvements the industry has spent decades pursuing remain frustratingly incomplete. And all of this just as AI itself is entering a period of more intense skepticism.
But Kedar doesn't think that skepticism is necessarily bad, because his optimism operates on a longer time horizon.
“I remain cautiously and stubbornly optimistic about our long-term future.”
AI, he believes, can ultimately create meaningful benefit if healthcare figures out how to use it in ways people can trust.
And perhaps that's the central idea running through all of Episode 18. Healthcare's next transformation won't happen because we finally create the perfect tool. It will happen because we’ll become better at combining the tools we already have. Between
AI may finally give healthcare a new mechanism for making change stick.
Humans will just still have to decide what change is worth scaling.
🎧 Follow Leadership Rounds on Spotify or Apple Podcasts to catch next month’s conversation. If you are building a leadership team or looking for your next career move, talk to us.
Kedar Mate, MD, is co-founder and Chief Medical Officer of Qualified Health, a healthcare tech company building AI infrastructure for health systems, and a faculty member at Weill Cornell Medicine.
Before co-founding Qualified Health, Kedar spent ~15 years at the Institute for Healthcare Improvement (IHI), ultimately serving as President and CEO from 2020 through 2024. His work at IHI spanned healthcare quality, patient safety, large-scale systems change, health equity, value, and global health, including earlier leadership of IHI programs in South Africa and across international markets.
Earlier in his career, Kedar worked with Partners In Health on HIV and tuberculosis care in resource-constrained settings and with the World Health Organization, where he served as a special assistant to the Director of the HIV/AIDS Department during a period of major global HIV treatment scale-up. He subsequently completed his internal medicine training at Brigham and Women's Hospital and joined the faculty at Weill Cornell Medicine, with clinical work at NewYork-Presbyterian Hospital.
Kedar's scholarly and professional work has focused on healthcare quality, health-system design, implementation and improvement science, large-scale change, health equity, and strategies for improving healthcare value. He has published peer-reviewed research, book chapters, and white papers and has served on multiple healthcare and healthtech boards. His recognitions include being a Paul & Daisy Soros Fellow, Fulbright Specialist, and Aspen Institute Health Innovators Fellow. In May 2026, Georgetown University awarded Kedar an honorary Doctor of Science. Kedar earned his bachelor's degree in American History from Brown University and his medical degree from Harvard Medical School. He completed his internal medicine residency at Brigham and Women's Hospital and is board-certified in internal medicine.
