Turns out conversations in healthcare already had the answers
Disruption in healthcare too often starts as an outside job. Health systems tend to attract non-physician outsiders eager to punch down and pontificate about how things should work with less regard for how they actually do.
Even in healthcare tech, where tools claim to be built for clinicians, talk of “transformation” can feel more like theater. Health system leaders may smile and nod at vendors’ grand visions, only to return to teams that roll their eyes and workflows that remain largely the same, minus a few keystrokes.
But what if teams flipped the model entirely? What if, instead of chasing scale, a company became obsessed with the smallest, seemingly ordinary thing?
What if transformation began with a conversation?
An earnest vision, built with exacting work
Shiv Rao, MD, decided to test that idea.
A practicing cardiologist and former executive at UPMC’s provider-facing venture arm, he’s the rare AI CEO who has spent more time inside a health system than outside it. “I love being in rooms with doctors and nurses, but also computer scientists and product managers and engineers,” he said.
In 2018, Dr. Rao and Zachary Lipton, PhD, founded Abridge on a simple, provocative premise: the everyday conversation between clinician and patient is one of healthcare’s most valuable and underused signals. Abridge’s generative AI powers clinical documentation across inpatient and outpatient settings, spanning numerous specialties and languages with precise understanding of therapeutics, pharmaceuticals, and diagnostics.
“When we started the company,” Dr. Rao recalls, “we already had slides with a statement that has turned into part of our mission, which is that we are building technologies that leverage the conversation to equip clinicians to be near-omniscient and omnipresent detectives, communicators, advocates and healers.”
Don’t mistake the eloquence for softness. Beneath Dr. Rao’s earnest language is steely resolve. From the start, he aimed higher than the low-stakes, high-frequency use cases that dominated the early AI scribe market. The company has leaned into the hardest tests, with Dr. Rao only half-joking that rigorous health system pilots are akin to “eating glass.” Abridge is live in more than 200 health systems today, including Kaiser Permanente, Mayo Clinic, Hartford HealthCare, UCHealth and UW Health.
Children’s Healthcare of Atlanta recently joined that list, having completed a randomized cross-over pilot with Abridge. “Overall, comments were extremely positive, with strong improvements in burnout using the Stanford Model of Occupational Well-Being,” the system’s CMIO, Evan Orenstein, MD, said.
As Abridge scales from pilot to platform, it moves toward a larger goal: rewiring the economics of care, where administrative costs alone amount to nearly $1 trillion each year.
Barry Stein, MD, Chief Clinical Innovation Officer and CMIO, and Founder of the Center for AI Innovation in Healthcare at Hartford HealthCare, said that drive was evident from the start. “It was obvious to me talking to them that they have massive ambition to transform healthcare, by taking the work out of workflow,” Dr. Stein said. “This is just the beginning.”
A high bar to clear
Vision alone doesn’t transform a health system. CMIOs and innovation chiefs have weathered decades of “transformative” technology pitches that ended up changing little at all.
“My team and I, we know that of the giant number of the things that we are looking at, we may never act on or participate in but we always learn something,” said UCHealth Chief Medical Officer and Chief Innovation Officer Richard Zane, MD. He frames his work — and partnerships — through three principles: Move fast and abhor bureaucracy. Create solutions that benefit people. And fear inaction, not failure. “Our bar is exceptionally high,” he said. “It’s almost impossible.”
In Madison, Wis., UW Health CMIO Joel Gordon, MD, operates with healthy skepticism, too. He recalls working in this space before it even earned the title of informatics. For years, his role meant trying to persuade weary clinicians to adopt clunky tools — work that felt more like salesmanship, less like problem-solving.
“We were putting doctors and nurses through days of training just to use this stuff,” Dr. Gordon remembers. “And how many days of training did they have to use their iPhone? None. It was just so intuitive, and there was just zero level of that happening in healthcare technology.”
Meanwhile, physicians across several specialties prepared to scale back or leave practice altogether, despite a dense catalogue of tools claiming to help. From 2022 to 2023, 36% of doctors said they had a moderate interest in leaving their current jobs within two years, according to the American Medical Association. The reasons were consistent: unsustainable workloads, bureaucratic burdens and the cumulative toll of burnout.
So when AI for clinical conversations arrived on the scene, leaders like Drs. Zane, Stein and Gordon kept their bars high.
Dr. Zane even passed at first. “We met Abridge a long time ago when they were in their infancy and in all candor, we passed on the opportunity, which is one of our big misses for sure,” he recalled. At the time, UCHealth opted against ambient AI before returning to Abridge and partnering in 2024. This fall, UCHealth will roll out a systemwide deployment across its ambulatory and inpatient settings across Colorado.
At Madison, Wis.-based UW Health, Dr. Gordon’s team ran “a very, very rigorous trial, not an observational analysis” at the six-hospital system. At 500-location Hartford, Dr. Stein’s team embedded two competing solutions directly into their EHR, assigning them to clinicians across specialties and switching after two weeks. The verdict was clear: “We got an overwhelming preference in almost all specialties for Abridge,” Dr. Stein said. “The voice of the customer was overwhelming.”
The next frontier
These health systems are each expanding Abridge’s footprint. UW Health adds 100 clinicians as Abridge users per month, with utilization rates approaching 80% post-training. Children’s Healthcare of Atlanta is targeting similar momentum, with plans to triple participation. Hartford HealthCare is scaling across physicians and advanced practitioners — and co-developing new solutions along the way, with emphasis on real-time, AI powered documentation for nurses, which the system will test this fall, as well as prior authorization and risk adjustment.
“We want to make sure that we leverage all the learnings that Abridge and our clinicians have had in primary and specialty care and accelerate the development of the nursing capability together,” Dr. Stein said.
The expansion into nursing underscores Abridge’s broader strategy: to strengthen the foundation, not just widen the surface. As Dr. Rao puts it, real efficiency comes from how much value a single conversation can unlock downstream. One patient-clinician exchange sets off a cascade — the note becomes a code, the code a claim, rippling into risk adjustment, quality reporting, prior authorization and clinical decision support. Users also point to its potential to streamline clinical trial enrollment and enable new research.
“As long as you unlock the point of conversation and that ability to shift left into it, you can start to bend the trajectory for everything that happens afterwards,” Dr. Rao said. “We’re going millions of miles deep on that idea.”
The view of the clinical note as infrastructure opens entirely new possibilities, but not before returning to its primary point. “As the note has become co-opted by many non-clinical needs, the extra demands can easily pull clinicians away from the core relationship with the patient and family,” Dr. Orenstein noted. By automating note creation through ambient listening, Abridge returns time and focus to the clinician-patient relationship while keeping documentation accurate and compliant.
As the AI listens, Dr. Orenstein said, it could surface relevant guidelines or extra questions that drive evidence-based decisions. “Many of our clinicians are next looking to AI scribes to queue up orders in addition to the note-taking capabilities they already demonstrate,” he said, noting that context — like whether a child’s ear infection is a first visit or a return after several days on antibiotics — can shape the right clinical decision but often isn’t captured in the EHR. “AI-driven order entry could easily become the next frontier after computerized provider order entry, with many risks and opportunities along the way,” said Dr. Orenstein.
That dot-connecting depends on rigorous science, which is both a source of investment and pride at Abridge. Chief Technology Officer Dr. Lipton leads a research team known for publishing its methods openly. This fall, the team released results of a homegrown system for catching hallucinations in clinical notes. It identified and corrected 97% of errors, catching six-times more than GPT-4o.
“The science [has to be] so good that we can truly honor the humanity of healthcare,” Dr. Rao said.
And while the science needs to be excellent, health system leaders also point to another dimension of their partnership that has stood out: how Abridge actually shows up at the table— how its people listen, respond, iterate and embrace the complexity of real clinical work.
“That’s what partnership should look like,” said Dr. Zane. “It’s not just the technology. We have deployed a lot of technology successfully and they are one of the best, if not the best, teams we’ve worked with.”
Listening at scale
This is where conversation takes on another meaning. What began as an idea — that clinical dialogue becomes infrastructure — now shapes how Abridge works. Its progress depends on ongoing, unfiltered exchange between innovators, clinicians, and engineers. Inside Abridge, “feedback is our oxygen” isn’t a slogan — it’s a system.
“This company is the first thing I think about when I wake up. It’s the last thing I think about before I go to bed, and it’s most of what’s in between,” said Dr. Rao. “So feedback means everything.”
Even with Series E funding and a $5.3 billion valuation, Dr. Rao still dives into the company’s feedback channels daily, where tens of thousands of clinician comments stream in. He used to do this manually — “I used to spend several hours every single day reading every single bit of feedback from every single doctor that was using Abridge,” he said.
Now, given the volume — the technology will support more than 50 million conversations in 2025 — Abridge built AI tools to synthesize feedback while keeping individual voices intact. “I make a habit of reaching out to specific doctors just to help them understand that, ‘hey, we hear you,'” Dr. Rao said. “You’re not speaking into the abyss.”
Dr. Gordon noted Abridge has a “special sauce” for responsiveness. “They are good at being what organizations need and flexing accordingly based on the small peculiarities between [partners],” he said.
For Abridge’s partners, that responsiveness reflects something deeper — a company grounded in humility and respect for the realities of clinical work.
“Aligned culture with Abridge is very important,” said Dr. Stein at Hartford HealthCare, emphasizing the company’s fierce focus on quality and safety. “To make sure that whatever we are doing is safe for our patient, for our clinician and for us — that’s paramount and we never compromise.”
Where transformation really begins
Turns out, conversation is both the medium and the mission. It’s what links Abridge’s engineers, clinicians and patients — a continuous loop of learning and listening. Each exchange, whether inside a feedback thread or at the bedside, helps ensure that the everyday work of health systems, long underappreciated and undercredited, is captured in all of its clinical, operational and human fullness.
For health systems like UW Health, UCHealth, Hartford HealthCare and Children’s Healthcare of Atlanta, AI for clinical conversations began as a way to strengthen the patient-clinician relationship and ease the weight of documentation. But with a partner committed to rigor and responsiveness, they’re finding more: deep intelligence at the point of conversation. Every conversation can unlock better data, sharper decisions and a renewed sense of purpose as care happens.
As Dr. Stein put it, “The sky’s not the limit anymore.”
That’s the quiet shift underway, with transformation that starts inside from the clinicians and patient interactions they have every day. It’s a reminder that innovation starts in a way too many before have overlooked: with a conversation worth hearing.
At the Becker's 11th Annual IT + Revenue Cycle Conference: The Future of AI & Digital Health, taking place September 14–17 in Chicago, healthcare executives and digital leaders from across the country will come together to explore how AI, interoperability, cybersecurity, and revenue cycle innovation are transforming care delivery, strengthening financial performance, and driving the next era of digital health. Apply for complimentary registration now.