Shiv Rao, featured on this year’s A.I. Power Index, has built what may be one of the most consequential applications of A.I. in medicine today. A practicing cardiologist turned founder, Rao saw firsthand how much of modern medicine’s burden falls on the complexity of the paperwork surrounding care. After long days with patients, hours of charting and documentation often followed, contributing to record levels of burnout across the profession. His healthtech startup, Abridge, set out to fix that by transforming one of healthcare’s most abundant yet underutilized data sources: the doctor-patient conversation.
Founded in 2018, Abridge uses ambient A.I. to convert spoken exchanges between clinicians and patients into accurate, real-time clinical notes that integrate directly into electronic health records. What began as an ambitious experiment in language understanding is now a $5.3 billion enterprise trusted by over 150 major health systems across the U.S. In a sector where most startups position themselves as disruptors to major EHR players, Abridge has instead chosen collaboration, partnering with Epic to embed its technology into existing workflows rather than rebuild them. Health systems adopting Abridge have reported reductions in after-hours work of more than 80 percent and dramatic improvements in clinician satisfaction. Yet Rao insists the true goal lies in restoring empathy and connection to clinical practice. “It’s not just about cutting costs,” he says. “It’s about bringing humanity back to healthcare.”
You’re a practicing cardiologist who built an A.I. company now worth $5.3 billion. What problem did you see in the hospital that tech entrepreneurs working from Silicon Valley completely missed?
As a doctor, nothing was more soul-crushing than working a full day helping patients only to come home and have hours of documentation. I knew the key was unlocking what was said in the exam room in real-time: gathering intelligence at the point of conversation. Now Abridge’s enterprise-grade ambient A.I. is trusted by more than 150 of the largest and most complex health systems across the U.S. and we are extending the value of clinical conversations across workflows for providers, payers and patients.
Abridge converts doctor-patient conversations into clinical notes in real time. How often does the A.I. misunderstand something critical, and how do you handle those mistakes?
Science is at the heart of Abridge. Every note is linked back to the conversation transcript so clinicians can verify evidence before signing off. We run continuous quality-assurance loops and ship models only after clinician-in-the-loop evaluations. Clinicians remain the final sign-off. Abridge’s pioneering models ensure that A.I. hallucinations are eliminated from draft documentation for market-leading accuracy that clinicians can trust. Abridge caught 97 percent of confabulations, or unsupported claims in draft clinical documentation, outperforming GPT-4o, which only caught 82 percent.
You partnered with Epic rather than trying to replace them. Most healthcare startups want to disrupt the EHR giants. Why did you choose integration over disruption?
Our partnership with Epic has been and will always be rooted in solving problems that matter most to health systems, clinicians and patients. By focusing on the conversation itself, we enable better workflows, deeper integration and meaningful outcomes at scale. We are meeting clinicians where they are and this is only at the beginning of what’s possible at the point of conversation.
Your platform now serves over 150 health systems, growing 50 percent in just four months. What’s driving that rapid adoption—is it physician burnout or genuine efficiency gains?
It’s both. Leaders feel the urgency to address burnout, but adoption only sticks if the efficiency is real. Abridge integrates seamlessly within EHR workflows—across care settings, 55+ specialties, and 28+ spoken languages. This year alone, Abridge will support clinicians across more than 50 million medical conversations, helping to reduce burnout by up to 60 to 70 percent. At Sharp Healthcare, clinicians who use Abridge reported an 83 percent reduction in note-writing effort. At Lee Health in Florida, 86 percent of clinicians reported doing less after-hours work.
Administrative costs are 30 percent of healthcare spending. If AI documentation works perfectly, how much of that waste can you actually eliminate?
The 30 percent stat is staggering, but it’s not just about cutting costs—it’s about bringing humanity back to healthcare. With Abridge, return on investment is more encompassing than improving work relative value units, increasing operational efficiencies and reducing clinician recruitment and retention costs. It’s also what happens when clinicians feel joy again, when patients feel heard and when health systems eliminate impossible backlogs and unsustainable workflows.
You’re expanding from general documentation into pediatrics and emergency care. What makes medical A.I. different across specialties—isn’t documentation just documentation?
Specialty context changes everything. In pediatrics, the “historian” is often a parent or multiple caregivers, so the A.I. has to attribute who said what, capture developmentally appropriate histories and encode family-centric plans. In emergency medicine, speed, brevity and structured clinical reasoning matter. Triage cues, time stamps and problem-oriented summaries have to be spot-on. We’ve built specialty-tuned models and workflows with children’s hospitals and ED teams so the note reflects the way those clinicians think and document, not a one-size-fits-all template.
