No PriorsNo Priors Ep. 108 | With Abridge Founder and CEO Shiv Rao, MD
Sarah Guo and Shiv Rao on aI-powered medical notes relieve burnout and reshape healthcare’s front line.
In this episode of No Priors, featuring Sarah Guo and Shiv Rao, No Priors Ep. 108 | With Abridge Founder and CEO Shiv Rao, MD explores aI-powered medical notes relieve burnout and reshape healthcare’s front line Abridge founder and CEO (and practicing cardiologist) Shiv Rao explains how the company uses specialized speech recognition and generative AI to turn clinician–patient conversations into high‑quality clinical and billable notes in seconds, dramatically reducing documentation burden. He describes the post‑pandemic burnout crisis, why Abridge chose the hardest enterprise health-system segment first, and how trust, integrations, and research depth let them compete head‑to‑head with incumbents like Microsoft. Rao details Abridge’s technical stack—from multilingual, medical-grade ASR to orchestration of models for documentation, billing, and patient summaries—and how feedback at scale fuels continuous preference tuning. He also outlines the broader roadmap: automating orders, revenue cycle, clinical trial matching, and ultimately clinical decision support, while emphasizing the enduring centrality of human clinicians and patient agency.
At a glance
WHAT IT’S REALLY ABOUT
AI-powered medical notes relieve burnout and reshape healthcare’s front line
- Abridge founder and CEO (and practicing cardiologist) Shiv Rao explains how the company uses specialized speech recognition and generative AI to turn clinician–patient conversations into high‑quality clinical and billable notes in seconds, dramatically reducing documentation burden. He describes the post‑pandemic burnout crisis, why Abridge chose the hardest enterprise health-system segment first, and how trust, integrations, and research depth let them compete head‑to‑head with incumbents like Microsoft. Rao details Abridge’s technical stack—from multilingual, medical-grade ASR to orchestration of models for documentation, billing, and patient summaries—and how feedback at scale fuels continuous preference tuning. He also outlines the broader roadmap: automating orders, revenue cycle, clinical trial matching, and ultimately clinical decision support, while emphasizing the enduring centrality of human clinicians and patient agency.
IDEAS WORTH REMEMBERING
7 ideasTarget high-frequency, lower-stakes workflows first to unlock adoption.
By focusing on documentation rather than direct diagnosis or treatment, Abridge found a high-impact wedge where clinicians remain in the loop, reducing perceived risk and accelerating enterprise adoption.
Depth in domain-specific models matters even when generic APIs exist.
Medical conversations require recognizing mispronounced drugs, specialty-specific jargon, and multiple languages; small error-rate differences in speech recognition can meaningfully affect both care quality and billing accuracy.
Enterprise healthcare sales hinge on simultaneously convincing CMIOs, CIOs, and CFOs.
Abridge framed its value as better clinician experience and quality (CMIO), robust integration and reliability (CIO), and measurable revenue and billing uplift (CFO), aiming to satisfy at least two of three to close initial deals.
Trust and ecosystem integration are as critical as raw model performance.
Becoming ‘core infrastructure’ for health systems required deep integrations (e.g., Epic), demonstrable reliability, and alignment with existing tech stacks, enabling Abridge to inherit and amplify institutional trust.
Real-world edits are a powerful training signal for continuous improvement.
Millions of AI-generated notes are corrected by clinicians, providing rich feedback for preference tuning (DPO, reward modeling, RL), making the system progressively ‘less imperfect’ in specialty- and institution-specific ways.
Conversations are a foundational signal that can drive many downstream workflows.
Starting from the clinician–patient dialogue, Abridge can generate notes, orders, claims, after-visit summaries, trial matches, and eventually decision-support suggestions, all from the same upstream signal.
Clinician and patient agency are central to meaningful healthcare AI.
Rao frames success not just as productivity gains, but as restoring control over time for clinicians and making patients ‘main characters’ in their care via clearer documentation and accessible summaries.
WORDS WORTH SAVING
5 quotesWe’re not compensated as doctors for the care that we deliver. We’re compensated for the care that we documented that we deliver.
— Shiv Rao
We decided to run into the hardest part of the market… the barrier to ‘good enough’ is really, really high.
— Shiv Rao
If we go down, the entire health system goes down. They’re not making money anymore because these notes are essentially bills.
— Shiv Rao
No technology has ever done this in healthcare… we’re reducing cognitive burden by like 60% within six weeks and burnout by about 50%.
— Shiv Rao
I literally took my phone out and explained to him that Abridge is a new tool that lets Mommy come home early and eat dinner with her family.
— Anonymous physician user (as recounted by Shiv Rao)
QUESTIONS ANSWERED IN THIS EPISODE
5 questionsHow should regulators and health systems evaluate and certify AI tools that affect both clinical quality and billing, given the high financial and patient-safety stakes?
Abridge founder and CEO (and practicing cardiologist) Shiv Rao explains how the company uses specialized speech recognition and generative AI to turn clinician–patient conversations into high‑quality clinical and billable notes in seconds, dramatically reducing documentation burden. He describes the post‑pandemic burnout crisis, why Abridge chose the hardest enterprise health-system segment first, and how trust, integrations, and research depth let them compete head‑to‑head with incumbents like Microsoft. Rao details Abridge’s technical stack—from multilingual, medical-grade ASR to orchestration of models for documentation, billing, and patient summaries—and how feedback at scale fuels continuous preference tuning. He also outlines the broader roadmap: automating orders, revenue cycle, clinical trial matching, and ultimately clinical decision support, while emphasizing the enduring centrality of human clinicians and patient agency.
What guardrails are needed to ensure AI-generated documentation and decision support don’t entrench existing biases in diagnosis and treatment patterns?
As AI begins to suggest orders and differential diagnoses, how does the responsibility and liability balance shift between clinicians, vendors, and institutions?
In what ways could giving patients more direct access to AI-explained notes and summaries change the doctor–patient power dynamic—for better or worse?
How might the widespread adoption of tools like Abridge reshape medical training, clinical workflows, and the skills future clinicians are expected to master?
EVERY SPOKEN WORD
Install uListen for AI-powered chat & search across the full episode — Get Full Transcript
Get more out of YouTube videos.
High quality summaries for YouTube videos. Accurate transcripts to search & find moments. Powered by ChatGPT & Claude AI.
Add to Chrome