EHRs Fight to Own AI Scribes

Diving deeper into

Brendan Keeler, interoperability lead at HTD Health, on GTM for AI medical scribes

Interview
EHRs are fighting top down to own the AI scribe experience
Analyzed 11 sources

The real prize is not note taking, it is control of the clinical workflow that starts with the doctor and patient talking and ends with codes, orders, billing, and follow up inside the EHR. That is why Epic has pulled ambient documentation into its own ecosystem through unusually tight partnerships with Abridge and Nuance, while Oracle Health is building native ambient and order creation directly into Cerner workflows. Once the scribe is embedded at that layer, independent vendors stop being wedges and start being add ons.

  • Epic has already made this concrete. Abridge became Epic’s first Pal in August 2023, with documentation and structured data flowing back into Epic workflows. Nuance also launched DAX Express for Epic in June 2023. In practice, large health systems choosing an ambient scribe inside Epic are often choosing between Epic approved options, not an open field.
  • The distribution advantage is massive. Abridge used the Epic relationship to expand from about 8,000 clinicians to more than 60,000 across 100 plus health systems in 18 months, reaching about $100M ARR by May 2025. That growth came from going deep inside the dominant enterprise workflow, not from replacing the EHR.
  • Bottom up tools like Freed can grow quickly with small practices because a doctor can buy a $99 per month tool without a big IT process. But once a group wants deep EHR actions like diagnoses, orders, and billing fields filled in, the sale shifts to IT, compliance, and enterprise contracting. That is where EHR aligned vendors pull away.

The next phase is expansion from note generation into order entry, coding, prior authorization, nursing, and emergency medicine. That pushes ambient AI closer to the system of record rather than a standalone app. The winners will be the vendors that either own the EHR or are trusted enough by the EHR to automate more of the visit without forcing clinicians to leave their main workflow.