Knowtex Building AI Layer Over EHRs

Diving deeper into

Caroline Zhang, CEO of Knowtex, on going from AI medical scribe to AI lab

Interview
an AI-native partner wins on both fronts
Analyzed 4 sources

The real wedge is not note taking, it is shipping new clinical tasks faster than the EHR can productize them. Knowtex is trying to become the fast moving AI layer on top of slow moving record systems, so a doctor can get chart prep, ambient documentation, coding, order drafts, and trial matching in one workflow instead of waiting through multiyear EHR roadmaps and upgrade cycles.1 2 3 4

  • In practice, this means the product is competing on how many clicks it removes after the visit, not just on transcript quality. Knowtex describes a workflow where the doctor starts with previsit chart prep, records the visit, then gets notes, codes, orders, and follow up tasks within about 30 seconds.1
  • The strongest comparable is Abridge, which won by going deep with Epic rather than replacing Epic. That path produced scale, with 60,000 plus clinicians across 100 plus health systems by May 2025, but it also shows the tradeoff, because the closest EHR partners gain distribution while losing room to challenge the EHR itself.3 4
  • The market structure favors specialists over general overlays. Healthcare integrations are deep, specialty specific, and hard to standardize, which is why buyers often choose the tool that completes jobs like diagnosis entry, orders, and billing steps inside their existing EHR, not the tool with the prettiest standalone scribe.2 4

This points toward AI scribes becoming broader clinical operating layers. The winners are likely to be the companies that turn ambient conversation into downstream actions and better decision support, while staying interoperable enough to sit above Epic, Oracle, and the long tail of specialty systems without being reduced to a bundled feature.1 2 3 4