Knowtex turns clinicians' playbooks into software

Diving deeper into

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

Interview
Traditionally, every doctor relies on their own memory, years of experience, or public guidelines to provide care.
Analyzed 5 sources

This is the real opening for Knowtex, because medicine still runs partly like a craft business where each doctor carries a private playbook in their head. Knowtex is trying to turn that hidden playbook into software by combining chart data, visit conversations, and clinician approved outputs so a doctor can see how similar patients were treated across a de-identified network, not just what a textbook says.1 2 3

  • Older clinical decision support mostly means public guidelines, drug references, or a second opinion from a colleague. Knowtex is aiming at a narrower and more practical question, what doctors actually did for a patient with this diagnosis, treatment history, lab pattern, and demographic profile, using its aggregated encounter data.1 2
  • This only works if the product lives inside the full visit workflow. Knowtex preps the chart, drafts the note, suggests codes and orders, and flags trial eligibility. That gives it access to richer context than a basic scribe, including history before the visit and actions taken after it.2 3
  • The broader category is moving the same way. Abridge, Ambience, Freed, and others started with note taking, but competition has pushed them toward deeper administrative work and clinical assistance. In that market, shared clinical intelligence is one of the clearest ways to become more than a commodity scribe feature.3 4 5

The next step is a split market. Simple note generation gets bundled into EHRs and free physician tools, while the winning independent products move up into decision support, order workflow, and trial matching. If Knowtex can prove better care decisions from its network data, it stops being just documentation software and becomes operating software for clinical judgment.2 3 5