AI That Fits the 30-Minute Visit

Diving deeper into

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

Interview
The missing piece is how that works in the real-time, game-time clinical workflow
Analyzed 5 sources

The real moat in clinical AI is not answering medical questions, it is fitting into the 30 minutes when a doctor is actually seeing a patient. Knowtex is built around that moment. Before the visit it pulls labs, scans, pathology, and prior notes into a prep view. During the visit it captures the conversation. Right after, it drafts notes, codes, orders, and trial follow up, which is the last mile that consumer style tools do not handle inside hospital systems.1 2 3

  • Tools like ChatGPT and OpenEvidence have helped by teaching clinicians that fast, natural language software is possible. But OpenEvidence spread first because it was free and did not touch patient records, while hospital grade products still need security review, EHR access, and workflow approval.1 3 4
  • In practice, real time workflow means a doctor does not copy data across tabs or finish charting at night. Knowtex says its system handles chart prep, ambient capture, then outputs notes, billing codes, orders, and care coordination in about 30 seconds after the visit ends.1 2
  • That is also why the market is moving past standalone scribes. Abridge, Ambience, and Freed have all expanded from note taking into coding, revenue cycle, and adjacent workflow products, because the note alone is becoming a bundled feature rather than the whole product.2 5

The category is heading toward clinical operating systems that sit on top of the EHR rather than simple AI note tools. The winners are likely to be the products that can turn consumer driven interest in AI into approved, secure, always on workflow software inside the exam room, where patient context, conversation, and next actions all come together.1 2 3 4