Data Assets Power Clinical Intelligence

Diving deeper into

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

Interview
the real unlock is also the data assets
Analyzed 5 sources

The data layer is what can turn an AI scribe from a labor saving tool into a much larger clinical intelligence business. Time savings gets Knowtex into the workflow, but the bigger prize is capturing the full visit record, what was in the chart, what the patient and doctor said, and what the clinician ultimately signed. That creates a richer dataset for trial matching, decision support, and eventually real world evidence products that can be sold to research and pharma buyers, not just care delivery teams.1 2 3 4

  • The immediate buyer case is still operational ROI. Knowtex users prepare charts, record the visit, review drafted notes, codes, and orders, then sign. In one oncology study, users saw 3.35 more visits per day and $433.61 more billed per working day on average, with those gains flowing to the practice rather than to Knowtex directly.2
  • What makes the dataset valuable is that it captures details normal records miss, like symptom changes, treatment tolerance, and why a therapy stopped. Knowtex describes this as a full clinical reasoning record, because it links EHR context, live conversation, and the clinician decision in one signed workflow.1 2
  • This matters because basic scribing is becoming a bundled feature. Abridge, Ambience, Freed, Heidi, Epic, Oracle, Doximity, and OpenEvidence are all pushing beyond note writing into coding, orders, search, and other workflows. In that market, proprietary visit level data is one of the few assets that can compound over time instead of getting priced away.3 4 5

The next phase is a split market. Basic documentation will keep getting cheaper and more bundled, while companies with permissioned, high quality clinical data will move upward into trial operations, shared clinical brain products, and pharma research. If Knowtex can keep winning workflow adoption and data rights, the software seat becomes the collection point for a much larger healthcare data business.2 3 4