The first wave of clinical AI was scribes — tools that listened to a consult and produced a structured note. Useful, often impressive, but ultimately janitorial: they cleaned up the paperwork, they didn't change the medicine.
What the next wave looks like
Evidence-backed AI listens to the same consult and surfaces, in real time, the guideline the clinician should be checking against. It doesn't make the decision — it makes sure the clinician has the right reference in front of them at the moment of the decision.
The mental model we use internally: a scribe is an output tool, a co-pilot is a thinking partner. The first speeds up writing; the second changes the conversation.
What clinicians do with the extra brain-space
Eight months in, the pattern is consistent: clinicians don't pack in more consults — they listen harder. The patient-rated 'I felt heard' score is the leading indicator we watch most carefully, and it's the one moving most.