Auditing what was said: the epistemic promise and limits of ambient AI in clinical practice

Traditional audits based on written charts may overlook aspects of clinical reasoning that shape patient care. Ambient artificial intelligence enables systematic analysis of clinician–patient dialogue at scale. We report a hypothesis-generating analysis of 124 single-centre urology consultations. A transcript-centred audit revealed inter-physician variation and expert disagreement that were not captured by chart review alone. We examine its epistemic value alongside its nonverbal, behavioural, technical, and regulatory limitations.