On March 20, 2023, Nuance Communications - by then a Microsoft company - and Microsoft announced Dragon Ambient eXperience (DAX) Express, describing it as a fully automated clinical-documentation application and the first to combine proven conversational and ambient AI with OpenAI’s then-newest model, GPT-4. The product targeted one of the most-cited drivers of physician burnout: the hours doctors spend typing notes into electronic health records after, and during, patient visits.
“Ambient” documentation means the software listens to the natural conversation between a clinician and patient in the exam room or over telehealth, then drafts a structured clinical note from it automatically, in seconds, for the clinician to review and sign. Earlier versions of DAX relied partly on human scribes reviewing transcripts behind the scenes; pairing ambient speech capture with a large language model was the step that made fully automated drafting plausible.
The announcement was an early, concrete deployment of GPT-4 in a regulated, high-stakes professional workflow, only days after GPT-4 itself was released. It pointed to where generative AI was likely to land first in medicine: not autonomous diagnosis, but the administrative burden around care - paperwork a model can draft and a human can check.
The caveat is the same one that shadows every generative system in healthcare. A model that drafts a clinical note can also introduce errors or invent details that were never said, so the clinician’s review is not a formality but the safeguard. The value proposition - give clinicians time back by automating documentation - depends on that human check holding.