Lee Health and the AI that doesn't diagnose but listens: the real hospital revolution begins in the paperwork

By Florida Weekly · July 12, 2026.

Lee Health, a hospital system in southwest Florida, hosted a public event at its Coconut Point campus called "The Intelligence Inside," with the stated goal of explaining to patients and the community how it is using artificial intelligence in its clinical operations. The piece, signed by the system's own CEO, Larry Antonucci, and featuring remarks from Wendy Victor, director of AI and digital transformation, describes three concrete applications: an "AI Scribe" that, with the patient's consent, listens to the consultation, transcribes it in real time and generates the structured clinical note; radiology tools that analyze imaging studies in the background to flag as urgent the cases that require it, in a context where a radiologist may review up to 100 studies a day; and digital whiteboards in the rooms of Golisano Children's Hospital that pull data from the electronic health record and offer support in Spanish and Haitian Creole. It was rounded out with an "AI Playground" of demonstrations for employees and visitors.

It is worth placing this piece in its genre: it is an opinion article signed by the top official of the hospital itself, published in a regional weekly, with no performance figures, no external audit and no data on errors, real time savings or adoption among staff. It is, in essence, institutional communication of trust toward the community, not a clinical study or an independent evaluation. That does not invalidate the content, but it does require reading it for what it is: the version the health system wants the public to understand, not an objective measurement of results.

That said, the pattern it describes is real and is repeated across the United States: the AI that is actually being deployed in hospitals today is not the kind that diagnoses cancer on its own, but the kind that eases administrative and cognitive burden. Ambient scribes (tools from the Nuance DAX, Abridge or Ambience Healthcare family) have spread precisely because they attack the number one documented cause of physician burnout in the profession: the late-night hours spent filling out records instead of being with family or resting. And radiology triage systems that prioritize urgent studies —already with FDA clearances for detecting stroke, pulmonary embolism or intracranial hemorrhage— do not replace the radiologist's judgment, but they do change the order in which that judgment is applied, which in emergency medicine can be the difference between a timely diagnosis and a late one.

This fits with something we have been arguing about AI's impact across sectors: in healthcare, the first thing to fall to automation is administrative friction —coding, notes, triage of work queues—, while direct care, the patient relationship and clinical judgment in the face of uncertainty hold up better and for longer. Lee Health fits that pattern with almost textbook precision: it automates the paperwork and the priority queue, and leaves intact —for now— the conversation between doctor and patient, which it in fact claims to strengthen by freeing up the physician's time.

The real short-term risk here is not technological but one of trust and governance: what happens to the transcripts of sensitive medical conversations, who audits the false negatives of an image triage system, and whether the transparency toward the community that Lee Health says it prioritizes translates into real accountability mechanisms or remains a well-intentioned public relations event. That the system itself acknowledges the need to "build trust" is an honest sign that this trust is not guaranteed by default, and that the healthcare sector —more than any other— cannot afford to sell promises that no verifiable data later backs up.

In the long run, however, it is precisely this kind of modest, functional deployment —not the headline of medical superintelligence— that builds the ground on which AI's greater promise in healthcare will rest: faster and more consistent diagnoses, fewer errors from human overload, and clinicians with more real time to think about the patient instead of the keyboard. Eradicating diseases and prolonging healthy lives does not start in a superintelligence lab, but in mid-sized hospitals like Lee Health learning, step by step and with data along the way, to delegate to the machine what does not require a human to be done well.