Apple Vision Pro surgery, the ChatGPT front door, and concierge care marked CHIME’s San Diego health innovation summit

CHIME held its Innovation Summit in late May, hosted at Sharp HealthCare in San Diego. Like the other regional summits CHIME hosts around the country, it featured a day and a half of fishbowl panels, where audience members rotate into open chairs to join discussions led by CIOs.

Much of the conversation during the San Diego summit mirrored the questions occupying healthcare more broadly in 2026. Leaders weighed buying innovation against building the capacity to create it, returned repeatedly to defining the problem before reaching for a tool, and debated the governance needed to scale a pilot. AI ran through every session, usually framed as a way to lift low-value tasks off clinicians rather than replace them.

Against this backdrop, the summit’s discourse included many highlights and surprises. The following four stood out the most.

Apple Vision Pro is being used in live cataract surgery at Sharp HealthCare

One of the most talked-about sessions at the summit came from Sharp HealthCare’s chief spatial computing officer, Tommy Korn, MD, an ophthalmologist who still performs surgery. Korn walked the room through a study Sharp launched in April testing Apple’s Vision Pro headset during real cataract surgeries using the ClearSurgery app. This is one of the first IRB-approved studies that puts Apple’s Vision Pro headset to work in actual cataract operations. Instead of looking through a traditional microscope, surgeons view a magnified image of the surgical field and patient information directly inside the headset. A standard microscope remains available as a backup if needed.

The main goal of the study was to address the physical strain cataract surgeons experience during their careers. Cataract surgeons spend years hunched into a fixed microscope position, and the neck and back strain that follows pushes experienced surgeons toward early retirement while demand for the procedure keeps climbing. He also walked through the economics, noting that the headset costs far less than the stacked monitors and microscope optics it could eventually replace.

Sharp designed the study to answer two basic questions: can surgeons operate while wearing the headset, and is it safe for patients? The work took place at a community outpatient surgery center— where routine, same-day procedures are done—rather than a university teaching hospital, and the surgeons handled patient consent themselves. The early results show surgeons can operate through the headset without putting patients at greater risk, and that it puts less strain on their bodies.

Patients are using ChatGPT before they see a doctor

On a panel about AI in primary care, leaders had to face the impact of ChatGPT on their practices; namely, that patients were seeking help from ChatGPT and skipping primary care altogether. One attendee described how his wife treated a worsening spot on her face by sending a photo to ChatGPT, following its suggestion, and watching it clear before a vacation rather than having to wait for an appointment weeks away. Another speaker sketched the path he expects to become routine, where a patient with a PPO plan and reams of wearable data, believing they need a specialist, simply tells ChatGPT their ZIP code and insurance information and asks it where to go.

Increasingly, patients with primary care appointments running weeks or months out are taking their symptoms, lab results, and wearable data to consumer chatbots, and then deciding on their own whether they need the system at all.

The obvious risk here is that a general-purpose model can be confidently wrong. Several leaders converged on the same solution: a health-system-provisioned AI that keeps a clinician in the loop for anything above low risk and routes patients back into safe, in-network care.

One health system is rethinking primary care around AI triage

The most provocative idea of the two days came from a Bay Area health system’s CIO, who said the math on primary care no longer works. The economics are upside down: a primary care visit costs more to staff and deliver than payers reimburse for it, and there aren’t enough physicians to see everyone who wants an appointment. Primary care loses money for many systems, and younger patients in particular will not tolerate a multi-week wait for a familiar face. Her team is weighing whether to stop funneling those patients into a traditional primary care relationship at all. An urgent care visit, supported by AI triage, would handle the immediate need, a less expensive nurse or care navigator would manage anyone with chronic disease, and specialists would pick up the rest.

The idea drew objections, including the counterpoint that pay-as-you-go convenience care widens gaps for the patients who most need a steady relationship with healthcare providers. One exchange reframed the conversation. Concierge medicine already delivers same-day access to the patients who can pay for it, which means the system knows how to provide it. Fast access is not scarce because hospitals lack the ability to deliver it, but because the economics reward delivering it only to those who pay extra.

Healthcare’s biggest AI wins are operational, not diagnostic

For all the talk of AI replacing doctors, the real deployments leaders bragged about were operational. One academic system described an AI voice agent that handles pre-procedure prep calls for colonoscopies, the reminder work staff never had time to finish. Because the agent could call everyone rather than a fraction of the list, the system reached far more patients and cut same-day cancellations sharply while freeing staff to shift to more complex cases.

Another system reported keeping roughly a fifth of its two million annual outpatient visits by telehealth well after the pandemic, then adding on-demand virtual primary care seven days a week, with average wait times under ten minutes.

The story that landed hardest was about an IT leader who had spent months surfacing efficiency data and getting no change in behavior to show for it. They shared an example of a clinician who simply texted a colleague “about to scrub in,” allowing the team to prep the next patient and save roughly thirteen minutes per procedure. This in turn added up to about $1.9 million for the hospital.

When the IT leader was playing the video game Fortnite, he noticed that the game’s on-screen scoreboard let players see what the top performers did and copy it. He rebuilt his reports the same way, highlighting high performers and the small tweaks that worked for them, rather than flagging who was behind. Buy-in soon followed, with everyone trying to match the best.