251: Rochester Regional's Dr. Everett Weiss: Rethinking Time Saved as the Main AI Success Metric, Why 100% Ambient Adoption Is the Wrong Goal, and How to Fund a Project When the ROI Is Soft

Subscribe on: RSS | SPOTIFY | APPLE PODCAST | GOOGLE | BREAKER | ANCHOR

On this episode of The Digital Patient, Dr. Joshua Liu, Co-founder & CEO of SeamlessMD, and colleague, Alan Sardana, chat with Everett Weiss, MD, Medical Director, Health Informatics at Rochester Regional Health, about "Rethinking Time Saved as the Main AI Success Metric, Why 100% Ambient Adoption Is the Wrong Goal, How to Fund a Project When the ROI Is Soft, and more..."

Audio:

Guest(s):

• Everett Weiss, MD, Medical Director, Health Informatics at Rochester Regional Health
• Joshua Liu, MD (@joshuapliu), Co-founder & CEO at SeamlessMD

Episode 251 - Show Notes:

[00:06:20] What sitting on every side of the healthcare table teaches a leader — Everett left clinical practice for informatics when the profession was still being defined, and his path took him across vendor and health system roles. The lesson, he says, is that it's "not always just about getting stuff done," but how you get there: partnerships and collaboration matter more than transactions. Goals across organizations, and even across departments within one organization, aren't always aligned, and "risk is not always shared," so asking questions and bringing empathy to every conversation is key to moving things forward.

[00:08:55] Why effective CMIO leadership is about listening, not politics — Everett doesn't love the word politics. To him, effective leadership means sitting across the table from someone, having a meaningful conversation, and understanding what is getting in the way of their work, whether they are a physician, APP, nurse, front desk staff, or back office employee.

[00:11:03] Why time saved may be the wrong metric for AI draft responses to patient messages — AI-generated portal replies are empathetic and complete, and patients respond to them, but someone on the care team has to proofread, review, and edit those lengthy drafts, adding burden. Everett frames it as a balance between "who is doing the work and who is deriving the benefit," asking whether more burden for the healthcare team is acceptable when the patient benefits.

[00:13:31] How physicians rewriting ambient AI notes raises a question about where time savings should go — a subset of physicians use their ambient scribe time savings to rewrite AI drafts to reflect their own personality, tone, and style. Everett asks whether that helps patients, and suggests redirecting some of that time toward reviewing AI draft responses to patient messages, where the net time is equal but the patient impact is "pretty tremendous." He notes AI tools are now learning to match a provider's previous notes, which may solve the problem anyway.

[00:15:32] Why a B-plus clinical note might be good enough — borrowing an idea from a speaker at an Epic UGM conference about where to be A-plus and where B-plus is acceptable, Everett asks whether a note that is complete, accurate, unbiased, and representative of the visit is enough, even without the physician's personal style, if the priority is "showing up and being genuinely present for that patient."

[00:19:12] Why leaders earn more respect by saying no — Everett believes leaders are better respected when they say no, especially "with empathy and with reason." Trust is built on delivering commitments, and organizations that commit to 10 priorities and work on them in parallel often delay or miss all of them, losing the trust of the people counting on them. Choosing three or four priorities you're confident in moves the organization further toward its strategic goals.

[00:21:27] What alignment looks like when not everyone agrees on priorities — Everett says it's not always about consensus. What matters is agreement on the three or four things the organization will move forward on, plus the ability to articulate the why behind them so people follow.

[00:22:55] How leaders mistake AI deployment for problem solving — if the metric is how many AI tools were deployed, that's easy to hit and report to a board. If the metric is whether you meaningfully advanced a strategic objective, the question becomes which AI solutions actually solve a problem. "Doing more... doesn't mean affecting positive change. It's like running in place," and "an organization really shouldn't be a treadmill."

[00:24:53] How to make the ambient AI business case to a CFO — Everett says it's not easy. If the goal is margin, ambient digital scribes may not be the answer, although he notes there is margin in them. If the goal is reducing burnout, there's a case. Downstream benefits translate into real dollars but aren't hard ROI: higher quality care, more complete continuity of care, increased patient satisfaction, patients more likely to stay with the health system, and better provider retention and delayed retirement.

[00:27:04] Why bundling technology deployments can fund human-centered priorities — since most conversations come back to "how are we gonna pay for it," Everett recommends discounting soft ROI expectations to keep them realistic and looking at the full portfolio of work. Pairing a revenue-generating deployment with one that is revenue neutral or a cost, but delivers value, can keep the combined case net positive.

[00:30:35] Why ambient AI scribes may help patients stay with a health system — Everett's organization chose to tell patients and ask for consent even where one-party consent rules don't require it. Because notes are drafted quickly, patients see a complete recap in the portal right away instead of a week later, and can share it with family and caregivers and better manage their own health. Patients who say they're "having a phenomenal experience" are more likely to stay and to attract others to the system.

[00:33:28] What an optional ambient AI rollout revealed about how clinicians value it — the ambient scribe deployment wasn't mandated, it was offered as a risk-free tool, which Everett says removes much of the resistance that comes with mandated deployments. About nine in ten clinicians responded positively, but value varied. Many went home earlier. Others spent the time saved on patient-related work that wasn't getting done before. Some actually spent more time documenting but valued the reduced cognitive burden and "returned some of that joy to practicing medicine." A few were already so productive it added nothing, which is why 100% adoption was never a goal: "don't fix a problem that doesn't exist."

[00:37:38] Why technically excellent AI can still be wrong for healthcare — medicine is about improving health and about human connection, which Everett calls both a means and an end. AI can replace clinicians and output probabilistic information, or it can assist, validate their thinking, and promote more human interaction. AI that removes the human element, or that distracts from an organization's real priorities, is "fundamentally wrong for healthcare." AI shouldn't be "in search of a problem."

[00:40:28] Why a shrinking workforce means healthcare needs AI — Everett agrees that with a diminishing workforce, AI must take on some current jobs so people can focus on more cognitively complex and human connection work, and so organizations can increase access far beyond today's levels. Doing it for those reasons moves healthcare "in the right direction."

[00:41:50] How "micro sunk costs" help leaders know when to pivot on multi-year projects — commitment to a strategy is essential, but leading indicators show whether you're on track. Being off track may call for a pivot, or for stopping entirely. Rather than fixating on the sunk cost of a whole project, Everett accepts "micro sunk costs" when leading indicators show something won't work, then switches to work with more potential value.

Fast 5 Lightning Round:

1. What is your favorite book, or the book you've gifted the most?

The Captain Class by Sam Walker, Algorithms to Live By by Brian Christian and Tom Griffiths, Unreasonable Hospitality by Will Guidara, The Gifts of Imperfection by Brené Brown, and Be Ready When the Luck Happens by Ina Garten

2. If you could instantly master any skill, what would it be?

"Cooking"

3. Would you rather have super strength, super speed, or the ability to read people's minds?

"Mind reading"

4. What is something in healthcare you believe that others might find insane?

"After decades of trying, we still can't figure out how to reconcile problem lists"

5. What is the last movie or TV show you saw, and what did you think of it?

"The Blacklist. Brilliant writing, and I loved the ending."

The Digital Patient has been recognized as Feedspot's #1 Patient Engagement Podcast of 2025. Thank you to our listeners for making this happen!