Rohit: Hi Sara. Welcome to the Big Unlock podcast. It is great to have you here. 

Sara: Thank you so much for having me. I’m excited for the conversation. 

Rohit: We are quite looking forward to it. So we will start with some very quick introductions and then dive right into the topic our audience is waiting to hear. I’m Rohit Mahajan. I’m the Managing Partner and CEO at BigRio and Damo Consulting. I will request Ritu to do the intro, and then over to you, Sara.

Ritu: Hi Sara, welcome to our Big Unlock podcast. We are thrilled to have you on the podcast and looking forward to a very insightful and engaging conversation. My name is Ritu Uberoy and I’m the Managing Partner here at BigRio and Damo Consulting, and also serve as the host for the Big Unlock podcast. I’m currently based out of Gurugram, India, and I travel frequently to the US. Really excited to have this conversation because we were in the Bay Area yesterday and every second billboard was an AI billboard—so you know this is what’s happening right now.

Sara: Absolutely. I’m so happy to be here with both of you. Just as a quick introduction for the folks who are listening—I’m Sara Vaezy. I serve as the Chief Transformation Officer at Providence, and I’ve been with Providence for about nine and a half years in total. The transformation role is relatively new for me.

I have dual responsibilities. I have responsibility for the consumer-facing, patient-facing front end for the organization. We serve 5 million patients a year. So how we do our brand, our marketing, our digital consumer experiences to really meet the needs of our communities—ensure that they can find our services, that we can bring them in, that we can retain them—that’s a big part of what we do across channels. Whether it’s web, mobile, proactive outreach, phone—all those types of connected digital experiences.

I also have in parallel this newer responsibility around transformation, which started out as a bit of a nights and weekends project back when GPT became publicly available. I had a feeling there would be a really important impact on our industry and that we needed to lean in. That morphed into a formal role around identifying and being responsible for driving AI adoption—responsible AI adoption—to get maximum value for it in our system. Do so in a way that keeps our patients safe, keeps our caregivers— we call all the folks that work at Providence caregivers—safe.

That’s been the undertaking over the last several months. We’re going to be really focused and drive it across a couple of specific domains. We’re not just going to do it spray and pray, but really focused.

That’s what I’m responsible for. I’ve been in healthcare for a long time. Prior to coming to Providence, I was in management consulting. Before that, I worked in health policy and health services research. I started out my career as a research scientist in med tech. Healthcare is something I love and is very near and dear to me. There are a whole lot of problems for us to solve. So the job is never done—and it’s a lot of fun.

Rohit: Absolutely. The landscape is changing so fast that it’s hard to keep up, but there are a lot of very interesting innovations coming our way.
So how do you think about the consumer-facing apps, and how does it make a big difference to the patient’s experience and care?

Sara: It’s an interesting question, and you’ve separated consumers and patients—we tend to do that as well. Actually, I kind of don’t like the term “consumer” either because it has the implication that you have the means to participate in the market or to buy services. That’s not really what we mean. What we mean is that each person has agency over their own healthcare, their own decisions, and we want to support that.

What we’ve really focused on is a few different things. The first is that we really need to make finding our services and transacting with us—like booking an appointment, paying your bill, getting financial assistance or counseling—a lot easier. Even things like just creating an account. A lot has changed with the electronic medical record companies too, making it easier to create accounts.

But back in the day, you needed a special code, or you needed to have a clinic visit already, and there were just lots of barriers to doing things easily. We’re really removing those barriers as much as we can—making it easy to find what you need, whether it’s a physician, content, or something else. Taking out the friction, making it easier—that’s job one.

We’ve got some pretty lofty goals. Things like online scheduling are still a battle for us. But we aim to make sure anything that is bookable has the option to be booked online. That doesn’t mean everybody will exercise that option, but it means it’s available to them. It’s just like booking a flight or hotel—there shouldn’t be secret sources of data. Everything needs to be available from the same source of truth and be frictionless.

The second piece we’re focused on is that each person is different. If we really want to keep people engaged, we need to recognize how important personalization is. I have a different experience than Rohit, who has a different experience than someone else. We live in different parts of the world. We have different healthcare needs. I have a six-year-old son who is a proxy on my account. All of these things are different. We need to deliver different experiences, and we know that makes a big difference in terms of keeping people engaged.

Rohit: Especially because you said you have 5 million patients that you’re serving. That’s such a large number. There must be so much diversity in that, and it must be spanning many different. 

Sara: Exactly. Many different states. We’re up and down seven states in the Western United States—Washington, Oregon, California, Texas, New Mexico, Montana, and Alaska. Very different states. Some have big urban areas like Los Angeles and Seattle, and some are very rural, like northern Alaska. It’s a very diverse footprint—different types of services, people, and expectations.

The last piece we’re really focused on from a consumer perspective is the expectation piece. We have to engage with folks in the way they’re used to engaging. I know we’ll get to a topic around AI and what we’re seeing there, but conversational platforms are big. People want to do things synchronously but not necessarily get on the phone. Helping them navigate in real life is another big area we’re focused on from a consumer perspective.

So—personalization, frictionless experience, and helping people navigate—those are the big ones. And they apply to everything—from marketing campaigns, to what people see on our web and mobile, to their experience when they call into our contact centers.

Rohit: That’s awesome. And Sara, how do you approach patient engagement and consumer engagement with product development? I know you have a very thoughtful way of doing these things. What’s some of the secret sauce you can share that leads to success in this area? 

Ritu: Yeah and that would be a great lead into like giving us a success story maybe around this topic. 

Sara: Absolutely. I can do that and I’ll give you a story that kind of bridges the gap into the AI work that we’ve been doing as well. The main difference between traditional sort of just how you might turn on features and just try to make an experience ever so slightly better versus a product development approach, which is like think about things end to end.

It’s not just about turning on a feature to be able to book an appointment, for instance. How do you actually even know, like how do you get directed to the right care in the first place as an individual? How do you make sure that care is discoverable, that it’s appropriate, that it really meets a person’s needs based on their intent and motivation and preference and their clinical needs? All of that is its very data driven. It’s very precise, not perfect, but it’s a little bit more precise than just saying, ‘we’re going to slap up an option of booking an appointment.’ Booking an appointment is a feature, a product development approach which is really seeing that whole customer experience end to end.

It supports the experience from discovery through delivery—for instance, the booking experience. We’ve done a lot of product development to facilitate that end-to-end experience. One example—and then we can extrapolate it to a bigger one—is we built a conversation and navigation platform that helps our patients get their needs met without having to message their physician.

That’s an important problem to solve. We have six to seven million patient-generated messages annually that go to our physicians. These go into what’s called the physician’s in-basket, and they have to be managed somehow. That’s a huge volume. So what we said was, let’s take a different approach. Let’s not just ask, “How do we manage the message?”—which is a proximal problem—but go upstream and ask, “Why are patients sending a message in the first place?”

We realized they either can’t find the content they need, or they can’t complete the task they’re trying to complete. For example, with appointment booking or financial counseling, they might not understand their bill or may need help paying. So they end up sending a message.

But what if we could understand what even a complex message is asking, use intent recognition, and then activate agents to fulfill that task? We’ll get to that AI agent world in a moment. But this approach helps patients get their needs met immediately, instead of waiting 24 to 48 hours for a response.

We’ve seen what we call “message deflection” up to 30%, and we’re aiming to deflect approximately 2 million messages annually over the next couple of years. That’s a massive impact on patient experience. And we couldn’t have done it if we just narrowly thought about reducing messages. It’s not just about that—it’s about the full experience. That’s where product development comes in.

Rohit: That’s very cool to know, Sara. Could you tell us a little more—for those in the audience who don’t know—what is “message deflection”? 

Sara: It’s a bit of a challenging concept. We’re using data science methods to say, “This message would have otherwise been sent, but it wasn’t—because the patient got their need met another way.” So we avoided or “deflected” the message.

Simply put, our patients interact with a chatbot. The chatbot walks them through the experience in a natural, conversational way. Instead of sending a message to their physician, they get their need met right then and there. We can understand about 90% of what patients are telling us. We call it “goal conversion”—and we’re able to meet their need 50 to 60% of the time.

Ritu: So you’re freeing up physicians’ inboxes. And like you said, patients don’t have to wait 24–48 hours, which creates frustration for everyone. If the chatbot can understand the intent and serve them right there—that’s a powerful concept. Thank you, Sara.

Rohit: And Sara, you mentioned agents—AI agents. Let’s get to that. We’re talking about a digital workforce transformation as well. What are your thoughts? What are you seeing down the pike? Agents are already becoming mainstream, especially voice agents. You also mentioned domain-specific implementations—what are your thoughts around those?

Sara: On the agent front, things are moving really quickly. If you had asked me this six months ago, I would’ve said, “Yes, agents are going to be very material, and a digital workforce is critical.” I still think that’s true, but we shouldn’t limit ourselves to only taking what humans currently do and asking computers to do it.

That’s not an optimized view of the possibilities. We can be more creative. A digital workforce doesn’t get sick, works 24/7—that’s great. But baked into that assumption is the idea that there’s no redesign of the process—they’re just running the same thing. I think we have to combine agents with actual process redesign.

It’s not just about automating dull or low-value work. It’s about doing things better. We don’t want to just automate bad processes. We want to improve them. That’s what I sometimes struggle with—I want to ensure that, as an industry, we’re reflecting on this deeply. It’s not just substitution—we need to think about it more materially.

Another thing is, there are tasks humans simply can’t do that we should be imagining as algorithm- or computer-powered. For example, when we do a marketing campaign, we look at our footprint and say, “We care for 5 million people, but we’re in communities with tens of millions.”

We can say: based on our data, here are 1,000 individuals who need a specific kind of care. And then we can proactively do outbound outreach and get them in. No human can parse through 10 or 20 million individuals and find the 1,000 who need a particular service. That would take decades.

But we can now do that easily at scale. So when we talk about agents, we shouldn’t limit the conversation to tasks we just don’t want to do anymore—we should be thinking about entirely new possibilities.

Rohit: I see what you mean. Yeah. That’s a very interesting aspect. 

Ritu: Sara you really hit the nail on the head. Rather than just thinking about agents to automate busy work or doing the same things the same way, she’s saying, look at it more holistically and think outside the box—think about new ways of doing things. It’s what we always talk about in our webinars: generative AI is not just about doing the same things better, but looking at things in a whole new way. The technology is so amazing, and with summarization and pattern matching—these are capabilities we just didn’t have before. So, you really need to leverage those to think about new ways to address a problem. That was very insightful. Thank you. 

Rohit: We previously touched upon in the conversation how you’re leveraging a data-driven approach. With such a big health system and so many disparate systems, how did you think about getting this data engineered in a way that would be useful for the applications or products you’re building? Just wanted to go back and talk a little about that interoperability and how you stitch together the diverse systems. What approaches do you take on that side of things?

Sara: This is a credit to our CIOs in the past, in particular. We’ve been on a multi-year journey to get some of our infrastructure in order and do some of the data platform work we needed. Now, we didn’t know that in 2022 we were going to have large language models available to us and go through this big revolution, but we had already done a lot of work getting alignment around our electronic medical record. So, we don’t have ten different systems—we’re on Epic, and there’s a tremendous amount of alignment around that.

The second step is that we’ve been on a cloud migration journey for many years, and we have all of our data in the cloud. We partner with Microsoft and utilize Azure. Cloud computing has unlocked so much of the potential we’re seeing today.

We’ve also spent many years on this data journey. We have an enterprise data warehouse—it’s very sophisticated. There’s been a tremendous amount of data engineering, unification, normalization, standardization, and de-identification to make the data usable for a variety of use cases. That was all thanks to our information services and IT teams over many years.

In parallel, we were doing a lot of the consumer experience and digital work, and these efforts converged over the last couple of years. That’s what’s really accelerated our progress. We were then able to partner with application companies on use cases like ambient assistance for documentation and charting, or in-basket management, which I mentioned earlier. But we could only do all of that because we had already made those multi-year infrastructure and data investments.

Rohit: Right. So, as we’re getting toward the end of the podcast, I’d like to talk about a couple of things—and please feel free to add anything from your side as well. One is innovation and incubation: how do you look at startups that bring value to the table? And could you talk about some successful startups spun out from your organization? 

Sara: I will say that—I’ll preface this—obviously I have a lot of fondness and warmth in my heart for this activity, but it is a very tough thing to do. It’s labor-intensive, it’s resource-intensive. And so a lot of systems don’t do it because they either don’t have the scale or they haven’t built up the capabilities. 

But many years ago, when we were in more stable financial times, we were able to look—and thanks to the foresight that we had—we were able to say, there’s missing technology, missing infrastructure out there that prevents us from successfully, for instance, matching supply to demand for our patients who need care, or delivering a truly personalized experience, because we know our patients and we’re able to create personalized experiences around what we know about them. 

And so, in those two cases, we went really deep into these problem areas. We had a team of folks that could build technology—big startup folks or big tech folks—product managers, engineers, UX, UI, product analytics folks. And we built two companies.

One was called DexCare. We spun that company out in 2021. That’s exactly what they do—they do supply-demand matching for on-demand care. So essentially help folks find the care that they need, make sure it’s discoverable, direct people to the right venue or modality of care, and ensure that there’s capacity. So you’ve got to manage the supply side. DexCare was spun out in 2021, and they’ve been off and running, doing amazing work.

We also spun out a company last year in 2024 called Praia Health. Praia is focused on personalization and driving engagement through knowing an individual and giving them exactly what they need—what’s relevant to them—as opposed to a more vanilla or generic approach.

So we built those two companies. They’ve raised venture capital in the tens of millions of dollars, and we continue to utilize them to power our consumer experiences in the system.

Rohit: That’s great. Sara, would you like to talk a little bit about what you see in the future—maybe over the next year or two? Things are changing fast, but what are some of your thoughts about generative AI, LLMs, and GenTech AI? How should one approach it? 

Sara: You know, if I knew that part, I’d probably put more money into the stock market—or any money into the stock market. Of course, I don’t. I’m very wary, especially with how quickly things are moving. But I think there will need to be some breakthroughs in the not-too-distant future, more focused on the energy that’s needed to power these sophisticated and energy-hungry models.

We’ll get more efficient, but we still have to solve some fundamental energy problems. That’s a big focus area—because without that, we can’t push innovation much further. There’s just too much GPU consumption.

Another thing I think we’ll see more of in the next couple of years, on the core AI software side, is a normalization. Right now, there are tons of little solutions out there, and some are getting massive valuations for unclear reasons. I think we’ll start seeing more focus on wholesale redesign—of business processes, data needs, operational needs—not just tech for tech’s sake. More AI-native thinking, not just substitution.

I’m especially curious to see what happens in the ambient assistant space. Some of those companies are at huge valuations, and I think a lot is going to change for them in the near term.

I’m also really excited about the monitoring and observability space—how we know what’s going on with these models. Before deploying, you have to test and validate, of course. But in runtime too—if you’re using an unsupervised, non-deterministic model, you better have a system to ensure it doesn’t go rogue. I think we’ll see advancements there.

And I’m very interested in the new workflow automation tools. I think the toolsets themselves will get better, allowing people to build the use cases they want more easily. That’s something I’m excited about.

Ritu: I just wanted to ask a last question, which was more like on the philosophical side. Because just a year ago when we were doing our webinars, we were talking about human in the loop. You know, AI is here to augment, not automate, and within a year we’ve gone, full 180 and now we are saying that it’s agents, it’s going to be totally autonomous. They’re going to be doing things on their own. So, like you said about rogue agents or shadow agents, how do you feel about seeding that kind of control to energetic workforce, which humans may not be completely in control of. 

Sara: Frankly, I think it is almost like fighting gravity and our job is to make it as responsible and humane and ethical as possible. I’ll give you another example, which is like, folks often used to say, back before we got really comfortable with sharing our own personal data all the time, folks would say things like, “I think that each individual needs to own their own data.’

That’s like a fine statement in a way, but it’s like a very facile statement. What does that actually mean? And like when you actually get into it, nobody owns their own data. Companies own your data and they do stuff with it and you have agreed that it’s okay, you know? I think we actually missed the mark on doing it humanely and ethically and with integrity and in this case, let’s not miss that Mark and say, ‘we know it’s going to happen, that we’re going to have huge displacement, huge amount of like change economically from a workforce standpoint, from an experience standpoint. It’s going to happen. How do we do it right?

Rohit: Thank you, Sara. This has been a great interaction. We really appreciate you being our guest on the podcast. And I hope the audience will like it as well, so thank you.

Subscribe to our podcast series at www.thebigunlock.com and write us at info@thebigunlock.com  

Disclaimer: This Q&A has been derived from the podcast transcript and has been edited for readability and clarity.

Q: Hi Vishisht. Welcome to The Big Unlock podcast. Very happy to host you. You’re joining us from Nevada, and it’s such a fun place. I was recently there at HIMSS, and we got to chat a bit. I’m Rohit Mahajan, Managing Partner and CEO at Damo Consulting and BigRio, and the host of this podcast. Would you like to start with an introduction? 

Vishisht: Absolutely. Thanks for having me. It’s good to reconnect after our initial call.
My name is Vishisht Mehta. I’m an interventional pulmonologist based in Las Vegas, Nevada. I’m the Director of Interventional Pulmonary at Comprehensive Cancer Centers of Nevada, a community practice with a large oncology presence. My colleagues and I are the lung specialists within the organization. We see a wide range of conditions—COPD, asthma, lung cancer, infections, lung scarring, interstitial lung disease, and more.

I trained in New York City, and working in Las Vegas has been a very different professional environment. Since starting practice, I’ve become increasingly involved in digital health, pulmonology, and emerging technologies. It’s been very interesting to be part of that space.

Q: Yeah, of course. So Vishisht, tell us a little bit about when and how you got attracted to the tech side of things. I guess you still practice even today, right? So a large amount of your time is spent seeing patients, but on the other hand, you have this strong interest in new technologies and AI, which we’re going to talk about. So how did you get involved with that?

Vishisht: So as far as the AI thing, it started because one of the vendors reached out to kind of just talk about their product, and I found that technology interesting. This is about three, four years ago now, but it did not seem ready for prime time. The concept was definitely intriguing.

So I tried to educate myself more about: What is AI? Is AI here yet in medicine and pulmonology? Is this something that will be technically successful but may or may not actually do anything for my patients? And then, all different AI technologies are different. Even though they may look somewhat similar, some of it is patient-facing, some of it is not.

So that’s kind of how I got involved from the AI perspective. And then COVID happened, and we started doing a lot of telemedicine. So that was another exposure to digital aspects of medicine.

There are other technologies emerging, for example, functional imaging in lung diseases. Most of the normal imaging we do is static—as in, a patient takes a breath, holds the breath, and we take a picture—chest x-ray or CAT scan.

But functional imaging is new and emerging now, where we are able to see how different phases of breathing affect someone’s lung. That’s information we did not have.

So I try to expose myself to these things because we just need to know. And it’s also very interesting. This is very nascent. I’m early in my career. These technologies are early in their deployment, development, and adoption. So it’s something I think will be a part of medicine going forward—and a big part of how I’ll be involved in medicine. So that’s kind of the background.

Q: That’s very interesting—how you segued into AI and technology. You talked about telemedicine as well. Just curious to know, on the telemedicine side of things—obviously the number of visits has dropped—but are you still seeing a mix, Vishisht?

Vishisht: Definitely. Telemedicine is here to stay, and our patients like it. We like it. It works very well for the right scenario. Sometimes if a patient is coming back to follow up on some imaging or blood test, they don’t necessarily need to be in the office. We don’t need to be sitting across from each other. We can do it just like how we’re doing now—have a conversation, look at labs, blood work, images, and make a decision.

I’ve had patients do telemedicine appointments from the park, the playground, from home. A few have done it at the airport. Routinely, people do it during lunch breaks or at work. Now they don’t have to miss work. They don’t need someone to drive them. There’s so much time saved.

Because it’s easy, you can have visits more frequently. I’ve had patients with bad flare-ups of COPD or asthma. We do a telemedicine visit, I prescribe what they need, and they avoid going to the hospital. They don’t have to come in. If they have an infection, others aren’t exposed. So our patients are also safe. It just works. As a concept, I think it’s very useful, helpful, and definitely here to stay.

Q: That’s amazing. The other thing I wanted to ask is something you mentioned—functional medicine or functional imaging. Could you talk about that? Are the machines different and new? That must mean more data, right? How does that work? 

Vishisht: Functional imaging is very interesting. There are different parts to it. For example, we do a test called pulmonary function test—typically called a breathing test. A patient sits in a transparent box. The box is closed because we need to control the environment. They get coached and do a series of breathing maneuvers—deep breath in, deep breath out, blow hard, breathe normally.

These tests assess how the lungs are functioning. We can see if someone has asthma, COPD. We can measure lung size and the ability to extract oxygen. That’s very helpful information—but there are no pictures involved.

Also, for each outcome, we get a composite answer for both lungs combined. But that assumes the condition affects all parts of the lungs equally, which is not always true. Some patients have emphysema more in the upper lungs, or conditions that affect the edges or bottom parts.

These tests can’t tell that. The hope with functional imaging is that we’ll be able to image the lungs entirely and see if different areas function differently. That’s very promising and may change how we understand lung conditions.

You asked about hardware. Some technologies, like scanners using hyperpolarized xenon gas, do require additional hardware, capital investment, and training. But other imaging technologies use existing scans with different protocols. For example, we might do a scan after a deep breath in, then after a breath out, and compare.

So for those, we already have the technology—we just need different protocols and people who can interpret the results. The scan machine is the same; the protocol and interpretation differ. The leap to get there is small. These are just some examples of where things are going. I believe it will move in that direction, but we’ll have to see. It’s exciting to have new tools to offer our patients.

Q: Right. And let’s talk about AI. You mentioned it a couple of times. What kind of AI is applicable in your area of practice? What are you seeing?

Vishisht: AI in medicine—and in pulmonology—we’re seeing it show up in different areas. Most prominently, we’re seeing it in lung cancer. A lot of it is focused on early detection of lung cancer, or identifying spots or nodules on the lungs.

We know lung cancer is the most lethal cancer. But if caught early, survival is much better. In some cases, we’re talking about cure—not just management. So we want to screen high-risk patients.

AI helps by identifying high-risk patients, currently by looking at smoking history and pulling that information from the chart. It flags those patients for lung cancer screening discussions.

Another area is incidental findings. A patient might get a CT scan of the abdomen, and a small lung spot is seen in the lower lung—even though the scan wasn’t done for that reason. Same with scans of the neck or thyroid—you might see part of the upper lung.

AI software can identify these nodules, whether in the report or image, and generate a list for the hospital or doctors. It flags them so we can follow up and determine if they’re important.

We’re also seeing AI used to judge the risk of cancer in these spots. Many people have lung nodules, most of which are not concerning. But AI can analyze data we can’t see with our eyes—data that’s in the scan—and help us assess whether a spot is likely cancerous.

It’s not diagnosing cancer yet, but it helps assess risk. So that’s where we’re seeing AI most—in early detection and identification of nodules.

Q: That’s great. You mentioned lung cancer screening—I have a curious question. In your health system or nationwide, since you’re connected with physicians elsewhere, what is the current percentage of eligible people who actually get screened? Is there a gap? 

Vishisht: There is a massive gap. It’s probably the biggest gap between eligibility and screening for any cancer. Ballpark, we’re screening about 5–6% of eligible patients. That’s a huge miss compared to other cancers like colon or breast, where screening is around 70–80%. 

Q: Any thoughts on why that gap is smaller for those cancers and bigger here?

Vishisht: Yeah, you know, at some level, lung cancer screening recommendations have been out for a while, but for whatever reason, they seem to not have percolated as much. I think the true number of patients who get lung scans is probably a little higher than this because patients may be getting scans for reasons other than screening, which maybe are not being tracked. But there’s no doubt that the gap is there.

I think there’s a lot of stigma also. Because lung cancer, because of the association with smoking, has some amount of blame attached to it, unfortunately. We should not be blaming patients, certainly not for actions in the past. And if someone is eligible, we should have the discussion with them.

One of the things to think about, if you think about AI in this space, is AI is doing a lot of interesting things when we find the lung spot. But what about finding the patients who need to be screened? I think that’s another place where AI is going to help us.

And I will also say we have to keep in mind that maybe the answer is not AI to improve lung cancer screening. Maybe we just need resources, education, manpower. That’s another long discussion. But we should not be thinking that AI is going to fix it, and it could be many things that are likely to help move the deal.

Q: That brings me to the point we were talking about a little earlier—that you actually have a website and a video that you’ve put up for people to learn more about this. You mentioned pulmonary.ai. Also a video on YouTube, right? So would you like to tell us a little bit more about that?

Vishisht: Yeah, for sure. Again, like two-ish, maybe three years ago, when I was trying to get more details and get more involved, I said, let me go and find AI and pulmonology resources. So I went online and I went looking and looking and looking, and I could not find anything where all the stuff that I was interested in was in one place.

I had to go and pull this article from here, that article from there, this clip from here, to try and educate myself. And then I said, why don’t I try and make the place where I can post some of these articles? It’s not going to be everything—because it’s impossible to keep up. One-man army, sort of.

But to my surprise, pulmonary.ai, the domain was available, so I took it. And I have been trying to add different pieces of pulmonary and AI literature, research articles in different areas, so other people who are interested in this subject, like I am, don’t have to run around as much. At least they can go to this website and have a good start.

This is not meant to be exhaustive and cover every possible thing—again, it’s not possible. But at least as a starting point, they can go on the website. I maintain it myself. There’s no commercial, no ad, no sponsor. I pay for it myself. It’s just my personal hobby and effort to find like-minded people.

That effort helped me generate my first large lecture on this, which was AI in Pulmonary Nodule Management. I went on YouTube, I did that for the Society for Advanced Bronchoscopy, a wonderful medical society that I’m a part of. And a year later, the follow-up lecture, which is the one you’re referring to, was Which AI Should I Buy?

Because after the first one, we started to see so many different vendors show up in this lung nodule space. And the question that I kept getting asked was, “Okay, that’s fine. Which one?” And not only which one, but “How do I decide which one?”

So that led to the second one, Which AI Do I Buy? And that’s, of course, a very important discussion. Sometimes people ask, “Do I definitely need AI?” And right now, I’m saying—I don’t know if everyone definitely needs it, but I think the conversation should definitely be had.

Because there are many challenges—cost, integration, reimbursement. The other challenges are the physicians, administrators, or nurse navigators—do they know how to use the software? Do they understand the pros and cons?

We do not want to have a problem of round peg, square hole—trying to fit things where they’re not supposed to fit. That’s mostly what I speak about—how rather than what. Because it’s not for me to tell any one person this is better or that is better.

Q: So, what’s one or two nuggets from the Which AI Should I Buy? video that you’d like to share here?

Vishisht: That’s a very hard question. I think one thing that’s important, and I’ve realized this over time, is that physicians, administrators—whoever is interacting with the software or making purchasing decisions—need to have a basic level of understanding of what they’re interacting with.

Even when we routinely read research articles, we’ve not—and for obvious reasons, because AI wasn’t there—we’ve not been familiar with the terms needed to judge AI literature. So when you look at an AI paper, you have to look at or be familiar with terms like structured/unstructured data, unsupervised learning, what was the development, what was the deployment, internal validation, external validation.

So I think one takeaway would be to familiarize yourself with those kinds of things. Where can you go? Shameless plug—my lecture is one place someone can start, but that’s not the be-all and end-all. There are resources online. Some medical societies have material. MIT has a course on this—I’ve taken it. I think it’s a very good course. Stanford has one through Coursera. There are places where we should go.

So I guess the one takeaway would be—if people are considering these technologies, then just understand how they are different, so the right people can make the right decision for the right circumstance, if it is going to be some AI technology.

Q: That’s awesome. I think we’re almost coming to the end of the podcast today. I’m hoping we’ll have more conversations in the future. 

Vishisht: I appreciate the opportunity to talk to you and your audience. In closing, all I’d say is that healthcare is evolving, changing—probably faster in the digital space than we’ve seen before.

So I’d want my colleagues to go out there, educate themselves, listen to lectures, and follow folks like yourself. A lot of the new information may not come from traditional medical societies—just because of the speed. Things change faster than we’re used to.

And at the end of the day, being well-informed, as much as possible, is important with these newer developments. So that’s all I got. Thank you.

Subscribe to our podcast series at www.thebigunlock.com and write us at info@thebigunlock.com  

Disclaimer: This Q&A has been derived from the podcast transcript and has been edited for readability and clarity.

Q. Hi Aneesh, welcome to The Big Unlock podcast. Very happy to have you here. I’ll do a quick intro and hand it back to you. I’m Rohit Mahajan, Managing Partner and CEO at BigRio and Damo Consulting. It’s great to know that you’re already familiar with Damo from your previous interactions with Paddy. We’re continuing to build on The Big Unlock podcast that he started. We’re now in Season 6, so welcome, and over to you.

Aneesh: Thank you, Rohit. I’ll begin by saying how grateful I am that you’ve picked up the baton. Paddy was a larger-than-life individual with had a lot of fans, myself included. I’m grateful you’ve continued this.

By way of background, I’m Aneesh Chopra, currently Chief Strategy Officer at Arcadia. We’re a market leader in putting healthcare data to work in value-based care for payers and providers, with a focus on helping individuals get the best possible care at every step of their journey.

Personally, I’m passionate about improving collaboration between the public and private sectors. I served in the Obama administration as the first U.S. Chief Technology Officer—a position that has thankfully remained bipartisan. I’m grateful for the opportunity to demonstrate what’s possible when we apply technology, data, and innovation to some of the country’s biggest challenges.

Q. That’s great to know. Could you tell us what first attracted you to healthcare?

Aneesh: Like many things, it was about timing. I was an undergraduate at Johns Hopkins in the early ’90s when President Bill Clinton brought national attention to healthcare reform. I had the opportunity to get involved—first as a student, and then as an intern for a Maryland initiative aligned with what Clinton was advocating, led by the university’s president.

That experience gave me an insider’s view of what it could look like to shape the future of healthcare. I’ve always believed that the best outcomes come from collaboration between the public and private sectors. Healthcare, in particular, is one of the defining challenges of our generation.

Back then, I was passionate about the role of internet technologies in democratizing healthcare. Today, I feel even more energized about the potential of AI to do the same—lowering costs and improving outcomes for everyone.

Q. Absolutely. That’s great to know, Aneesh. That brings me to the comment we were discussing before the podcast—about making use meaningful again. That’s a great catchphrase. Can you explain your thought process behind it and what it means to health systems, patients, and the larger population?

Aneesh: Thank you so much. By way of background, your audience likely remembers the term meaningful use—two words Congress wrote into the HITECH Act that led us on the journey of accelerating health IT adoption starting around 2010 when subsidies kicked in.

There’s always been a trade-off. What does meaningful use really mean? If you go to Best Buy and purchase software like TurboTax, you expect it to help you file taxes. But in healthcare, just installing software doesn’t guarantee better patient care. It may only serve the back office for billing or administration, without clinical benefit.

So, Congress allowed us to define what it means to be a meaningful user. That sparked debate. Should we focus on maximizing productivity in a fee-for-service model—like seeing twice as many patients a day—or should we focus on outcomes?

In the Obama transition team, our vision was to shift toward outcome-based models that reward better care. If we fix how people get paid, and incentivize longitudinal care, then technology becomes essential—not just during visits, but also between them. It can remind us of care gaps or flag things like unfilled prescriptions that signal problems with adherence.

So, meaningful use for organizations focused on long-term outcomes differs from those operating under fee-for-service. Our original intent was to prepare for the future—“where the puck is heading,” to quote Gretzky. But the industry found that difficult. We ended up with a half-measured IT program—subsidizing tech mostly built for fee-for-service, with only partial support for value-based care.

Now, the Trump administration’s CMS-ONC RFI is a chance to finish what we started in 2009-2010.

Q. The RFI you’re referring to is the CMS-ONC RFI, right? Can you explain what this RFI is for those who may not be familiar? 

Aneesh: It’s a great opportunity for the private sector, nonprofits, providers, and plans to come together and say: here’s how we want healthcare IT to operate, how information should flow, how patients should access their data, and how clinicians and health plans can coordinate care to improve quality.

It’s an open invitation—across a dozen or more specific use cases—to help define the future we didn’t quite finish under meaningful use. It asks each stakeholder: What do you need? What do consumers want? How can physicians interact with data in real time? Historically, health plans were kept out of public health information networks—so how do we bring them in, and when?

It’s a short 30-day comment period. By the time this airs, it may be over—but that’s not the end. The executive branch will likely take action over the next year. I hope early adopters raise their hands and participate voluntarily before any formal regulation. That’s always been a hallmark of my work in government, and I’m grateful that the Trump administration carried that idea forward.

Q. That’s great to know, Aneesh. Thank you for sharing that. Based on your experience with the Obama administration or your current role, are there any success stories involving AI or interoperability that you’d like to share? 

Aneesh: Yes. We’re grateful to support over 150 health systems, health plans, and ACO networks. We serve as their core data platform to build the longitudinal patient record. The great news is, once you’ve done that work, you can apply use cases on that foundation to help people be more successful.

There are three areas some of our customers have been advancing. The most exciting is conversational AI. One challenge in population management is taking responsibility for individuals who may not come into your clinic regularly. You need to reach out—remind them to get flu vaccines, check blood pressure, manage diabetes.

One academic medical center, mostly a fee-for-service organization, had underperformed on longitudinal measures. They were scoring one or two stars in the Medicare Advantage program, where four stars or more are needed to earn incentives. Their enablement partner—our customer—brought in a conversational AI workflow. It used the physician’s office phone line, declared the automated agent was calling on behalf of their doctor, and helped close hundreds, if not thousands, of care gaps over a 90-day sprint. As a result, they went from a one or two star rating to a four-star rating across several MA contracts—earning over $6 million in bonuses they previously hadn’t qualified for. That’s an example of the co-pilot concept in care management.

Another use case is evidence-based medicine. There’s debate around health plans denying care via prior authorizations. But if you frame it as evidence-based decision support, it becomes more of a nudge. One of our large academic centers, with a library of approved guidelines, has started piloting an AI co-pilot to monitor whether the patient in front of the doctor is a good candidate for a certain treatment. This reduces the cognitive load on physicians and integrates reminders at the point of care. It’s promising.

Finally, interoperability. CMS rolled out a daily FHIR API for health networks in the ACO REACH program. It moved claims data from being available 60 days after service—useful mainly for actuaries—to more real-time use for care managers. One of our advanced primary care clients, focused on high-need senior populations, saw their ADT (admission-discharge-transfer) data coverage improve from about 40% to nearly real-time. It may not be real-time for every patient, but it’s now fast enough to take meaningful clinical action for many more patients.

Q. That’s great to know. And now, thinking a little bit into the future—before the podcast, we were talking about the idea behind healthcare information fiduciary. Could you explain more about your thoughts on that and where this concept might go? 

Aneesh: Yes. A little bit of history first. In the Obama administration, we proposed a fiduciary rule through the Department of Labor for self-insured employers managing 401(k) plans. At the time, there wasn’t much transparency. Many firms were charging 8–10% in fees, when it should have been less than 1%. That became a massive cost to employees’ retirement savings over time—estimated at nearly $3 trillion.

More importantly, brokers had incentives to sell products that paid them higher commissions, rather than products that met the individual’s risk profile or goals. So we proposed a fiduciary rule: if you trust me with your information in a complex domain like financial services, I must act in your best interest—not based on how I get paid.

That standard inspired me to carry the concept into healthcare. As we build health data hubs and consumers gain the right to control access to their information—or request and keep copies—we need to create a marketplace of apps that people can trust. These apps might use AI or other tools to help consumers not only aggregate records, but make better decisions at every step of the healthcare journey.

The key is that these apps must act in the patient’s best interest—not in the interest of a health plan pushing for lower-cost care, a hospital steering to more expensive services, or a pharma company promoting a branded drug when a generic might be better. So a healthcare information fiduciary is both a technical and ethical standard. It’s about creating a trusted ecosystem that organizes and activates personal health data responsibly, for the consumer’s benefit.

Q. Is this something already concrete and coming down the pike, or is it still just an idea? 

Aneesh: It’s a bit of both. I’m one of the co-founders of the CARIN Alliance—a bipartisan effort with Governor Mike Leavitt, Dr. David Blumenthal, and the original National Coordinator, Dr. David Brailer. The four of us conceived a program now led by Ryan Howells to create a category of consumer-facing apps that follow a voluntary code of conduct.

This code addresses one part of being a fiduciary: transparency in how my data is used so I can trust the app. It goes beyond HIPAA, requiring disclosure of even de-identified data use—something hospitals and insurance companies don’t typically do. So the CARIN code of conduct is a step toward a full healthcare fiduciary model, but we’re not there yet.

Combine that with data-sharing rights and an upcoming AI code of conduct, and we may soon see apps that say: “Share your lab results, and I’ll give you next steps”—as we’ve already seen with GPT-like tools.

Q. That’s great. Before we wrap up, I’d love to hear more about your book The Innovative State. It’s been a while since it was published, what lessons remain relevant today? 

Aneesh: I appreciate that. I was inspired by Sam Pitroda. Back in the ’80s, India had just 300,000 phone lines for 300 million people. Sam, after selling a business, returned to India with a symbolic salary and a bold goal: connect every remote village. He rejected both the public subsidy model and deregulation-only approach. Instead, he chose a third way: innovation.

He recruited hundreds of brilliant minds and built a digital-first telecom network. Within a decade, villages had STD booths and modern phone access.

That inspired my approach to public sector innovation—looking for that hidden “third way.” The heart of my book focuses on strategies that empower both government and private actors through handshakes and handoffs.

Innovation is often bipartisan. Washington can create the pathway, but it’s the private and nonprofit sectors that must execute. That includes opening government data, setting interoperability standards, introducing outcomes-based payments, and recruiting talent for “tours of duty” in government. These ideas are still highly relevant today.

Q. Absolutely. So. as we end the podcast, Aneesh, any parting thoughts or any final words for the audience that you would like to share? 

Aneesh: Yes. In the spirit of the conversations Patty used to lead—this is our moment to raise our hands. If you see the future and want to help build it, now’s the time. The Trump administration’s RFI will kick off a series of actions this summer, fall, and maybe winter.

If you’re ready to be an early adopter, we want to hear from you. Reach out to me or connect with Washington. Let’s organize the early adopters, test what works, and then scale it across the sector.

Q. Amazing. Thank you, Aneesh. This was a great conversation. Hope to have you back again soon. 

Aneesh: Thank you, Rohit. Appreciate you having me. 

Subscribe to our podcast series at www.thebigunlock.com and write us at info@thebigunlock.com 

Disclaimer: This Q&A has been derived from the podcast transcript and has been edited for readability and clarity.

Q. Hi, Rob I’m Rohit Mahajan, Managing Partner and CEO at BigRio and Damo Consulting, and host of The Big Unlock podcast. It’s been a popular show for many years, and we’re continuing the tradition started by Patty Padmanabhan, the founder of Damo Consulting. Many healthcare leaders have been on this podcast, and it’s great to welcome you. Over to you for your introduction, Rob.

Rob: Terrific, Rohit. It’s a pleasure to spend time with you and with your audience. I’m Rob Posner. I’m the Chief Technology Officer for AbsoluteCare. I’ve been with AbsoluteCare for about two and a half years now. I joined because of the game-changing mission of this organization. I’ll get into that in a moment, but first, a little background.

Prior to AbsoluteCare, I spent the last decade in similar private equity–backed, provider-centric organizations that are changing healthcare. My passion for healthcare has really been about the transformation that’s needed—not just for those organizations, but for the country and the world. I truly believe in that mission and the role technology plays in achieving it. That’s why I decided to move into the healthcare industry.

Before that, I was in hospitality and entertainment. I live in South Florida, and the cruise lines are big here. I’ve worked with major cruise lines and Disney Parks in particular. I built a team and worked backstage at Disney Parks where we developed the MyMagic+ experience and led major aspects of that global rollout—transitioning to a managed guest experience. It was one of the early efforts in what’s now known as the experience economy, leading products and services by experience for consumers and guests.

Q. That’s awesome, Rob. I’m actually a fan, and I don’t think I mentioned this to you last time—Disney has a university where they run programs on leadership, quality, service, and a few other topics. I’ve been through all four of those. So great to know that background. Now that you’re at a healthcare organization, could you tell us more about AbsoluteCare—what the business model is, and how it aligns with your experience in provider-centric organizations that are really changing the healthcare industry?

Rob: Absolutely. AbsoluteCare’s mission is to improve the healthcare of the nation’s most vulnerable populations. We do this by improving outcomes through holistic care and care management.

We refer to those we serve as members—not patients—because we contract with payers for a panel of their members. So they become our members, and we take full responsibility for their care. Our tagline is “Beyond Medicine” because we deliver primary care, care management, pharmacy, behavioral health, and address social determinants of health in a comprehensive way. That’s what differentiates us from traditional care models.

We operate under a full-risk, value-based care model that demonstrates improved outcomes. We’re fully accountable—clinically and financially—for delivering on the triple aim of healthcare. We don’t just talk about it; we live it. That accountability is central to being a sustainable and impactful organization.

Q. And, and I understand you are in several markets and tens of thousands of patients now, so there must be some early learnings. 

Rob: We are rapidly growing. We were in seven markets at the end of 2024, and we’re about to be in 11 markets over the next couple of months. We serve tens of thousands of high-acuity, high-utilization managed Medicaid members. That’s important to understand—these are some of the nation’s most vulnerable individuals. They often have multiple chronic conditions, compounded by social determinants of health. We need to address all of those factors to be successful. 

Q. That’s awesome. And your delivery model is partly center-based and partly community-based, right? That seems like something unique AbsoluteCare brings to the table. 

Rob: Absolutely. We deliver about half of our care in our centers and the other half in the community. In each of our markets, we have a center located in the urban core to serve our members. But the reality is, it’s often difficult for our members to reach us. Since we’re responsible for their outcomes, it’s on us to go to them—wherever they are—to ensure we’re delivering care and care management that drives better outcomes.

Q. That’s amazing. I have a curious question—from my perspective. You mentioned that you get the cohort of patients from the payer side, right? Because they obviously want better outcomes. So, do you work with providers at all?

Rob: Our organization includes providers. In other words, we deliver the actual care. We have our own employed providers who deliver primary care services. We also employ care management teams, behavioral health specialists, and pharmacists who work in our pharmacies. So we provide comprehensive care and care management in our centers. Additionally, our providers and care teams go into the community to deliver care in members’ homes and other facilities. 

Q. So I understand that there’s a lot of technology at play here. Obviously, we are all leveraging all kinds of technologies to help better outcomes for these members and for these patient populations. So, would you please talk to us about your journey? Like how was Absolute care and then the how did you fast track? I think you were talking about fast tracking transformation. 

Rob: Oh, absolutely. Yeah. So I’m the first C-level IT executive in the organization. And not surprisingly, that means that usually there was—coming into it, I think the teams were doing the best they could with what they had, but they didn’t have a seat at the executive table. And so we found that the state of technology was not where it needed to be. And of course, that’s why they brought in a Chief Technology Officer.

So nothing was really surprising from that perspective. But systems were not configured for our mission. It’s not unusual because systems are not really made or designed—typical EMRs are not designed—for value-based care, full-risk provider models, right? So it’s not surprising they weren’t configured correctly.

The technology team was following the business rather than leading. There was a lack of appropriate governance. And like I mentioned, the teams were working really hard, and I’m proud of the work they did to get to that point. But clearly the organization recognized that more needed to be done.

Just to explain the point—when I got there, within a month or so, we had a terrible situation where we had a significant outage of our EMR that went on for more than a day. And no one in the organization thought to tell me. Our EMR—which all of our providers and care management count on to do their jobs every day—was down. And she let me know that we were down. So it was the kind of thing where we had to change the culture, change expectations, and deal with accountability.

That was kind of the start of that story—saying, okay, that’s where we were. I installed new leadership within my department, got people who understood where we need to go, how to set expectations for our teams, reset what operational excellence means, and establish that culture of accountability.

We worked very rapidly to start reconfiguring our EMR. Even in terms of the governance—we had a steering committee for our EMR, but what we found was a whole bunch of leaders were sitting on calls that should be about strategy, and they were dealing with day-to-day issues. That pointed to the fact that we didn’t structure our governance well. We didn’t have core teams and the right people dealing with the day-to-day so we could address those, and then separately deal with things like: should we be moving to the cloud? Should we be on one instance of the EMR?

These things have to run in parallel, and you have to have the right people engaged in those conversations—and the right cadence. Those are some of the things we had to do very quickly to start dealing with the rapid transformation that we needed to make as an organization.

Q. Right. And you really had to do groundbreaking—or from the grassroots—you had to build a care management system, because the EMR really isn’t suited for that purpose, right? So please talk to us about that. It’s something very different, I guess, that you’ve done in the organization, and it’s like the bedrock for engagement.

Rob: Absolutely. And just about all EMRs you could categorize as being focused on delivering care in an office, in an ambulatory environment. What does that mean? For a patient visit, a member visit, in a center. That’s really what it’s designed to do.

But when you talk about all the work we need to do to provide longitudinal care for that member—it’s the things happening 99% of the time when they’re not in the doctor’s office. That’s care management, and that’s where a lot of the outcomes for our patients happen. But it’s not really addressed in EMRs. EMRs look at things like decision support for the doctor and care gap closures, but care management is its own thing. It has its own workflows, and we need to make sure they’re focused on what produces improved outcomes.

So, we determined that EMRs don’t really do that, and we brought in a full-fledged care management system. We implemented that and went live about a year ago. We also integrated it with our EMR, which at the time was a fairly significant step, because the objective was to let people work from one pane of glass.

If they’re in the EMR, they shouldn’t need to jump into the care management system to check on something—and vice versa. That’s been a journey. We’ve done the first couple of iterations to make it work. There’s more to do, but we’ve gotten our MVP product into the hands of our markets, and it’s being used successfully. It’s been a great success story, showing how we can integrate care and care management in a way that reflects our model. We have a care model, and we need to make sure our systems are aligned with that—not force people to work within systems that don’t fit.

Q. Absolutely. That’s significant. And it’ll always need to be taken to the next level, based on user feedback and the people who engage with it. So we talked about care management, Rob, and it’s fantastic that you were able to build a product to engage members and support all the stakeholders. How did you think about the tech stack? Because you’re operating at two levels, right—like we discussed before, also out in the communities?

Rob: Absolutely. And from what I’ve seen in the industry, these systems are typically designed to work in a facility—an office or a hospital—or maybe a hospital-at-home type setting. The assumption is that the technology stays in one place: someone is a remote worker, someone is working in a hospital, etc.

But in our integrated model, that’s not the case. We’re delivering half of our care and care management in people’s homes, which means our workforce is mobile. They’re in the field, moving between members’ homes. We have transitional care managers who go out into facilities when we get an alert that a member’s been admitted to the hospital or has visited the ED.

So we need systems that can work across a variety of environments. Our original tech stack wasn’t built for that—it was built for more stationary settings. So we had to completely rethink it. We tested different laptops, connectivity solutions, and carriers until we found a solution that worked. We looked at each market individually, since different carriers perform differently depending on the location.

We landed on a solution that includes new high-powered laptops, MiFi devices, and iPhone 15s. It turned out that the 5G technology—and specifically the antennas on that hardware—allowed us to overcome many earlier challenges. 5G really opened up the capability and gave us the bandwidth we needed to connect to an EMR in the field.

And once we did that, it really unlocked the power of our solution. I saw that firsthand when I did a round with our transitional care managers. One of them was at the bedside of a member who had been admitted to the hospital. Their job is to coordinate care, make sure follow-up appointments are scheduled so we can continue to support the member.

And right there at the bedside, the care manager was able to open the EMR, schedule the appointment, confirm it with the member, and ensure continuity of care. That’s exactly what we need to see—technology working in the field, making a real difference in the care and care management we deliver.

Q. That’s awesome. Yeah. Being able to schedule appointments at the bedside is, is fantastic. Even today to reschedule my appointment with my physician is a big task.

Tell us, you know, uh, Rob, that in building all these solutions to solve the problems that you saw, how did you go about the governance aspect of it? 

Rob: Yeah, governance is a really important aspect here. It’s really easy to focus on what system you use—there are great enterprise solutions for most of the challenges we face in healthcare, broadly speaking. We have our in-house pharmacy, so we implemented a pharmacy system. We have a care management system, our primary EMR—so we have the big pillars, if you will, of our clinical applications.

But what’s really important to unlock the value of those systems is establishing product teams, having effective steering committees, and creating a proper intake management process. It’s so easy to get lost in all the new requests that come in. You need a way to manage that, making sure we stay aligned with the business—both in terms of the strategic plan and the day-to-day changes happening.

So you have to manage between the strategic and the tactical. You can’t just do one or the other or you won’t be successful. Keeping that executive alignment—engaging the appropriate executives at the right times—and having an overall change management and governance approach are all key building blocks. If you don’t have these things, it doesn’t matter what systems you put in place—you’re not going to be successful.

And I think another point—not governance in particular, but related—is not forgetting the small things. Like I mentioned earlier about the community tech stack—one of the things we didn’t think about was the equipment itself. You’ve got to pilot everything. Don’t go live with anything without piloting it, because that’s where you learn the small things that make a big difference.

For example, we provided all this technology, but we didn’t have the right type of briefcases for the care management team. We eventually got them rolling bags—and they had to be locking bags—because of PHI. We’re a HITRUST-certified organization, which is business-critical to us. So we take the handling of patient information very seriously.

Those are the kinds of things that make a difference. If you don’t take care of the little things, you don’t get the adoption—and then you don’t see the business results. Which is why, at the end of the day, you’ve got to think about the big things and the little things. And together, that’s what makes a solution really work for the organization.

Q. That’s great. So Rob, no podcast would be complete without touching on AI and innovation. So, would love to get your thoughts on innovation, AI, and now GenAI. What are you thinking, and what are some of the use cases you might be coming up with? 

Rob: Sure. Look, AI is a really exciting topic in the industry, and for us in particular, I’m really excited about what we’ve been able to accomplish recently, and equally excited about the opportunities in the future.

So on the clinical side, rolling out ambient experience—ambient listening for clinical notes—is the critical use case. We’ve been able to implement that successfully. We went from pilot very rapidly to full rollout. We saw the results very quickly. And look, there’s a lot of change management to do with the providers—to get them used to the fact that this ambient listening device is there, making sure they’re talking to their patients about it and what it means, and how to leverage it effectively. And so that’s a learning process, and we’re definitely still going through that, but we’re already seeing results.

One of the immediate results is that our providers can engage our members more effectively, right? At the end of the day, they’re spending less time hands on keyboard, and more time engaging with our members, having the important conversations that they need to have to deliver care. And so that’s really exciting in terms of the impact on care delivery and outcomes.

Another piece on the clinical side—we implemented a system that detects diabetic retinopathy with fundus cameras. The solution takes the images in the office, sends them immediately to our partner in the cloud, they do a read of those images, and send back—within 30 seconds—a result and a recommendation for referral. What we’re seeing is that because we get that instantaneous result, our patients—our members—are actually going forward and getting that referral and follow-up appointment. And that’s really what we’re talking about: we’re changing the behavior of our members so they get better outcomes, address diabetic retinopathy early, and take care of it before it leads to something as serious as blindness. It’s a really urgent issue, and here’s an example where technology is really motivating our members to take care of their health. That’s amazing and exciting to see.

Additionally, we’re starting to see AI throughout all of our systems—it’s just almost happening organically, I would say, just by vendors providing it. So, certainly Microsoft Copilot, which has become pretty ubiquitous—we’ve rolled that out. We piloted it, saw great results, saw great adoption. I’d say of all the technologies that we’ve released, Copilot was one where we just put it out there, gave some training and tips and tricks, and the uptake was amazing. Unlike an EMR, where providers are required to use it and follow detailed workflows, with Copilot there was no requirement—and still the adoption was high. We’re seeing great productivity results and people learning how they can use AI in their day-to-day work. And that just rolled out recently. We expect to do a lot more with it. We’re doing workshops to enhance learning and help staff understand what’s possible with AI.

So that’s what we’ve implemented so far—those are active in our organization at the enterprise level. In terms of where we go from here, there are a couple of areas that are very exciting to us. Robotic process automation—though it’s not new—is an area where we can continue to refine the EMR experience for our providers and frontline staff. We’ll continue to look at automation opportunities and other AI capabilities within the EMR, like documentation and note summarization, translating into different languages to communicate better with patients, and reducing the multi-click environment of the EMR by automating routine tasks. The next area is really AI agents—looking at what’s happening outside our core platforms that could be managed with more automation and integration. That way, we can free up our team members to focus on what really matters—our members.

Q. That’s awesome, We actually did a seminar today, Rob, on Agentic AI in healthcare, and there was a great response to the webinar. It’s such an interesting topic—people are actively looking for use cases. We’ve worked on several use cases with our clients, and we definitely see Agentic AI as one of the key options to explore.

So I think we’re toward the end of our podcast, Rob. Any final thoughts or remarks you’d like to share with the audience?

Rob: Yeah, I would just say that it’s a really exciting time to be in healthcare technology. I believe we’re at a point of inflection—where not only do we as healthcare technologists see the opportunity, but the business side clearly sees it as well and is relying on the technology function to step up.

And I think that’s happening—as executive leaders begin to expect more from their technologists, but also expect the business side to think about how to leverage technology. That’s where we’ll really start to see the tech and business teams working together to solve meaningful problems and drive real impact.

Subscribe to our podcast series at www.thebigunlock.com and write us at info@thebigunlock.com 

Disclaimer: This Q&A has been derived from the podcast transcript and has been edited for readability and clarity.

Q. Hi Siva. How are you doing today? Welcome to The Big Unlock podcast. Very happy to have you as our guest today. For our audience, as you might be aware, this was started by Paddy Padmanabhan, and I’m building on his legacy. We’ve done many episodes. Let’s do some quick introductions. I’ll start. I’m Rohit Mahajan, Managing Partner and CEO at BigRio and Damo Consulting, and also the host of The Big Unlock podcast. Over to you.

Siva: Wonderful. Appreciate you having me on the podcast, Rohit.
I’m the CEO and co-founder of a company called Cohere Health. We started in late 2019, and we are solving the burdensome issues related to prior authorization.

I’ve been in the healthcare industry for more than 25 years. Prior to Cohere Health, I founded another company called Coda Health Analytics in 2007, built it over 10 years, and sold it in 2018 to EXL. It was a successful analytics company in the healthcare market, funded by VCs including Sequoia Capital. It was a good exit for everyone. We were serving health plans at the time.

Before Coda Health, I started a small company in the provider space and sold it to Perot Systems. I also started my career at Intel, then went to business school at the University of Michigan, got my MBA, and moved to Connecticut. I live in Connecticut now and came here to work for Gartner Group. From there, I wanted to start something on my own, and that’s how I began my healthcare career.

Q. That’s amazing, Siva. Thank you for that introduction. With such successful exits, I’m sure Cohere will also be on a great footing. Could you share how the idea for Cohere came about? You’re so familiar with the healthcare space—I’m sure you saw a big problem to solve. How did it start, and what’s your journey been like?

Siva: As I indicated, in my previous company, I had been working closely with the health plans—health insurance companies. We were doing analytics, and at that time, we were focused a lot more on payments, population health management, care management, etc.

During that process, I came across the prior authorization process, which the health plans were involved with. I’ll get into more detail about prior auth later. So I got involved in that, and it was always in the back of my mind that the process was highly manual and caused a lot of operational issues for both providers and patients.

It was a major pain point for the health plans. After I sold the company, I had to work there for a bit. And then I started thinking: how can we apply AI and other advanced technologies to solve this problem in the healthcare ecosystem? That was my angle.

While I was working with the health plans earlier, some of my clients had indicated they’d be willing to work on this problem—if I could come up with a better solution. That’s how I kind of got into this.

Q. That’s amazing to know that you were able to discover a gap and then build a new business enterprise to fill that gap. So tell us, Siva, a little bit more about prior auth. Most people in healthcare know what prior auth is, but tell us some of the intricacies and more details about it. You’re the expert on it, right?  

Siva: Sure. Prior authorization is, you know—as a patient, for example—when you go to a specialist, they’ll usually examine you. Say you go there for knee pain. Depending on the severity, they might just take an X-ray first to see what the problem is. If it seems more acute, they might order an MRI.

Now, the moment a physician orders something more expensive—like an MRI, which can cost between $1,500 and $2,000—the insurance company wants to know why that test is being ordered. So the physician’s office has to fax or submit information explaining why I need the MRI.

Then, the health plan looks at their policy and decides whether to approve it. So that’s the process. Anytime there’s a costly or potentially unnecessary procedure being considered, this process acts as a check and balance to ensure it’s appropriate.

On the insurance company side, the submission process itself can be confusing. It might happen through a portal, a fax, or a phone call. Then, the health plan assigns it to a nurse, who reviews the information and determines if it aligns with policy. If it does, they approve it.

If the nurse thinks it doesn’t meet the criteria, it goes to an MD for review. The MD might then say, “You don’t need it,” and deny it—or they might approve it. If it’s denied, it’s usually by a specialist on the insurance side who gives a reason—like saying based on the X-ray, the issue doesn’t look serious, so physical therapy might be enough.

My physician might not agree with that, but that’s the process. So then they might send me for conservative therapy, etc. That’s the prior authorization process.

Q. Okay. And Siva, in the first part of what you were explaining, you used the word abrasion, right? I’m very curious—what is this abrasion that’s happening? And second, how long does this process take? Because now the person needs to apply, right? 

Siva: Yes. Right, exactly.

So the process hasn’t always been very clear in terms of what information needs to be submitted. What happens is there can be a lot of back and forth between the physician’s office and the insurance company. For example, the office sends some information, and the insurer says, “No, no, we’re looking for an indication of something else.” Then it gets sent back. The provider looks at the documentation and says, “No, we actually did provide that—here’s where it is,” and they send it again.

So that back and forth adds time and creates more administrative work on both sides.

All this paperwork, documentation, back-and-forth communication, and waiting can take anywhere from five to 14 days. For very complex procedures, it could even take 13 to 14 days. For example, if someone needs surgery, they might have to wait while going through multiple rounds of paperwork and approvals.

Meanwhile, the patient is the one who suffers. The final decision—whether it’s approved or denied—won’t be known until that whole process is complete, and only then can the surgery be scheduled.

That’s what causes the abrasion: the administrative burden, the delays, the unclear requirements, and the possibility of denial at the end of a long process. And that’s still the case in many areas today.

Q. So because of your prior experience with payers, in this particular case insurance companies, you chose to focus on prior auth with them. And there is the healthcare system in the loop, which is the physicians and the providers. How do you distinguish between the two? Because prior auth is important from both perspectives, right?

Siva: For the health plans, it’s a cost. The main reason why there is prior authorization is because, as we all know, healthcare costs have been going through the roof.
There is quite a bit of waste, and a lot of it is due to unnecessary procedures, unnecessary imaging. For example, there’s no need to do imaging if it’s sufficient to have just an X-ray, which costs like 50 bucks instead of a $1,500 or $2,000 scan.
Because of the excessive use of high-cost items, there’s waste in the system. Health plans, being the intermediaries, manage the dollars for employers or the government, like Medicare.
So one of their tasks is to control for this. The health plan’s viewpoint is to prevent unnecessary things.

Obviously, the physician thinks something is very important for the patient, and that’s where the tension is.
The reason we decided to go with the payer side is that payers have the volume, and a lot of things can be controlled from the health plan side using technology.

There’s no point in just speeding up the process on the physician side. There are benefits to it, but you’d have to do it for every physician office.
If you go to the health plan, you can address all of this in one shot.

Q. Awesome. So Shiva, you mentioned that you started in late 2019. So that’s actually before COVID, right? 

Siva: Yeah. That’s like three months before COVID. 

Q. And then COVID hit. It must have impacted your go-to-market and your plans. But you stuck to the mission. You have very good investors who’ve supported you in your journey.
Tell us a little about the bumps on the road, how you overcame them, and where you are today. How many employees, and how are you going about this?

Siva: One of the things is that we actually managed to partner with a large health plan—okay, Humana—it’s on our website. What happened was that I had hired like four people or so. We were actually working in a WeWork office in Boston in February and were in the process of finding an office and recruiting people, etc.

I remember in early March 2020, while working in the office, they called all of us down and said, “Hey, we found somebody with COVID today in the offices. So you guys have to go home. We will call you, and we’ll see when you can come back.”

That was the last time we all saw each other—the four of us. And we came home, and after that, we didn’t see each other for more than a year.

But then we changed our entire plan—worked remotely—and built the product out. They had a deadline of January 1, a client. So January 1st, 2021. We said, “We can’t just sit at home and wait for COVID to go. We need to develop the product and everything else.”

We actually took advantage of the remote situation because initially our office was going to be in Boston, and we were going to recruit engineers in Boston—everybody in Boston. But because of COVID, we said, “You know what? We can hire people anywhere in the country.” And so that actually opened up the pool for us. We went around the country and recruited people from all over.

Q. That’s amazing. And I understand you’re still fully remote, which is very different from many companies shifting to hybrid or back to the office.
So tell us—what’s the secret sauce for keeping people engaged? You’re up to several hundred people now, so how do you keep such a large team engaged remotely?

Siva: It’s not easy. By the beginning of 2023, when things were becoming more normal, we were already up to 400 people across the country.
We didn’t have a choice. A substantial number were in Boston, but that was only about 35%.

So we continued with the remote model but tried to make it more efficient. There are pros and cons. We manage it by making sure management and teams meet regularly.

Our travel budget is high, but since we save on office space, we spend on getting people together. From a management team perspective, we meet once a quarter.

We also have regular team meetings—sales, clinicians, operations, technology, AI, product—each meets in different parts of the country throughout the year. That’s important for building camaraderie.

Q. That’s amazing. And from a time zone perspective, since everyone is in the U.S., that works well. We’ll talk about expansion plans in a bit, but you just mentioned AI. Tell us how you’re applying AI, GenAI, and agents in your product development. Things are moving fast with GenAI.

Siva: In fact, from day one, our goal was—let’s try to provide real-time approvals instead of the usual five to seven days. At the end of those five or six days, if it’s going to be approved anyway, why not do it immediately if the information is there? So we focused on how to approve things faster.

We found that at the end of the prior authorization process, 80 to 85% of requests are usually approved. So we said, let’s focus on that and use AI to approve—not deny—because denial still needs to be reviewed by a nurse or MD. So we focused first on solving that piece.

Today, we approve about 80 to 85% of requests in real time. That’s where AI comes in. We use AI in six or seven different ways on our platform. One of the main ones is this: we get the EMR or medical record from the provider’s office and ask what service is needed. Then, we analyze the unstructured data—diagnosis, patient history, etc.—and determine whether the treatment is clinically appropriate based on certain policies.

For that, our physicians review the algorithms to ensure they’re clinically sound. We have about 50 physicians in the company across multiple specialties. They review the information and help us encode that into the algorithms. It’s a painstaking process, but that’s how we reached 80%, and we’re still improving.

If there’s any doubt about a request, it goes to an MD. We never use AI to deny care—we leave that decision to physicians, who then communicate with other physicians. That’s one big area where we use AI.

We also use GenAI for scheduling patients, retrieving missing information, and automating tasks like converting faxes into structured data. We have intelligent agents that complete entire workflows. Summarization is another area—we use GenAI for documentation and generating letters. We’ve been an AI-native company from day one.

This has helped reduce abrasion because users know that 85–90% of the time, they’ll get an answer immediately. That’s a huge win—they don’t have to wait or reschedule.

We do quarterly user surveys. Our NPS is between 65 and 67—very high. Providers are saying, “Okay, someone is finally solving prior auth,” and that’s one of our biggest outcomes.

For the remaining 15% of requests that still need more review, we’re now working to bring that timeline down to one or two days using AI. We’re able to summarize and present all necessary information so physicians can quickly review and approve it—or reach out to another doctor for a quick consult. So AI is helping us shrink that review time, too.

That’s how we’re deploying AI across the board.

Q. Very interesting. Siva. So that brings me to my next question actually, that when you consider the benchmark of companies or your landscape in which you are doing your competitive positioning, are there any other large players that are in the same space and different and unique and how do you position yourself?  

Siva: The process has been there for more than 30 years. So there are legacy companies that have been doing this for health plans. Yeah. So this is not a new process, right? We didn’t invent this process.

They’ve been doing it, and they are the ones with seven-day, 40-day turnarounds, paperwork, old technology—you’re seeing all of that. So we are completely disintermediating them. We’re creating a completely new category, where we’re actually differentiating ourselves from them.

We’re kind of coming in and changing the way things are being done in this industry.

Q. That is great to know. So, I think we have covered a lot of ground Siva. Any other closing thoughts or any other information or news that you would like to share with the audience? 

Siva: I know that there is a lot of press around prior authorization. To listeners—especially providers and patients—almost everyone goes through this. Just know that companies like Cohere are now using AI to solve the problem. Relief is on the way.

Subscribe to our podcast series at www.thebigunlock.com and write us at info@thebigunlock.com 

Disclaimer: This Q&A has been derived from the podcast transcript and has been edited for readability and clarity.

 Q. Hi Alvin, welcome to The Big Unlock Podcast. It’s a pleasure to have you on board. As you might be aware, this was started by my colleague Paddy Padmanabhan from Damo Consulting, and we’re building upon what he left us as his legacy. I’m Rohit Mahajan, Managing Partner and CEO at BigRio and Damo Consulting, and I’m also the host of The Big Unlock Podcast.

Alvin: Rohit, thank you so much for having me on this podcast. I’m excited about the interesting topics we’ll be chatting about today.
So yes, happy to give you an introduction and some sense of where I came from and what I’m interested in.

My name is Alvin Liu. I was born and raised in Hong Kong. I came to the U.S. as a teenager to attend a boarding school in New Hampshire. After that, I did most of my schooling on the East Coast. I’m a practicing retinal surgeon. I did my ophthalmology residency and retinal fellowship at Johns Hopkins Medicine, and I stayed on as faculty.

I actively practice and take care of patients with a variety of retinal problems. Outside of my clinical work at Hopkins, I’m focused on artificial intelligence in several areas.

Within Johns Hopkins Medicine, I wear several hats. First, I’m the inaugural Director of the Gills AI Center at the Wilmer Eye Institute—this is the first endowed AI center at the Johns Hopkins School of Medicine, made possible by a generous $10 million donation by Dr. Gills.

Second, I’m a clinician-scientist involved in the development of clinical AI tools.

Third, in recent years, my focus has been on the implementation of AI tools for both clinical and operational purposes at the health system level. I’m sure we’ll dive into specific examples later today.

And fourth, I’m involved in AI governance. As you can imagine, there are many developments in AI in healthcare. In response, Johns Hopkins Medicine recently established a leadership team to oversee AI efforts across the entire health system, and I’m part of that team. I’ll be happy to talk about the AI governance work we’re doing at Johns Hopkins.

 Q. That’s amazing, Alvin. I wonder—with so many responsibilities, how do you even find time? Do you sleep at all? 

Alvin: I do sleep and try to get seven to eight hours sleep every day. I think that’s extremely important because, I myself cannot think very well if I don’t get enough sleep. So, I do put a premium on the amounts of sleep that I end up getting.

 Q. That’s amazing. So tell us, Alvin—you studied here on the East Coast and you’re a practicing physician. What attracted you to technology, especially emerging technologies, and when did you get involved with it? Also, talk to us about some of the work you’ve done in this area.

And even before that, if you’d like to talk about the health system itself, the geography, and the kind of patient population it serves, feel free to do that as well.

Alvin: Sure, I can start by talking about how I got involved in AI.
Near the end of my clinical training, around 2017–2018, I first got started with artificial intelligence. That was when a specific kind of AI technique called deep learning really started gaining traction.

Deep learning is the underlying architecture that powers much of what we know as AI today in 2025. It’s especially good at two things: image or video analysis, and more recently, natural language processing through large language models.

Back in 2019, most deep learning applications in healthcare focused on image analysis. As a retina specialist, I’ve always worked closely with images. If you look across medical specialties, radiology and ophthalmology are the most image-intensive, both in research and clinical care.

That’s why, when you look at AI research and real-world implementation today, the two medical fields leading the way—in the U.S. and globally—are radiology and ophthalmology.

What really got me interested in deep learning’s application to ophthalmology, and to medicine more broadly, was a study published by Google a few years ago. They showed you could train an AI model to predict someone’s age, sex, blood pressure, and smoking status just by looking at a retinal photograph.

That’s a superhuman capability—no doctor can do that. That one paper convinced me that AI would change medicine and society as we know it. And that’s something I want to dedicate the rest of my life to.

 Q. That’s amazing. So, tell us a little more, Alvin, about Johns Hopkins as an organization and the kind of patient population you serve. And then we can dive into some of the use cases you’re seeing or currently working on. 

Alvin: I’ll start by giving a sense of what Johns Hopkins Medicine is about, and then we can dive into specific examples.

Johns Hopkins Medicine is headquartered in Baltimore, Maryland. As an integrated health system, we operate six hospitals and around 50 outpatient sites. We serve a wide range of patients, most of whom are urban residents. Over the past several years, we’ve been working on a variety of AI initiatives. I’ll give you two specific examples.

The first is a clinical one—the deployment of autonomous AI for diabetic retinopathy screening, which we started in 2020. This is a significant application. When this technology was first approved by the FDA in 2018, it was the first-ever fully autonomous AI system in any medical field to get FDA approval. So my field, retina, actually made history. A recent study published in the New England Journal of Medicine AI showed that this technology is now the second most widely used clinical AI tool in the U.S. I think it’s a great gateway example to explain the broader medical AI ecosystem.

The idea is simple: everyone with diabetes should get an eye exam once a year. Diabetic retinopathy is the leading cause of blindness in the working-age population globally, and it’s expected to worsen with rising diabetes rates. It’s also well studied—we know that annual screenings, early detection, and timely treatment are effective and cost-efficient in preventing blindness. However, the challenge is that even in the U.S., only about 50% of patients with diabetes undergo these recommended screenings each year. The rate is even lower in many other countries.

Autonomous AI changes that. Traditionally, a primary care doctor would prescribe medication and manage diabetes, but eye screening required a separate appointment with an eye specialist, which creates friction. With autonomous AI, screening can now happen right in the primary care office. Imagine going in for a routine visit—your vitals are checked, medications refilled, and now, photos of your retina are taken. These images are analyzed in real time by an AI model in the cloud. Within a minute, the AI autonomously determines whether or not you have diabetic retinopathy.

If the answer is yes, you’re referred to an ophthalmologist. If no, you’re done with your screening for the year. We started using this at Johns Hopkins in 2020 and reviewed the data to evaluate its impact. The result? Yes, it worked. We saw improved adherence to the annual screening guidelines.

When we looked closer, the greatest improvements were seen among historically underserved groups—African Americans and Medicaid patients. The positive impact was outsized for these communities, and we published our findings in Nature Digital Medicine about a year ago. The second example is operational—using generative AI for revenue cycle management.

For those unfamiliar, revenue cycle management is how health systems like Johns Hopkins get reimbursed for the care we provide. It’s complex and involves many steps and a lot of paperwork. Traditionally, automation efforts have relied on older machine learning approaches like robotic process automation (RPA), which require a lot of rule writing and don’t handle exceptions well. This is where generative AI, particularly large language models, shine. They are adaptive, understand text and unstructured data, and can handle edge cases much better.

We’ve used GenAI specifically for prior authorization. It has significantly reduced the time needed to complete and submit each case, making the process more efficient overall. So, these are two real-life examples—one clinical and one operational—where we’re currently using AI at Johns Hopkins Medicine.

 Q. That’s very interesting. So, I have just some curious questions. Alvin, on the first example I. That you talked about in the primary care physician setting, that a patient can go and get their eyes checked. So, does it need specialized equipment at this time, do you think? At some point in time, it may be that I can just use my iPhone camera and or look into some kind of a kiosk. And, you know, kind of get it done at the airport or, you know, I always look into this when I do the security clearance. 

Alvin: That’s a great question. You’re touching on a really important point—the nuts and bolts of implementation. Implementation is key when it comes to scaling any kind of technology, including AI.

The short answer is yes, it does require some specialized equipment, but these are very common. In short, you need a way to take a picture of the back of the eye, which we call a fundus camera. These are already widely used by ophthalmologists, and there are many different brands and models. So, if you step back, there’s already an existing supply chain and industrial process in place for producing these cameras.

Now, the traditional cameras are desktop-based. They’re not very portable—they’re a bit heavy, and you can’t easily carry them yourself. But their footprint is relatively small—about two feet by two feet—and they can sit on a mobile table. So they’re easily accessible, and the image quality is quite good.

Of course, there’s been work on developing more portable cameras, and many of those already exist. You can even use an adapter with a smartphone to capture retinal images. So the technology is there.

However, in real-world settings, most of the AI models for diabetic retinopathy—especially the ones used in clinical deployment—are designed for use with the more common desktop-based fundus cameras. While they’re larger, they typically deliver better image quality, which is why they’re still preferred.

 Q. And then a curious question on the prior auth side—are you implementing and experimenting with prior auth across the board, or is it for a certain set of disease conditions, CPT codes? And then, is that a software that the team has developed, or is it something you’re using from the outside in? 

Alvin: That’s a great question. So, what you’re getting at is the nuances between the different service lines—who would benefit from prior authorization or not.

Broadly speaking, there are certain fields that require a lot of prior authorization, and that’s how insurance payers do utilization management. And I’m painting with very broad strokes here.

Typically, the service lines or medical specialties that require prior auth tend to give out more expensive treatments—things like infusion medications in oncology or dermatology, or in our case, retina. We do a lot of injections into the eye—what we call intravitreal injections—for diabetes and age-related macular degeneration. These are examples where, because the treatments are expensive, they’re more likely to require prior authorization.

So when we did our pilot at Hopkins, we focused more on those specialties that require a lot of prior auths, versus ones where the care typically just goes straight through without it.

But that’s a great question, and you’re absolutely right—the devil is in the details. Even for a relatively specific step in revenue cycle management like prior auth, designing a pilot that makes sense, that demonstrates ROI, and establishes relevant KPIs—requires a very deep understanding of how medicine works and operates. And not in a vacuum.

 Q. So shifting gears a bit, Alvin – with the macroeconomic factors now impacting the whole ecosystem, including digital health (which is a very large part of the U.S. economy, as we all know) – what are some of the things that you feel are coming in the near future?

Alvin: I’ll answer your question from two opposite ends of the spectrum. First, from the startup angle—because in my role at Hopkins, I end up interacting a lot with startup companies in the AI space. And then I’ll speak from the enterprise perspective.

So on the startup side, I think one of the common mistakes startups make in the healthcare AI space is not considering—or not understanding—the reimbursement issue from day one. And I think that’s the most important thing.

One could argue that healthcare AI is still a very new field, so the payment mechanisms in the market aren’t yet mature enough to handle an influx of new products. It’s a tough situation, honestly, for healthcare AI startups. If you’re on a founding team that doesn’t have a deep understanding of how medicine works, you probably don’t know what a CPT code is, or how that’s how services get paid for. If you want to get a CPT code, very likely—especially if you’re in the AI and medical device space—you fall under the FDA’s purview.

And if you want FDA approval, we’re talking about $3 to $5 million off the bat. One mistake I see is startups being hyper-focused on building the product—both in terms of execution and how they spend their funding—without accounting for or budgeting for that FDA process. And even if you’re lucky enough to get FDA clearance, then you have to think: are there existing CPT codes that will reimburse you for the AI service? Very often, there are not. So then you have to go to the AMA to negotiate a new applicable CPT code.

That process takes a long time. And even if you succeed in getting a new CPT code, there’s no guarantee the payers will reimburse you. And even if they do, the rate might not be financially sustainable.

So from the startup side, you really have to think long and hard about your reimbursement pathway. Of course, there are other ways to get paid—not just through CPT codes—but that requires a deep understanding of healthcare business models. And in some cases, you may need to invent a new one.

Now, on the enterprise side: AI is here to stay. But for health systems, it’s chaotic. We—as an integrated health system—get many, many sales calls from AI companies every day. It’s a crowded, noisy space. That’s why having a robust AI governance structure that looks at multiple aspects—clinical, operational, ethical, financial—is absolutely necessary. And I think Johns Hopkins Medicine is one of the first major integrated health systems to give this serious thought.

It’s still evolving. We’re learning. But building a thoughtful and industry-friendly governance system is critical. And if you zoom out even more—on a very macro level—the billion-dollar question is: how will value-based care and AI come together? These are two very big trends that will intersect soon. What that intersection looks like is going to be very interesting.

 Q. That’s very good insight, Alvin. So could you talk to us about any digital health programs that have been implemented and that you’ve been involved with, which improve access to care—or any other examples you’d like to share from the digital health side? 

Alvin: The example I would give is the autonomous AI for diabetic retinopathy screening program. Yeah, that’s a good example. We already talked a little bit about it. What we learned is that, again, like 80% of a successful program is all about implementation and how you execute things.

So, for example, even if you have a successful screening program at the level of primary care, you still have to figure out how to get the patients who screen positive to ophthalmologists. That’s a different line of work.

You can extend this analogy to other areas as well—for example, in omics. Just to set the stage, omics is a relatively new field that connects biomarkers found in the eye—mostly retinal biomarkers—with systemic health conditions. I’ll give you a couple of examples. Right now, we can already use retinal images paired with AI to predict someone’s future cardiovascular risk, risk of kidney damage, or even dementia.

So, I think diabetic retinopathy is just an early example. We’re going to see an explosion in the adoption of omics. But the question is: even if you have an AI-based omics screening program in a community or primary care setting, and you identify patients at risk for various systemic conditions like Alzheimer’s or cardiovascular disease—what do you do next?

How do you set up a workflow to get these people to the subspecialists they need to see downstream? That’s still in the works. It’s very fluid. But I think that kind of thinking—being able to implement and execute things efficiently at scale—is going to determine the success of a lot of AI programs, especially when it comes to AI in omics.

 Q. Yes, that’s amazing. So, Alvin, we talked about governance a bit. How do you structure prioritization and funding, and what kind of operating models do you look at? 

Alvin: Sure. I’m happy to talk about that. I’ll take a step back and give you a brief background on how this all came about.

Back in 2024, the executive leadership at Johns Hopkins Medicine started a task force to develop an implementation strategy that ensures Hopkins becomes a global leader in the responsible use of AI.

One of the key tenets the task force identified was that they wanted this to be a clinically led responsible AI program—meaning physicians like myself would and should play a major role.

The task force then identified seven core principles that are critical for responsible AI: fairness, transparency, accountability, ethical data use, safety, evidence-based effectiveness, and so on.

From these, we identified several implementation plans. A key one was to establish a governance process and framework that would integrate with existing governance structures. As a result, an AI oversight team was created. It’s an eight-person leadership team drawn from across the health system. I’m one of the eight, and we have purview over all things AI-related across clinical, imaging, and operational domains.

In a nutshell, what we’ve developed is a standardized framework for how AI vendors should interact with Johns Hopkins Medicine. So, for example, if you have a clinical AI product and want to engage with us, there’s a standardized intake process. First, you need to find an internal partner at Johns Hopkins who will advocate for you.

We then have standardized questionnaires—what is the tool used for? Do you have data cards? Model cards? What’s the expected ROI? How do you demonstrate it’s safe? And so on.

Based on the nature of the tool—whether it’s clinical, operational, or imaging—the application gets routed to different sub-teams. Then there’s an internal review committee that dives deep into the responses. We grade them, bring them back to the committee, debate, and often go back to the vendors with follow-up questions.

Ultimately, the committee does an up-or-down vote based on a variety of criteria and decides whether the tool can be implemented at scale across the enterprise—or not, and why.

 Q. That’s very robust process Alvin. Thank you for sharing such good examples, thoughts and advice so far. Any final closing comments? I think we are coming to our end of our conversation. Any other things that you would like to bring up? Any announcements, news items? Or anything else that you would like to share that’s upcoming on your horizon?

Alvin: What I’d say is that, at a high level, most people agree that AI is going to change medicine—and society—as we know it. The train has left the station. It’s no longer a question of whether we’ll adopt AI, but what the future will actually look like.

When it comes to healthcare specifically, it’s one of the most heavily regulated industries—and also one of the most personal. At the end of the day, we’re in the business of taking care of people and reducing suffering, and there’s a deeply human, emotional component to that.

I do believe that, for the good of humanity, we need much more collaboration in this space. And in particular, I see venture capital and startups as major engines of innovation.

What’s been missing—but is starting to improve—is a strong connection between the VC/startup world and integrated health systems. I think that relationship needs to get better. In the U.S., integrated health systems deliver the majority of care. So whether startups like it or not, their products will ultimately have to go through these enterprises.

That said, health systems don’t move as quickly as the tech industry. And that’s understandable—but I also think there’s room for improvement, particularly in how quickly decisions are made. Technology is evolving at an exponential rate, and AI is no exception. Things move fast—and for good reason.

So, there’s work to be done on both sides. I’m hopeful that we’ll see much stronger and closer collaboration between startups and health systems in the near future. If that happens, I think a lot of good will come out of it.

Subscribe to our podcast series at www.thebigunlock.com and write us at info@thebigunlock.com 

Disclaimer: This Q&A has been derived from the podcast transcript and has been edited for readability and clarity.

 Q.  Hi Patrick, welcome to the Big Unlock podcast. Great to have you with us today. I’m Rohit Mahajan, Managing Partner and CEO of BigRio and Damo Consulting. The Big Unlock podcast was started by Paddy Padmanabhan, founder of Damo Consulting, and it’s been a successful series with many healthcare leaders. Great to have you here.

Would you like to introduce yourself to the audience? 

Patrick:  Sure. I’m Patrick Mobley, CEO and founder of Vivid Health. A bit about my background—I’m speaking to you today from Raleigh, North Carolina. I grew up around healthcare; my dad was a physician. I eventually found my way into the startup world. I was an early employee at a company called Evolent Health, where I did a bit of everything. I started on the clinical side, ran their Virginia market, and oversaw value-based care operations.

Then I moved to Aledade, where we aggregated independent providers into shared savings arrangements. We grew from about 20 practices to 220 while I was there. Next, I joined Bright Health and ran a large health plan in the Mid-Atlantic. Across all three companies, I saw what high growth looks like—but I always wanted to build something myself.

Eventually, I connected with the leadership at Redesign Health. I had hired a company that spun out of them. For those who don’t know, Redesign is a venture studio based in New York, backed by General Catalyst, UPMC, CVS, and Aetna, among others. We came up with the idea for Vivid, which really reflects a lot of that experience. I know we’ll get into the details, so I won’t spoil it all now—but it’s really focused on making clinicians’ lives better. 

Q. That’s really interesting, Patrick. You’ve had such a deep career in healthcare. What was the early catalyst that set you on this path?

 Patrick: As I mentioned, I grew up around my dad’s clinic. It was a very familiar environment.
I saw how hard he worked—the long hours, the behind-the-scenes frustration with paperwork and administrative work. It was inspiring to watch, and it made me want to solve some of those pain points. Thinking I could build something that might help with the things that gave him headaches was definitely exciting. 

Q. That is awesome to know. So, you mentioned Vivid already and it came out redesign. So Patrick, please tell us more that. What is your thesis at Vivid Healthcare and and what kind of problems are you solving and what are you working on?

 Patrick: In my prior roles before Vivid, I worked closely with nurses or had them report to me. What stood out was how much time it took to call a patient, assess them, build a care plan, and follow up. These are four essential steps in any risk-bearing organization looking to manage cost, high-risk individuals, close care gaps, and improve risk adjustment. I kept thinking—how do we leave the nurses with nothing left to do but provide care? And when I say “nurses,” I also mean LCSWs, community health workers, the whole care team.

So, in partnership with Redesign, we started exploring what generative AI could do—specifically, how we could automate tactical workflows already used in care organizations. That’s exactly what we’ve built: from intake and referral, where an AI agent can call, text, or email a patient, to assessment, care plan generation via a large language model, clinician approval, and automated follow-ups. Whether the patient is with the practice for a day or six months, our platform supports it all.

Q. That’s impressive. What makes the platform so unique? You were also one of the earlier adopters of GenAI in clinical workflows. Was it hard to incorporate? And what results are you seeing now? 

 Patrick: It definitely wasn’t easy—but we had great partners, like your team at BigRio, to help build it out. We focused on covering a wide range of specialties and conditions. For nurses, it’s about understanding both mental and physical health needs, gathering that data, and turning it into a care plan—while always keeping the provider in control.

In fact, we actually own the trademark for “Provider-Led AI.” We strongly believe that no matter how helpful AI is, the final say should always rest with the clinician. Our platform lets AI agents handle tasks like calling patients, enrolling them in chronic care management, or conducting assessments like OASIS in home health. It builds care plans and allows nurses to focus on care coordination.

One of the most important aspects was making sure we could scale. If we were effective at engaging, assessing, and managing patients up front, then scaling the backend workload was critical. We wanted to amplify our nurses and care teams—individually—but also make sure they weren’t overwhelmed. That’s why we designed the platform to automate follow-ups. The agent can call, text, or email patients. Nurses don’t need to do anything manually. They simply turn on the platform, see patient stratification, view notes from calls, and take action from there. It delivers scale and reach—whether you’re a risk-bearing or home health organization—that you just can’t achieve with any other platform.

Q. So, is that where the positioning of the platform is also Patrick? Tell us a little bit more about who are the kind of potential customers or clients or users of the software platform.

 Patrick: Yes. There are a few distinct markets, though I often describe our platform as the perfect puzzle piece for any organization.

One key market is the primary care space. It’s ideal for risk-bearing organizations, but even those that aren’t can still use it to deliver extra care and generate additional revenue.
We can deploy our AI agent to enroll patients in chronic care management, conduct assessments, and complete all required documentation for chronic care, annual wellness visits, and transitional care management. We’re seeing a 100% reduction in outreach specialist labor using our voice AI tool and over 50% cost reduction compared to competitors.

We also target post-acute care—specifically home health and hospice. These settings have some of the most burdensome documentation requirements. In home health, for instance, the OASIS form is 27 pages with 200+ questions. Nurses typically can only see two patients a day because of this.

We’ve automated that entire process. When the nurse enters the home, the OASIS answers are already received, the care plan is generated, and the visit can focus on actual care—not paperwork. We’re seeing over 50% reduction in documentation time, which directly impacts revenue. If a nurse can see even one more patient per day, that’s a significant gain.

Hospice is also going through a big shift to a form called HOPE, which is like a shorter version of OASIS. We’re applying the same technology there and expect similar results.

We currently cover 100 conditions across 16 specialties. That’s generated interest from palliative care providers, non-skilled nursing organizations, and even some specialty groups. Once they see the platform in action, it really resonates. We’re proud of what we’ve built.

Q. That’s amazing, Patrick. And I know you’ve built a robust chronic care management capability across many disease conditions. You also mentioned a bunch of surveys—that’s your proprietary IP, right? So, that is what you built early on the core of the system. So, could you describe that a little bit more in detail on how it adds value and what it actually does?

 Patrick: Yes—and there’s a bit of a story there. Many organizations, rightfully, focus on five big conditions like CHF, COPD, diabetes, depression, etc. But I always felt that if a patient has COPD and also a kidney disorder, that second condition could significantly affect their overall health—physically and mentally.

So, we designed our system to evaluate patients across a broad set of conditions, not just a narrow few. We also wanted our surveys to assess both mental and physical health. So we ask questions like, “Do you have chest pain or swelling?” but also, “Do you have anxiety about paying for your meds?” or “Do you have transportation and social support?”

This creates a holistic view of the patient. The feedback we’ve received is that when patients interact with a clinician directly, they may feel pressure to answer a certain way. But when we deliver the assessments through text, email, or voice, patients respond more honestly. Nurses tell us the responses they get are clearer, more detailed, and more accurate than before. That’s been very cool to see.

Q. So when you are thinking of this in a larger setting, Patrick, obviously you might need to think about how it integrates with the systems organizations might already have in place. So what is your approach to that? How do you make it easy for your customers to use it?

 Patrick: Yeah, I think there are three paths there. One, just straight out of the box, the platform works really well as a standalone. It can do everything we’ve talked about so far—it does really well.

Second part of that answer is we use FHIR server—that creates a data standard for us to push and pull information from and integrate, frankly, with most EMRs. So we’re fully capable of integrating with just about every EMR in the market.

The third, and it’s the most interesting, is we partnered with an organization called NO2, which is part of something called a QHIN. And I know I’m getting kind of technical, but QHIN is a Qualified Health Information Network. What that is, is an interoperability layer to our platform that allows us to push and pull data from just about every EMR instance in the country.

For example, every single hospital in the states of Washington and Oregon is on this QHIN. So today, while we may not be directly integrated into OHSU in Portland, we can access their EMR—we can push and pull data, we can make requests. And I really think, like bigger picture beyond Vivid, that capability is going to be table stakes for any organization entering the digital health space.

Q. That is great to know. So tell us where you think the future is, Patrick, and what are some of the early findings that you have from your client implementations, and where are you headed in the next few weeks or months? 

 Patrick: Well, hopefully a lot of growth. We’re certainly seeing a lot of interest. There’s no limit of folks that want to talk to us, engage with us, and understand what the platform does.

I think what we’ve seen—and it’s been kind of entertaining to watch—is when we go through our demos and what the platform can do on the voice side, we’ll have them call the agent—her name is Sage—and just the faces light up. Their creative juices start flowing. They start thinking, what are the many different places they can deploy this agent?

You’ve gotta think of her as like an employee that doesn’t get tired, that can work 24 hours a day, that you only have to train once. And it’s incredibly powerful. So I think all phone calls that don’t require clinical decision-making will ultimately be done by agents like this long term—and across specialties, it doesn’t really matter.

Then the third thing—more going way out in the future—I think that at least within our platform, we’ll start to deploy different types of agents. So we talked about voice agent, which is one type. Another is one that will take all the data we’ve acquired—and maybe we’re not directly integrated with the EMR—but we are using the agent to go into that EMR and deploy data into specific sections.

I was at a conference last week and part of the conversation was, does the EMR just become the file cabinet for everything, but all the action happens in applications like Vivid? The idea is that because of these QHINs that I referenced earlier, it’s going to allow pushing and pulling data, and the agents can take it and put it into the specific spots it needs to be.

So maybe the EMR stays the source of truth or system of record, but the actual technical capabilities and advancement—and frankly, the efficiencies that AI will bring—will live in a layer above that. And that’s where the clinician does a lot of their work. I could see that definitely happening, because I think agents will be able to operate everyone’s computer at some point.

Q. Yeah, that’s true. Patrick, just shifting gears a bit—because you mentioned value-based care, and that is something you’ve been very closely associated with—what are the macroeconomic or other trends that you’re seeing, whether it’s value-based care becoming more prevalent or anything else coming our way?

Patrick: Yeah, so I think, having lived in that world for many years, I don’t think it’s going to entirely go away. We’re not at 80% market saturation with primary care being in a risk-based arrangement. I think the MSSP numbers are around 40 to 45%, somewhere in that range.

What’s interesting is, pre-AI, the way organizations worked with independent primary care—or even a health system—was that you’d deploy a lot of nurses and staff to find the sickest of the sick, manage them well, try to bend the cost curve, and then make money based on how much you saved.

A lot of those organizations had deal terms where, for every dollar, 50 cents went to the company and 50 cents back to the provider. I think the revenue opportunity for the provider is going to go up because the cost of providing those services is going to go way down, thanks to AI.

So instead of deploying an army of people into an ACO, the agent can make all the same phone calls, do all the same engagement, at a fraction of the cost. And now it’s going to look a lot more appealing to a primary care doctor who, instead of making 50 cents on the dollar, can make 80 cents.

As those models mature and the technology merges with them, it may accelerate—but we’ll see. There are contingencies, but there’s definitely a path that could be really interesting.

Q. Right. Any other changes that you’re seeing that might impact the business model—or anything else coming up in the future—that you’d like to share as part of your closing remarks? We’re getting to the end of the podcast here. 

Patrick: I think it’ll be interesting to see whether it’s federal or state governments that ultimately dictate individual AI regulations in healthcare. You’d prefer it to be more federal than state, or else you end up with 50 sets of rules that every company has to manage.

Had this discussion last week—you want a standard. You want everyone playing by the same rules. And if it doesn’t happen fast enough on the federal side, states are going to figure it out themselves, which could lead to unintended consequences for organizations wanting to operate in multiple states.

That’s not just true for us—that’s true for OpenAI too, who may have different rules in 50 different states. So, having a set of standards and regulations to help manage what’s coming—not just for Vivid, but AI in general—is probably something we should all be keeping an eye on.

Subscribe to our podcast series at www.thebigunlock.com and write us at info@thebigunlock.com 

Disclaimer: This Q&A has been derived from the podcast transcript and has been edited for readability and clarity.

Ritu: Hi Shekar, welcome to The Big Unlock podcast. It’s really nice to have you on the show. We’re now in our sixth season, with over 150 episodes and a great listener base. We’re excited to have you here and look forward to a lively discussion.

I’m Ritu M. Uberoy, Managing Partner at BigRio and Damo Consulting, and also a co-host of The Big Unlock podcast. I’d like Rohit to say a few words before we hand it over to you.

Rohit: Short intro from my side as well, Shekar. I’m Rohit Mahajan, CEO and Managing Partner at BigRio and Damo Consulting, based in Boston. As I mentioned, we’re very happy to have you on the podcast.

Shekar: It’s a pleasure to be here, and I’m looking forward to a great conversation with both of you.

I’m Shekar Ramanathan, Executive Director of Digital Transformation at Atlantic Health System. We’re a seven-hospital—soon to be eight—health system based in New Jersey, with over 20,000 employees and about half a million patients in our ACO.

We’re a fairly large organization, and I’m responsible for the integration of digital, AI, and data—essentially, using technology to solve business and clinical problems.

We’re really focused on working outcomes backward: identifying what we want to accomplish, the metrics we aim to achieve, and then developing solutions to meet those needs.

Ritu: Great. Great. We all know generative AI is the buzzword right now. I just attended two conferences—HIMSS and Human X—both heavily focused on AI, generative AI, and now the latest: agentic AI.

We’d love to hear your thoughts on these technologies, your AI maturity model at Atlantic Health, and where you feel the organization stands in terms of AI maturity. Also, tell us about some initiatives you’ve led or are currently working on.

Shekar:  Sure. I don’t think I’ve been at a conference or had a talk in the past two years that hasn’t been about AI. Even if it starts with something else, it ends with AI. It’s definitely the hot topic no matter where you go.

For us, it’s very much focused on the “what are we still like…” I think our strategy hasn’t changed in terms of what we’re trying to do from a digital or organizational perspective. What we’re really trying to do is see how we align that with the new capabilities that are emerging—continuing the same business strategy, which is always somewhat challenging because people keep asking, “What is our AI strategy?”

I used to say our AI strategy is our business strategy—it’s not any different. But I’ve somewhat changed that over the past two years. Now, our AI strategy is really about building the framework. It’s about enabling the business and understanding where the technology is going so we can be in a position to fully leverage it.

That means setting up the right governance and the right processes to monitor AI—to ensure it’s the right solution and at the right cost. I think that’s been a challenge for many organizations. You see a lot of piloting—we started there too, with plenty of pilots. But scalability becomes a challenge.

Then comes the question: who are the right people to use these tools? How do we extract value and not just get excited by the art of the possible?

We’ve done a lot of what other health systems are doing—ambient voice, note summarization, routing of messages, and so on. But we’ve also done some novel things, like focusing on a virtual MA where we use a quasi-agentic approach. Not fully agentic, but using some of those tools for outreach, patient communication, helping manage care—with escalation to a clinician or care provider whenever necessary.

We’ve seen a lot of success. At the same time, we know things are changing rapidly. That’s probably one of the biggest challenges—not just for us, but for healthcare in general. We’re used to a fairly slow process—just selecting a vendor, signing a contract, going through implementation—it’s usually a long timeline.

Now, by the time you select a vendor, the next one is already out, doing it better. So how do we shift to being truly agile in our thinking? Solving problems in smaller pieces, being more iterative—those have been some of our key focus areas and challenges.

Ritu: Great answer, Shekar. It’s amazing that you mentioned the top three use cases—ambient, scribing, and message in boxing. You mentioned that going from pilot to scalability is a challenge. Could you pick one of these initiatives and talk a bit more about your experience—specifically, the timeline and what that looked like?

To give some context, I attended a talk at HIMSS about innovation using GenAI, and one of the takeaways was that culture can be both an enabler and a barrier. You have to be open-minded and ready for these accelerated timelines, but if there isn’t buy-in across the organization, change becomes difficult. Would love to hear your thoughts on that.

Shekar: Absolutely. I think one of the key things is that people often start by piloting with a highly engaged, super excited group—folks who really want to leverage the technology. They get great results in that small setting. But when it’s time to scale, it becomes difficult to replicate the same level of adoption, utilization, and value.

We’ve had more success when we focus on narrow workflows—being very intentional about what problem we’re trying to solve. That allows us to have the bandwidth to do proper education, integrate the technology into the workflow, and not just introduce a tool that people are playing with.

With a lot of the GenAI tools, what we’ve seen is a burst of initial excitement—people want to try it, see what it can do, maybe generate a song about COPD, and that’s great. But then reality kicks in. When people are back to caring for patients, they ask: Is this actually saving me time? Do I know how to write a good prompt? Do I understand when it’s useful—or not?

We’ve had more benefit by being very prescriptive: “Here’s the use case, here’s the prompt, here’s the button to click.” That helps users adopt the tool more effectively and ensures they see real value.

It also helps with cost control. These tools can get expensive as they scale, much like how cloud costs were a challenge to predict a few years ago—but AI takes that challenge to another level. So we want to manage rollout carefully, ensure users understand the benefits, and then scale in a controlled, thoughtful way.

Ritu:  Okay. Thank you. Next, we would like to talk about the role of a Chief AI officer and if Atlantic Health has a Chief AI officer, and what do you think would be the pros and cons of, you know, that role and what are your viewpoints about that role?

Shekar:  So, we don’t have a Chief AI Officer per se. We have a lot of people who kind of wear the Chief AI Officer hat—myself included—where part of my role is to drive what our AI strategy is. And that means different things to different people, right?

For us, it’s really about how we lay the infrastructure so we can support the different ideas and initiatives that are coming in. It’s also about identifying what’s truly different between an AI project versus a digital project or a regular technology project—what do we need to think about differently?

Then, depending on whether it’s a business use case or a clinical use case, we need to make sure we’re bringing in the right stakeholders. Especially on the clinical end, we need to have the right clinicians involved and really understand the potential impact—and make sure we have the right processes around that.

So it ends up being a group effort. I definitely see the role evolving, but the question is whether it becomes a purely dedicated position or if it stays tied into roles like data leadership or digital transformation. I think it really depends on the organization—what makes sense for them, and the size and scale of their AI ambitions.

That said, I do think we’re going to see a lot more Chief AI Officers emerge, especially as the space grows, the opportunities expand, and there’s a greater need for structure and oversight.

Ritu:  Yeah. What we’ve seen with other folks we’ve been talking to is that the real need for a Chief AI Officer, like you said, is around strategy. Multiple people can wear those hats and do the work, but the real need they felt was around governance, ethics, bias, and some of the other thorny problems that crop up.

Would you like to talk about any challenges you’ve faced—primarily in terms of AI implementation—like hallucination, bias, or data integrity? And how you’ve overcome those challenges?

Shekar:  Yeah, and I think one of the biggest challenges is really understanding what a lot of these vendors are doing, especially given the pace at which innovation is happening. And then there’s the challenge around black box AI, right? I mean, that’s the so-called “vendor secret sauce.”

But at the same time—are they really doing something truly innovative? Are they actually getting results? How is it working? Do we know what data they trained it on? Patients in a rural area may be very different from those in an urban area. Or maybe the model was only trained on adults and not pediatric patients. There are so many variables that can introduce bias.

There are also a lot of things that can make a model either work for you or not. So how do you really evaluate that? Right now, a lot of these companies are coming to us without the level of research and documentation we’re used to—things like clear evidence of efficacy or quality outcomes.

It’s sometimes hard to get that information, because this is a new, shiny object and people are excited about the art of the possible. That’s especially challenging for us on the operational side. People come in with a great idea, and they’re promised big results that maybe they don’t fully understand.

It’s like—someone says this tool is 99% accurate. But then, when you look at the positive predictive rate, you realize that nine out of ten times, it throws a false positive. So now you’re getting ten alerts for every one useful one.

Is that really helpful? Sure, it’s 99% accurate—but it shows up all the time, and that affects how people experience it. So we have to interpret that correctly and make sure the business is fully aware of what the actual experience will be—before we sign a contract, implement the tool, and then find out later that it doesn’t meet expectations.

Rohit:  Shekar, I’m very interested about your journey in healthcare so far. The audience always likes to know what got you started, what interests you. What are you thinking about the future as well? So if you can share with us what motivated you to take on this role and how you walked into the healthcare industry segment, and where are you headed?

Shekar:  It would be great. Yeah, no, absolutely. I’ve always been in the healthcare technology space. I went to grad school for medical informatics—back when nobody really knew what that meant. And now it feels like everything is kind of coming together.

I started more on the development and consulting side, working with a number of state governments to develop syndromic surveillance systems and similar initiatives. I also did a lot of research around patient experience during grad school.

Eventually, I ended up at Epic and spent some time there. That gave me a lot of exposure to electronic health records and large health systems. After that, I worked across several large healthcare systems—BayCare Clinic, Kettering Health Network, Wake Forest Baptist Health—and eventually landed here at Atlantic Health System.

Over time, I’ve been focused on clinical applications, digital transformation, generating value from data, and process optimization. And now I’m at a point where I can pull all those different pieces together and apply them more broadly.

I’m really excited about the potential of AI right now. I’ve been talking about the future of healthcare technology for a long time—what we could do with EHRs, how to collect and use data—and now it feels like we’re finally at a tipping point. We’ve spent years burdening our clinicians with documentation, all with the hope that it would one day lead to better care, and I think AI is finally enabling that transformation.

That’s why there’s so much excitement in the space. People are energized—maybe even tripping over themselves a bit—trying to figure out the right solutions. I share in that excitement, but I also want to make sure we do it right. We’ve moved at a glacial pace for a while, and now we’re ready to sprint—but we need to do it ethically and consciously, always focused on outcomes.

I feel really fortunate to be at this point—where everything in my background is converging, and I get to be part of pushing healthcare forward.

Rohit:  Absolutely. I’ll add one more thing here, Shekar—I think you’re right we are at an inflection point. I completely agree with you there. A lot of changes are coming at us fast, and we have to adapt and adopt the technologies that are actually useful.

From a patient perspective—since you mentioned Atlantic Health System is based in New Jersey and you serve around half a million patients—can you tell us more about that geography? Is your patient population very diverse? And is there anything you’d like to share about patient engagement?

Shekar:  Yeah, absolutely. So, we’re primarily based in New Jersey, with a bit of presence in New York and Pennsylvania, but mainly focused in northern and central New Jersey. We have 500,000 patients in the ACO, and even more overall.

New Jersey is very diverse, and that diversity really comes into play when we start talking about things like health equity and digital engagement. One of the challenges is figuring out how to reach patients with varying levels of digital literacy.

Interestingly, the patients who could benefit the most from digital tools often face the most barriers to accessing care in general. So, the question becomes: how do we make it easy and accessible for everyone? There’s a portion of the population that will be really excited that there’s an app for everything, but the ones who really need it may not be thinking in those terms. So, it’s about how we reach out, educate them, and truly enable them to be partners in their own care.

Ritu:  Yeah, I think we’re almost at time. Shekar, would you like to share any closing thoughts? It’s been a really engaging discussion. Maybe you could tell us what you see as the top three future trends with generative AI?

Shekar:  I think the next big thing’s gonna be agentic AI. It’s the next evolution as things become more and more “autonomous.” I think we’re gonna see a lot of kind of a hybrid, kind of a mix of agent traditional generative AI solutions.

I think a lot of this comes down to how do we start removing a lot of kind of that burden of healthcare. We’ve spent a lot of time asking providers to do more and more, and there’s where people are excited is that, maybe more of that can get offloaded so clinicians can really focus on direct patient care and a lot of those other things that may be more administrative or tangential.

That’s really where I think we’re gonna see a lot of technology that’s going to be able to kind of help solve some of those problems. 

Subscribe to our podcast series at www.thebigunlock.com and write us at info@thebigunlock.com 

Disclaimer: This Q&A has been derived from the podcast transcript and has been edited for readability and clarity.