No Pilot Graveyards: How Novant Health’s CAIO Turns AI Into Patient Outcomes

0:00We're going after several other conditions that the population at large either are ignorant or unreachable by normal human means.

0:10Peter High: Welcome to Technovation. I'm Peter High, president of Metastrategy. My guest today is Vijay Sankararaman. Vijay is the chief artificial intelligence officer of Novant Health, a not-for-profit health care system serving communities primarily in the southeast of the United States. The organization generated nearly$12 billion in operating revenue in 2025 and includes more than 43 ,000 team members, 21 medical centers, and over 900 locations. As CAIO, Vijay leads AI strategy, product development, engineering, platforms, and data science. His team works closely with clinical, operational, and technology leaders to reimagine how care is accessed and delivered while ensuring the new capabilities are developed responsibly.

0:49Peter High: Prior to joining Novant Health, BJ held product and engineering leadership roles at Walmart and Lowe's. In this conversation, BJ and I will discuss how Novant Health is using AI to make care easier to navigate, expand patient access beyond traditional hours, and reduce the administrative burden of clinicians. We'll also explore the role of autonomous agents in identifying and reaching patients at risk, how Novant determines when human involvement remains essential, and why the organization is concentrating on a select group of high-value opportunities. rather than creating a graveyard of disconnected pilots.

1:22Peter High: Vijay, welcome to Technovation. It's great to speak with you today. Nice to have you. Nice to meet you here. Today's episode is brought to you by Tynes 3B. Organizations are pushing teams to move faster with AI. That's creating a wave of AI-generated code and unsanctioned automations, leading to rising costs, security risks, and huge IT burdens. Meet Tynes 3B, a single, secure environment that enables employees to build with AI while giving IT and security complete control to run and govern that work. 3B enables effortless building. Describe what you need in plain English, and 3B builds a fully functioning workflow.

1:56Peter High: It uses IT-authorized connectors, meaning credentials, state-protected, and non-technical teams can innovate without taking on unseen risk. Also, with full visibility across every workflow, IT gets total governance, while features like Auto-Tune and Auto-Fix ensure workflows improve and self-heal continuously over time. Employees get the freedom to build, while IT gets the control to govern. No shadow AI, no data leaks. Tines 3B moves IT and security from blockers to enablers, giving teams the speed to innovate without compromising control. Visit the Tines website at tines.com. That's T-I-N-E-S dot com to learn more and try 3B for yourself.

2:35Peter High: And now on to the interview. Yes, it's a great pleasure. I'm so glad we have time for this conversation. Vijay, let's begin with Novant Health, if you don't mind. the company you help lead. For those who may be less familiar with it, tell us a bit about the health system and its background, if you would. Absolutely. No One Health is a Catalina-based multi-state provider system. 45 ,000 strong team members who represent about 20, 22 plus large hospitals and a number of other clinical outlets that we provide remarkable experience. We have a little over about 8 million patient encounters we make annually.

3:22Very delighted to be here to speak all the AI-led journey we are going through.

3:28Peter High: Speaking of which, you are the chief AI officer of the company. Talk a bit about your purview, a role that's emerging in some corners, not all, but clearly, obviously, within Novant Health. Talk a bit about the role of CAIO as you are applying it in that organization. Absolutely. Very humbled and privileged to represent the role for an amazing company here at Novant Health. So, Peter, for us, when we think about AI, our fearless leader, the CEO, and his amazing team, we look at the business as a whole, most important thing for our business is patient, remarkable patient experience. How do we bring that to life?

4:08How do we make sure that we look at all the clinicians doing their job, staying top of the license, bring that to life? So when we look at that, AI providing itself a transformative moment for us to reimagine a craft of rendering healthcare as, you know, situates very important. So me and my team are privileged to directly represent through the transformation office to the entire executive team. Peer groups run across the organization and we look at all the problems in its own merit and leverage where the automation can help to play that part.

4:49Peter High: And the organization also has a chief digital and Information Officer, Onyeka Nchega. Talk a bit about the interplay between your roles, if you will. Yeah, Onyeka and his team are a very healthy peer group for us. We work hand-in-glove with the team, very similar to some of the roles within the health informatics, some of the roles we work alongside the chief medical office practices, you name it. So So the practice of AI, we see it more than just a figment of technology. We see it more as an element of business transformation. So Onyek as a team, for example, for me and my team, we work together where we leverage EHR, where we leverage an element of data, where we leverage infrastructure pieces to it.

5:38Within the AI organization that I run, I have product engineering, I have a technology arm within my own arc that works all things system of the future with AI in it.

5:50Peter High: Let's talk a bit more about your organization. You've begun to describe that. Talk a bit about the team that you lead. You've begun to mention a little bit about their areas of responsibility. I'd be interested in the skills you have built. And also, as this role has come online, as I believe you were the first to hold the post, to what degree was it some combination of bringing in existing employees into your part of the organization versus recruiting others? Absolutely. No points for guessing. It's a very heterogeneous and a healthy mix of sorts. To begin with, we approach with a very problem-first mindset, which means a squad of my leaders or product leaders who obsess over what is the problem going on.

6:34So think of a product pillar that I have. And then I have a team that is pure software engineering, but obviously an AI-led. It's a team that completely built around AI-driven software engineering. We build, we integrate, we co-build, all of the above. So there is a very healthy engineering team that directly reports to me to sort of work on these things. Then there is a platform team. The platform team is the one that ultimately puts these pieces together, extremely working hand in glove with the IT organization, with the marketing organization, with who have you, to ensure that both the internal products we build, AI products, and the external products that we work alongside and integrate comes together.

7:25No matter what way we sort of touch a patient, touch a clinician or touch a team member, it is unified in a way. That's what the platform team gets together. Now, given all that, I also chair and run the AI Trust Safety Committee for the company. So there is a team that helps put that piece together in the safety umbrella, if you will, runs across. us.

7:51Peter High: A great overview, Vijay. Thank you so much for that. I wonder, as you think about the value that AI can drive in an organization like yours, I'd be curious about the things that have you most excited in terms of, as I say, value creation within a healthcare organization like Novant Health. What are some of the areas that you're pursuing currently? Yeah. So, you know, we have verticalized into three broad buckets. No particular order. First one, all about enriching patient experience. And I'll give you an example on each of these in a minute. Enriching patient experience. The second pillar focuses on internal, which is our clinicians, our team members, non-clinicians.

8:35The third one, all about operational excellence. How do we look at various facets of the operations, Think of finance as an operation. Think of, you know, your hospital operations. Wherever there is friction, wherever there is opportunity for us to mine better value out of that with automation AI, that's the excellence team goes under. So those are the sort of verticals I have. Given all of that, take about patient experience. I'll give you one use case, you know, which we've been rolling out for a while and rolling very well. We have amongst us population, a number of what we all call as a sort of a high risk or a surgical moment that, you know, comes across in some of our lives where it takes an element of effort across all of it.

9:26Think of hip and knee surgery. Think of hysterectomy or colorectal surgery. These are multi-day in the minds of a patient, how they sort of go through a battery of preparation to sort of, you know, get there. Then the day of the surgery arrives. And then there is a recovery period post the surgery that they go through. Given all that period of time, put yourself in the shoes of a patient. That's exactly what my product team and engineering team work together. Putting ourselves in the shoes of a patient, what are the elements of friction? Take a step back. We truly believe that the healthcare system over its time, over its existence of time, has amassed friction for a variety of reasons.

10:16All these friction have gotten to a place where the patients have come to, let's call it a grim reality of acceptance. Now, I'm not referring just for no-one health, I'm referring for health systems as a whole, at least in the U.S., right? Now, when we look at that in the example I just mentioned, a patient who is about to go through, let's call it a hip and knee surgery. Let's just anoint the person to be in his early 60s, you know, 11 p.m. in the night. The person has a question, do I bathe before my surgery tomorrow, the day after? Do I, what sort of a meal do I, you know, take or not take?

10:59Is that going to help or hurt my bowel moment? You get the idea, right? And the day after the surgery, coming away from it, when do I, what side of the pillow do I keep? I mean, trivial questions to many of us, but not so much for the person who's sort of enduring through that, right? Can they make a phone call? Absolutely. we are all about remarkable experience. But given the time of the day, given the availability of

11:24Peter High: what they can do, they may not be necessarily motivated to make a phone call or in a position to. Now enter the AI care component that we have rolled out for all these surgical scenarios. 24-7, an agent is helping to help answer a question, voice or text, doesn't matter. I mean, whatever the modality is. Now, at any point in time, either the patient or the AI agent has the ability to bring the human in the loop to answer things that perhaps the moment calls for a human in the loop. The fun fact there, Peter, is ever since we rolled out earlier this year, we are finding little over 40 % of the conversations, very, very rich conversations that's happening with the patient and the AI agent after hours, the traditional hours and on the weekends.

12:14Now think about it. In the traditional world, they would have to sort of get on a line, get a human and nurse on the line. And sometimes a question may be beyond the nurse's capability. They have to get somebody else to answer. Or even the ugly moment of, do I even ask somebody about, can I wear a particular clothing before the surgery, after surgery? Some people now with the AI agent, all those hesitancies taken off the door. So we're finding a lot of traction with that. Now, with the second pillar, I talked about clinical and team member enrichment. In that space, our prized position is our remarkable experience delivering clinicians, nurses, right?

12:57We have a few thousands of those. My job running the AI team is to figure out every possible way to minimize, if not eliminate, the administrative burden they go through day in and day out. So they can keep on top of their license and focus as much, if not all the time, on a patient, which no machine can do, only a human can do. So leave human to do what a human can do and the other way around with the machine, right? So we have introduced a few years now, ambient listening, right? You know, that helps to reduce the administrative burden that a clinician has to go through. Keep an eye contact and keep the perspective automated insights.

13:37Now, recently, we have also launched what we call as an in-basket alleviation. All that geared to optimize the clinician away from all the administrative gear that holds them back in providing a remarkable service to its fullest extent, right? The last but not the least pillar, I'll give you all things around operational excellence, right? Now, operational excellence is hinges on the premise that our job is to render the remarkable experience, reducing the clinician burden, but also wherever we have friction in the supply chain, in the revenue cycle, in the hospital operations, where we can have a better way to, for example, create a better draft appeal letters for a patient who's been denied for an appeal that the payer system made for them?

14:34How do we automate and make it easy for them? How do we look at thousands of contracts we have with our supply? With the variety of clinicians we have, how can automation at speed, at scale, can help synthesize that and say, well, I can look through a thousand contracts for a particular set of products I'm bringing to the hospital and make it a lot more optimized and value added for the hospital and for the patient at the same time. These things were simply not possible with the likes of a human alone doing these things. At some point, humans can only go so far. AI crosses all that threshold.

15:16Peter High: Fantastic examples all. Thank you so much. Double clicking a little bit on a couple of those points, if I may, Vijay. Talk a bit about the pathway to getting the voice of the patient, if you will, to understand pain points or opportunities like the ones that you've articulated? How do you collaborate either directly or indirectly with them in order to derive some of the insights that you've described? Yeah. I mean, first things first, we work very closely through our clinician population, understanding what the underlying problems are. Where does the friction begin with? Knowing what the problems are, variety of ways we do to source that information.

15:57When we get that, we first test it out, for lack of a better word, in a small market, in a small clinic, it's a small population, make it as better, close to perfection. I would never call anything perfect because perfection is always the one that keeps us going. Get us as close to as good as it before we sort of take it from that and roll everywhere. Now, I will caution the way we do, Peter, is that We are strong believers in our AI organization for vital few versus trivial meaning. We do not want to have a pilot graveyard of sorts where we try out everything everywhere and nothing takes off from the airport pilot.

16:38So we are very curated about it. I'll give you another use case on that. About a year, year and a half ago, we did some predictive analytics to understand with the hypothesis that we have certain high risk conditions. And I'll explain what they are in a minute. That sometimes, and scary though, oblivious to the population themselves. Lung cancer is one such. So we did the study and within our own segment of population, we had a little over 20 ,000 population. that potentially could have that risk of, you know, getting into a lung cancer. And like I mentioned before, oblivious to that. They probably have conditions that may lead up to it.

17:25And our goal, obviously, is to help reach to them and say, hey, you may want to get, you know, checked out. You may want to get tested. You may want to sort of, you know, get a scan in place. So, early detection is the best thing than anything else. 20 ,000 people. Peter, it is not humanly feasible to make that number of phone calls at a faster time. Now remember, time is against all of us in that scenario, right? And we made human phone calls like many other provider systems in the countries are. We made a grand total of a few hundreds across a few months. And this is not an ability for us to make the phone call.

18:04It is an ability for us to how influential we can be We didn't get the person at the time they're available. Now, what we did was we developed a conversational AI that now has the ability to not just make a phone call, but connect it to our electronic health record, connect it to the physician information, make a lot more informed, purposed call that allows the patient to be, you know, truly provided with information why the call needs to be made. And what help can the call be, can the agent be beyond just a phone call, help them schedule, help them provide with other contextual information, giving them a better chance to sign up.

18:51Happy to tell you that we are well past the halfway mark in a very short amount of time. The key I'm trying to make here is in this whole pursuit, our goal was not about can AI do or not do, not about what the technology element or figment can help or not. That's all part of the recipe. Most important piece is the goal here is clear. The problem is clear for us. How do I reach out with speed and with purpose and clarity with these patients? By the way, we didn't stop with lung cancer. We're going, lung cancer was an element of it. We got other, like an uncontrolled blood pressure. We're going after diabetes.

19:30We're going after several other conditions that the population at large, either are ignorant or unreachable by normal human means.

19:41Peter High: Powerful examples. The technology, artificial intelligence actually helping to save lives, no doubt. Thank you so much for sharing those, Vijay. You clearly have a product mindset, which emerged for you in among other areas of responsibility and past experiences at companies like Walmart and Lowe's, where you were a product and an engineering leader. And I wonder, what did you learn in a retail setting that serves you well in a healthcare setting? To what extent are the analogies do they translate across the fulcrum in what would appear to be very different industries? Yeah, I think…

20:18One of the biggest things that I've carried with myself across industries, to your point, consumer, retail, and obviously with now healthcare is doing technology for the sake of technology, rarely, if any, has a merit to it. But instead, pivot yourself, don the persona of the character that you're serving to. In the case of consumer, obviously, whoever is shopping or doing. In the case of healthcare, the patient out there. When we put that perspective on, all of a sudden the narrative shifts different. And we now look at less about a hammer looking for a nail, more about the problem takes on the precedence.

21:01And then you think about, is AI even the right thing to solve a problem? There are a number of, in fact, a harder part of my job is to say no when somebody approaches me to say, can I solve this with AI? It's not a question of can AI solve? Should AI solve? Is this problem even deserve a merit to solve? So those are the things we look at. Those are the lessons I've sort of brought with myself working across industries and particularly for the bigger ones like Walmart.

21:28Peter High: Vijay, now that you've been in healthcare for some time at Novant specifically, of course, I wonder what perspectives do you have about the areas within healthcare that will remain for some time uniquely human. You know, we're talking about remarkable advances that artificial intelligence is making and no doubt bolstering, even taking on perhaps some of what caregivers once had as burdens for them, for example, or tasks that were repetitive, for example. What do you see as the areas that where humans for the foreseeable future will not have competition from artificial intelligence? Yeah. I mean, living in healthcare, long ways to go.

22:06I mean, you know, end of the day, we are all about getting the right care at the right time for the, for the population who deserve that, right? Who need that. You think about that simple narrative I'll give you, you think about having a nurse, how much time, how much more time can a patient get from the care of a nurse? The answer lies in the question itself, right? There is not an automation that I can think of that is going to substitute that. Now, can there be automation to make life of a nurse better so their caregiving moment can only be enriched while focusing on the patient? Yes, but that doesn't deduct the presence or the purpose that in this case a nurse can do.

23:00I just picked nurse as an example. Obviously, we were working on a number of things to improve the life of a nurse while doing the job. The list goes on and on, right? Now, we can look at large and go, what purpose is that a care in its purpose of getting rendered to a patient goes through points of friction, like I mentioned earlier? And if those points of friction, by the way, friction comes not just internally, friction comes externally. Now, when you think about when a patient or a population, rather, has a need for care, some of the questions goes beyond just the physician themselves. The question could be, can I afford it?

23:42Can you tell me more about it? Do you guys have pharmacy? Can I afford the pharmacy? And the list goes on and on, right? And there are humans who help answer the question. And there are humans who help clear the pathway to physically, you know, render a service. the job of automation as I see as science evolves with time is increasingly deducting the friction that gets the care closer and closer to the patient making it much more easier so some automatable functions obviously will exist where you know time taken to go create a schedule time taken to reschedule and I'll give you another example we recently created an automated organizational agent that helps to intake a call from a mammography scheduling rescheduling like traditionally those calls occur during the hours of the day and a lot of times it ended in a hold time we called it called abandonment abandonment right you know naturally every healthcare organization every organization out there when they have a call that they take through a human, there is only so many humans you can deploy at any given time to take a call.

24:58If the incoming call outnumbers the number of human call agents that can take a call, there you go. You're on a wait, right? That's a simple, simple, you know, let's call it elementary math. That translates to friction to a patient who, let's just say, you know, after work, She's come home. She's in her late 50s. She's trying to take care of her family. She's got a mammography schedule the next day morning. She now cannot make it for a variety of reasons. She now wants to go move it to another day. In the process, she also has a question about the billing, either because of a prior one or the current one.

25:38All these runs in her mind. She's trying to make a phone call. Traditional systems, traditional pathways. We'll get her an answer. perhaps after a few minutes of wait, perhaps after a certain amount of time that has to be routed for another day, another call. With the agent that we now have, there is no wait time. We literally dropped the call abandonment rate from whatever higher number it was to near zero percent. That's simple yet powerful experience in a remarkable way given to the patient. No longer the wait time. right and and that person out there at five o 'clock in the evening or got you know or you know even if it's 8 30 in the evening she has that is where we are looking at essentially the idea here is where all friction exists in the collective ecosystem that we are we are working to essentially take it away give the patient what they deserve the most fantastic i wonder also

26:38Peter High: So how do you think about the calculus between buying and building of artificial intelligence and where buying? I'd also be curious about the ecosystem that you have developed of partners to help you deliver what you've described. Yeah. Look, you know, it's a very heterogeneous way the AI is building, and we are no different than that. What I mean by that is, I support a very humble and yet powerful team of AI engineers, obviously in partnership with the CIOs and his team. The technology build is not our shortcoming. Having said that, we are prioritizing where the build makes the most sense, where we can partner with a certain set of up and coming or strong build out, including an amazing partnership we have with our own EHR provider.

27:36All of that gives us a healthy meal plan, a healthy construct that says, hey, let's put it all together and have it all work coexist. Fortunately, the AI platform team I mentioned earlier, we have built out an ecosystem. We call it the Arbery ecosystem that houses all these build, buy and co-develop pieces together in a unified way, giving us the flexibility to choose when we need to choose. And by the way, the choice is also not regrettable choices. My point I'm making is that we could make a choice to build something because we're waiting on, let's say, in a particular scenario for the EHR provider to get that built out in a couple of years.

28:20When that comes around, we can quietly switch over and vice versa, right? So those choices allow us to go execute with right quality and speed without that becoming a bottleneck. So it's a very healthy, but the ecosystem belongs to us. We have unified it. That way, it doesn't become, you know, everything, trying to do everything else.

28:43Peter High: And now having occupied the chief AI officer role, I wonder how broadly applicable do you find the role to be? I realize it's an unfair question for you to say, you know, the specific companies or industries that it would serve best. But what are the circumstances under which you think an organization ought to contemplate bringing in a CAIO as a new leader within the organization separate from the existing executive team? Yeah, I mean, it comes down, Peter, to me, the DNA of the company and the DNA of the purpose they had wired in. there have been companies or industries where I have been part of both in my past life and I look around where the role for a chief AI officer subsumes under the broader technology organization.

29:35There have been scenarios where the other way around works out too. In the case with NoWantWareFC, we coexist as peer groups and both purposed to solve for the problem. It ultimately comes down to what the company is truly trying to solve for, where they fit in the spectrum of things in their own world of suite. Are you a product company or are you a service company? Are you rendering what sort of mass do you account for? How much of a commoditized, let's call it the AI offering that you want to use? And probably even more a philosophical question to trade is, what does AI mean to the CEO on the boat?

30:17Do they treat that as an element of technology and no more? Or are they treating it as an element of truly a transformative aspect that brings along technology and everything else to truly reimagine the craft of the business? The answer to those questions will give answers to what you're talking about.

30:37Peter High: Yeah, fascinating. I also wonder, given the pace of change currently, in your area of focus. I wonder, what do you read or listen to in order to stay current that you might recommend to others? Are there certain thought leaders, certain podcasters, authors that you find provide you the greatest sense of where things are going in the evolving state of the art related to artificial intelligence? Yeah, I mean, I'll start by sucking up to you. I listen to a lot of media's podcast, but in true honesty, the speed at which, Peter, the science of AI is evolving, it is arguably more than just mind-boggling to the point where it's more of a noise, right?

31:27But the reality, though, is applied science is different than the underlying science, right? Working for different companies in the past are now serving for no-armed health. For me, it is all about how do we apply, in this case, AI and technology? How do I apply AI and tech that moves the needle for the company in a much different element of stratosphere than before? in the sense it is all about compounding the general purpose of the company and challenging if the goal of the enterprise is to do X, Y, and Z in a certain number of time can the transformative moment with AI that is presenting to all of us cut the time down and increase the quality with speak that is is a broader notion of how we are thinking.

32:31Peter High: Well, Vijay, thank you so much for sharing your perspectives here. It's been fascinating to learn more about the remarkable work you and your team are doing in Novant Health, literally saving lives through the ideas that you're bringing to bear within the organization. I appreciate you taking time with me today and sharing your insights with this audience. Peter, pleasure to be here. Thank you for having me here as well. Thanks for listening to Technovation. If you enjoyed this conversation, subscribe and leave a rating. It helps more leaders find our podcast. Thank you, and I look forward to seeing you next time.