Newsday: How PhD-Level AI Agents are Changing Healthcare with Dan Schubert

[00:00:00] Revuud is the healthcare IT platform built for speed and control, giving your team direct access to over 4,000 pre-vetted professionals, epic Cloud, cybersecurity, you name it, without the delays in overhead of traditional staffing firms. The Revuud platform handles the contracting so it's fast and compliant and low risk.

Your team saves time, your budget goes further, and you stay in control. Learn more@revuud.com. That's R-E-V-U-U d.com.

I'm Bill Russell, creator of this week Health, where our mission is to transform healthcare one connection at a time. Welcome to Newsday, breaking Down the Health it headlines that matter most. Let's jump into the news.

Bill Russell: It is Newsday and we are joined by the usual suspects and Dan Schubert, CEO, co-founder of Revuud the Usual Suspects of course being. Sarah Richardson and Drex de Ford.

You, do you want adjectives? I've been giving you [00:01:00] adjectives now. You like hunger for the adjectives? No, we're not. No, we're, no, we're not. Or actually don't, oh no, I, no that, that, I'm gonna go back. This is one of the key use cases for AI right here is coming up with adjectives to utilize.

Dan. Welcome welcome to the show.

Dan Schubert: Thanks for having me. Really appreciate being here.

Bill Russell: we had our state of the CIO call today with our partners and shared a bunch of things and. Top three conversations were were ai re-skilling and ret staffing or re-skilling and up upskilling our teams and then cybersecurity and the rise of resilience really is a key.

And I sort of made the joke early on that we could have talked about AI for, you know, an hour of that could have been the whole show. I wanted to talk about AI longer. We're gonna talk about AI and here's how we're gonna do it. I wanna go to a LinkedIn post and isn't it a LinkedIn post that was published eight hours ago?

How's that for timely? Matthew? Call EVP, chief Information and Digital Officer at In Nova Health. [00:02:00] Wrote this morning, and I like this. And this will be a backdrop for some of our discussion around ai. He says dare to think crooked. The secret sauce of healthcare's next leap. What if straight line thinking that's dominated healthcare for decades is the very reason we're stuck?

Linear logic built our systems efficient, predictable, profitable. But the future, it demands minds that zigzag obsess over the overlooked and shatter assumptions. We didn't know we had a dash of diversion. Thinking isn't just tolerable, it's the catalyst for breakthroughs that redefined care. Take some of our recent AI pivots at Inova.

We were grinding on traditional models, tweaking variables by the book. Metrics, looked solid on paper. Then a rogue insight, systemness mission values, questioning the sacred sequence of typical inputs and outputs. The script was flipped. Suddenly accuracy [00:03:00] jumped, readmissions dipped, and we unlocked major efficiencies, not from more data or fancier tech, but from embracing thought patterns that diverge from the norm.

Relentless hyperfocus pattern spotting in chaos, comfort with complexity that makes most teams flinch. This isn't fluff. McKinsey reports. 85% of digital initiatives fail from human blind spots. Why group think being enemy number one, non-traditional styles cut through that unrelenting questioning, fueled ideation that sparked wild prototypes and persistent rigor.

Test stress, tested them to indestructibility in the future world where AI mediates every patient touchpoint. We need thinking that bends rules without breaking ethics. It's how we build equitable algorithms, resilient platforms, and access that feels human, not robotic. Healthcare loves to pat itself on the back for innovation, but are we brave enough to invite [00:04:00] the discomfort of diversion thinking into our war rooms to reward the questioners over the yes people?

The evidence says yes, systems do. See outsized returns faster, pivots real impact. Provocative may be necessary. Absolutely. Who's rethinking their team's cognitive mix today? Drop your takes on the comments below. I use that as the backdrop to say it's it's pretty natural for us to take a new technology like AI and look at what we're currently doing and say let's apply it to this.

Is that the right approach here?

Dan Schubert: I mean, I think at the end of the day, yeah, it kind of does come down to that, some of that stuff. Right. You know, one of the things I'm interested in learning more about on this whole journey with CIOs is like, how are they prioritizing what they're working on? Right. Some of that's fairly easy, right?

Um, But some of it's not. So it's like. How do you address all the different areas that AI's gonna [00:05:00] tackle for the healthcare system? That's a pretty complex issue. You know, from my perspective, there's so many things that it can apply to. So how do you prioritize that?

Drex DeFord: That's a good question. Drex, are we at, are we at some point where we start to think about this as like, AI is just like the oxygen in the air that we need to be able to.

Make progress anywhere. Like, it's not a separate thing, but it is everything. It is all the things I like is the new way I like Matt's post. And part of the reason I like it is that I think, you know, these are my words. I feel like what he's calling out also is the. This is the way we've always done it, and we're super proud of the way that we've always done it, and we have emotional attachment to that way of doing things.

And what we need now as we look into the future is a willingness to sort of dump out the toy box and start over. And it's really hard. I don't know a lot of people that have really been [00:06:00] trained to kind of say. We're gonna go in and take everything that we've done. We're gonna disrupt ourselves and throw it all out and build something brand new that's just better because we've got the capability of doing it now.

Sarah Richardson: Well, here's the thing is how is ai, how is all this automation? How's even rethinking some of our processes that exist today? Changing the patient experience because we're still not fixing the backend issue of access and insurance and all the other components that say, Hey, I can get you, I can get you.

An appointment on the books twice as fast, but you still have to wait six months. Now if some of the ambient capabilities are opening up, more capability for the doctors to have more time, which you don't lead with that by the way, in your use case. Hey, Dr. Russell, you get to see five more patients a day.

But we're speeding up a lot of things, but we're not fixing the intrinsic problem that still exists in that you're managing multiple portals, you've got all these password resets, you can't get an appointment anytime soon. You, your care [00:07:00] coordination is all over the map. If you have multiple specialists, like we haven't really fixed that for the patient yet, and I'm grateful for all the things that are coming forward.

I'm hopeful that continues to be a lean in for most of these conversations.

Dan Schubert: No, I think that's a great point. Because my immediate reaction to that is like, I do think it's going to help. The process of patient flow, if you will, through the hospital system and things like that, because I think we're gonna be able to, through things like genic AI and predictive ai, we're going to be able to do things remotely that we've never been able to do before.

As simple as a, an online kind of interaction or checkup without going into your doctor's office. And maybe that's done with a, with an AI agent instead of a person initially. There's just a lot of things I think that we're gonna, that we almost can't even think of it right now.

Things are moving so fast. I think we're still in the very infancy of this and the more on the generative AI side versus some of [00:08:00] the more advanced stuff that we're gonna get into. But I really do think you have to kinda like, obviously crawl before you run, but, and that, that's the phase we're in.

I think it's I think it will eventually impact greatly the, you know, the speed as to which a doctor can process patients and things like that. But, you know, I think it's going to lead up to everything up to that point.

Bill Russell: You know, here's how I'm thinking about it. Unlimited number of PhD level candidates that are about this big, right?

They're about this big. So you can put them in a little camera. And you put them in a room and they're PhD level candidates and they say, Hey, that person's getting up. That's a fall risk. Or they say, oh, the nurse just walked in. Let me document that for her. Oh, the doctor just walked in. Let me document that for him.

Oh, he's going to want. She's going to want, you know, predictive, they're gonna want this information. Let me pull this information up. Supplies, oh, they wanna see it on the tv. Let me connect to the TV and push that up. So when you think about a hundred PhD level [00:09:00] candidates about this big sitting in the room, this is something we couldn't do before, right?

We could never hire enough people to be everywhere, all at once. Same thing on cybersecurity. I could never hire enough people. Same thing on, on data stewards, data quality, whatever. I could never hire enough people. All of these things I'm going to be able to get with just power and a subscription and some thinking around what do I want this agent to do?

Is it perfect today? No, it's not perfect today, but I will tell you this. My gosh. And I feel like I'm saying this over and over again, but because I'm spending so much time coding right now, I'm seeing the progression of these coding models and it is astounding. Like it used to make a bunch of mistakes.

Now it like predicts the mistakes it's gonna make and it doesn't make them. Now it's, you know, and the other thing is in some of the things I'm using. It's using multiple models, so it's like using a model, different models to like one, to [00:10:00] test, one to do qa, one to, it's doing a lot of these different things.

Well, those same things apply. When you think about unlimited number of PhD candidates, this PhD candidate says, I think this is a fall risk. This one says, no, he is not a fall risk. That person's just, you know, turning over in their bed and you know, these three agree. And they go, no, it's not a fall risk.

Don't make the, don't make the alert. That's how I'm thinking about this. It's like unlimited number of PhD level candidates that we can program and deploy. Everywhere. Research projects on everything instantaneously. Yeah, exactly. I wanna come back to the question that you asked Dan, which is, how are CIOs approaching this?

because I'm not sure Matt Cull is in the majority, and I'm not sure what he described as in the majority because. What he described is not only Matt and the IT team, he described a system where it sounds like the system is [00:11:00] taking an approach that says, Hey, we're going to rethink everything.

We're gonna look at everything a little different. because Matt from the CIO chair. It's probably not going to be able to do much unless he has from a readmission, from the areas he was talking about without support of the entire executive team. So how are they approaching this? How do you bring this conversation into your organization in a way that your organization is I don't know.

You bring them along so that they understand what you're talking about for starters, and then bring them along to the point where they can go. Okay. Wow, man. If you gave me a hundred PhD level candidates right now, where would I deploy them? I'd deploy them here, and here. That's a great exercise, by the way.

because I could sit there and go, all right. That's the thing

Drex DeFord: I can wrap my head around, right, is like, if I had a hundred free PhD, where would I have them go? Do studies right now to take a look to figure out how we could, like, throw processes that we have today completely out or rebuild them in a whole new different [00:12:00] way.

I mean, Sarah, back to your point about you know, I can make an appointment right now, but it's not until six months from now. It could be a capacity issue. It might also be. There's some really slow process I have to go through to get approval from my insurance company to be able to have that appointment.

So I can't really even make the appointment. I can hold the spot, but I don't know that I'm really allowed to have the appointment. What if it was an agent talking to an agent and like that could just happen and it could happen because. The insurance company would also understand that if they don't do, the sooner they do that appointment, the better it is for the patient.

It's less likely that they're admitted into the hospital, which is the expensive part of the system. You can have a whole set of little bill, little mini PhDs out there looking at all those different diagnoses, all those different things that are going on with patients that can keep them healthy as opposed to in the system and driving bills and consuming capacity that could be better used somewhere else.

Bill Russell: We had a really [00:13:00] interesting internal meeting this week. There was different groups that actually different subsection of people who were actually in it. And one of the things that sort of dawned on me as we were having the conversation is it was really creative. And the reason it was really creative is we have automated a bunch of things and now people have margin, right?

Yeah. People have margin. When people have margin, they can actually think. It's not like, oh my gosh, if I had five more minutes, I'd just work on that project, or the next project or the next 20 projects that already lined up for me. What if we created margin where people actually had time to look at healthcare and the delivery of care and patient perspective and all those things, and actually ask those questions that were generative and they go, Hey, maybe.

We have the patients do this instead of this, and we can make a much better experience. A lot of times we just don't have the time to ask those questions.

Drex DeFord: We don't have time to think about them. And Dan, I know like you, you're thinking about [00:14:00] those same kind of things when it comes to you know, how folks hire outside staff that a lot of this outside staff is also tied to capacity problems and other things too.

And a lot of it's about looking at this whole problem differently.

Dan Schubert: we're trying to attack it from an application standpoint. Like, you know, on the true software side, we're trying to make processes a lot more efficient. You know, cost efficient and process efficient. And put a systematic approach to how you manage the resources that really you can redeploy very quickly on projects.

Because I do think that when you really think about what I guess the CIO's going through and how many things that you can apply this automation of AI and things like that to, it's a daunting list of stuff that you've gotta approach. So. Prioritization is key. But then what? So what we're doing is trying to make it easy for the right talent to be deployed at the right time for the right period, take off as much burden off of the health system as possible, and off that hiring manager [00:15:00] so that they can not only find that resource, but get them working on their project as fast as possible without the hiring manager have to, having to worry about all the.

The little processes that go into that from interviewing to contracting with, to onboarding and all that kind of stuff. So our, you know, our view of it is like how we can just impact the CIO in that capacity. And deploy the right talent, trusted talent, you know, bench talent that you really know that you can redeploy quickly.

I'm fascinated by what we're gonna see over the next five years and how these CIOs are gonna navigate prioritizing all of these things. I mean, I think you guys would know, right, pretty quickly, like what is the biggest thing on my biggest pain point right now on my plate? And then how can ai, or if we understand what.

AI and that kind of stuff is gonna do for us. You know, how do I tackle that now and how do I find the right resources to deal with that?

Bill Russell: Well, I,

I know what G would be doing. because what G would be doing is [00:16:00] he would have this entire process stood up. If anyone doesn't know Gee is is Sarah's alter ego.

He is the complete nerd that she hired who just does amazing stuff with technology. But he would have this whole process stood up so he could stand up agents tomorrow and we would just, and you would be out talking to the business and you'd come back and say, okay, gee, we need five more.

One needs to do this, and this. And he would go, done. Yeah. And Sarah would look like a hero because she'd come back out and say, you know what, normally you would hire five people but we just, or hire three people or two people. And we just deployed those agents and they are now reviewing every fax that comes in.

They're doing this, they're doing this, they're doing this. And you don't have to do that anymore. because they're learning, iterating.

Sarah Richardson: We call that hold my beer

Dan Schubert: Hold my.

Sarah Richardson: What's so interesting about that though, the maturity of the organization, this is why I love what Matt wrote about, is that Matt has the gravitas inside his organization to have those different teams think about things differently.

If you go in and you completely [00:17:00] transform an organization, you create a baseline where everything we've just talked about can be true, the organization still has to want and know how to do it. And so the technology being an enabler only goes so far as the fact that our legal risk, compliance, security, and everybody else part of that equation too, and I love Dan, when you think about how you.

Provide talent to organizations. That critical thinking aspect of being able to use AI along with workflow. Come out with different outcomes for the organization. It is more than just filling a seat anymore. It's what are the additional capabilities they're bringing to the organization, especially when they source talent.

The expectation is you're sourcing talent because they provide a little something extra that you can't from within, and either you keep it longer term or it becomes part of that project, but that leap behind needs to be that the organization is better off than you found them because of the type of talent you're helping them source for a Myriad of projects.

Dan Schubert: Absolutely. And you know, we try to we try to educate on like, look, there's. [00:18:00] nothing better than like trying the right resources out, finding the ones that are truly, you know, beneficial to the organization and then bringing them on later on if needed as an FTE. But we're trying to bring the whole cost structure down so that you can be more iterative and like, use contingent labor for three to six months, get those projects done.

You know, one of the things with AI that's gonna be so transformative is that you're gonna build the right. Again, I keep talking about agentic ai, but it's going to be game changing because it is going to learn how to take care of itself in the simplest form. So it will learn when, there's a mistake made.

It will iterate from that. It will change direction. It will correct. It will push, you know, a simple kind of like example is. You know, we're a software company, so we, you know, develop code that has to be written for one thing and then it goes through QA and then it has to go through the different paths to get through to production, post-production, tested and all that.

[00:19:00] Well, down the road, not too far from now, a year from now, probably maybe less. agentic AI is gonna manage that entire process. There's not gonna ne necessarily be a developer or QA analyst doing that kind of stuff. It's going to be done and managed and put through to production. So it's just a simple example of how these things are going to like manage themselves.

So then it's more about how you continue to train the models, things like that, to then, you know. Continue on, but that's a lot less effort at that point once it's built and built correctly.

Bill Russell: So let's end with this drex. I'm gonna give you the last word on this. So we just described deploying a hundred PhD level candidates, which rely on compute power storage.

Network, all that other stuff. You know, when we had an outage before, it was one thing that, you know, it caused people some consternation. But essentially, if we had an outage, now we're essentially saying that a hundred people didn't show up for work.

Drex DeFord: it's a [00:20:00] whole new version of like a supply chain problem.

Right. When we went into the pandemic, what we realized is that it wasn't just. Once we were outta stuff in the warehouse, well, we had built warehouses that only held like two days worth of supplies instead of the 45 days worth of supplies we used to have three or four or five years ago. And a lot of that was because we could rely on the supply chain.

The trucks were gonna be there every day. Delivering tomorrow's supplies and we didn't have to worry about it, and they were gonna be able to pick it up from the port where it was gonna be delivered by a ship that arrived every day and suddenly the whole system was disrupted. So I think there's these unintended consequences.

When we think about resilience. We talk about like, how does the system work when something in the system breaks? This is all great. I love this idea of a bunch of PhDs running around, PhDs being able to spin up their own little agent PhDs to work on side projects because they see something that's really interesting.

Go do [00:21:00] a bunch of research on that and tell me why that happens so often. That all sounds great. And it would be amazing as long as everything works. And that includes everything. Like the power stays on and the water continues to flow to the data centers and, you know, all of this stuff has to happen.

All those things that

Bill Russell: Are always gonna work. We just know they always

Drex DeFord: work. Just take them for granted

Bill Russell: until, Until they don't. Yeah. That's why we're building for resiliency. It's like you, you by design has to be part of it. Dan, always great to see you. Thanks for being on the show. Drex, Sarah, fantastic as always.

And everybody else. Thanks for listening. That's all for now.

That's Newsday. Stay informed between episodes with our Daily Insights email. And remember, every healthcare leader needs a community they can lean on and learn from. Subscribe at this week, health.com/subscribe. Thanks for listening. That's all for now.

[00:22:00]