[00:00:00]

I am Bill Russell, creator of this Week Health, where our mission is to transform healthcare, one connection at a time. Let's take a look at what's working today.

Drex DeFord: absolutely not.

Sarah Richardson: It's, you know, to me it's a three way tie with governance. The never ending discussion, and also talent and upskilling relative.

Drex DeFord: It's like, what do we want ai, when do we want it? Ai I don't know. It seems like there's just Yes, it is. My gosh, we've turned it into a drinking game. I mean, it's really, we

Bill Russell: did turn it, we into a drinking game and at one of the places.

They started saying a one instead of ai because they're like, enough drinking, we're done. I was like, all right. It is amazing how much it's talked about though.

Sarah Richardson: Yeah. For governance, for staffing, for everything

Bill Russell: really. I, I would call it a fairly large health system, CIO, said in one of our city tour dinners that his CEO that afternoon had come into his office and said.

What are we doing with ai? We need to do more with ai, to which his response was as all of ours would be. What do you wanna do with ai? Like, yeah what's the, where do you wanna point that screwdriver? Like what do you wanna screw in with that? And he says, I don't know. I just know we need to do more ai.

I literally said it was like a, you've heard

Drex DeFord: it.

Bill Russell: It was like a me. I've heard

Like what are we doing with ai? I, and I think there's a little bit of a FOMO thing. Like I also wanna go to the conference and make a presentation about all the cool AI things that we are doing. So, you know, be prepared. If you haven't gotten that question, you're gonna get it.

Bill Russell: You know, Sarah, I throw out all the time with healthcare leaders.

You are not behind. Because they're all worried that they're behind. I'm like, you're not behind. But you could get behind, like you should be doing a whole bunch of prep work of training your team and understanding what's out there understanding what you're gonna need in the future. You know, it's, it's. it's

It's you know, it's not, you know, whatever. And it's going to work and then it doesn't work. And the person who fixes it is the night nurse who has, you know, 15 years experience and knows where, you know, all the data is and every, where all the things are buried. It's not the technologist who solves it, it's not the ai, it's not the data scientist.

It's the person who has deep knowledge, deep roots in the workflow and deep and knows the edge cases that's able to look at it and go well. I know why that's not working. It's not, this is why it's not working. It's not taking this into account. It's not doing this. It's just so we're, it's almost like, it needs context.

Sarah Richardson: actually takes me a little bit to what I would consider a prediction for 2026 more than anything else, is that I don't believe, honestly, the organizations that have the most AI pilots are gonna be the ones who do well to me. It's gonna be the ones who go after something that is very specific, trusted.

They can audit it, they can have enterprise scale in that pipeline because. We've heard a quite often this year, Hey, we have a head to head with six different partners for this specific solution, et cetera. I always question that because when you can transition from pilot to production, then that is where your governance is working, where your lifecycle management is working, or even some of the sustainability and the adoption for clinicians comes into for their workflows.

Because when you just go pilot without understanding, when it goes into production or gets scaled, then you just have. Pilots to go into perpetuity, and we've all been in the spaces where those

Sarah Richardson: Yeah, because you hear it and if you don't specifically say we're only picking one solution outta these six as an example, then you're gonna get into trouble.

But I do believe that our community and smaller health systems are gonna have challenges because the academic medical centers, they can use, you know, an agile platform to improve, whether it's core coordination, population health, whatever they're looking at for Constituency. But if people can focus on what the models can do and how they're governed and how they impact a workflow and how they integrate into care, then you're gonna be on the right track.

And those are the types of relationships you need to have with your partners or otherwise that really let you make a difference. So don't do 30 little things. Focus on one or two things that really are a problem for you and your healthcare system, and then get the right partner to help you do it really well.

And then that becomes a basis by which you can start to replicate that success.

Bill Russell: that's awesome. Drex number one, use case. People throw out, this is the easy one. What's the number one use case? Everyone throws out, Hey, are you using ai?

And where are you using ai?

Drex DeFord: I mean, it's ambient listening. Ambient listen. So ambient

Bill Russell: listening wins.

Drex DeFord: Everybody's doing it. Yeah. . Hey, here's the question I have for you, bill, though. I mean, it's kind of interesting, right? So Sarah, a lot of these things wind up being connected together. You've got people doing tons of pilots and working on lots of different stuff.

You know, you've heard me say this before, like I think it's okay to be in to moonshots, but what you really have to be into is moon landings. You gotta actually make some of these things stick. And work. Right. And that's a big problem right now. But it's tied into our inability to prioritize back to the governance conversation.

Is AI a feature of products that we use or things that we're buying? Or is it a product that we're. Building ourselves. Do you see a lot of builders out there or do you see a lot of people who are just like, turning on the button on Thursday when it shows up the app?

Bill Russell: Less than 5% are builders.

And so let's be real clear what builder means. It doesn't mean building out a foundation model, and it doesn't mean it doesn't necessarily mean coding a whole heck of a lot of stuff, but it does mean. Knitting things together. It's almost what we used to call middleware. It's identifying the different workflows that could benefit from things that you haven't done before.

We didn't have time to do that. It can do. Incredibly well and yeah, the

Drex DeFord: boring grindy stuff that we don't

wanna

Bill Russell: do. So we put it off. Yeah. His example was there's 15 years of documents in a repository that the organization knew it had a lot of value in that if they could just go through all those documents, but you know, it, it just would've been arduous to do that.

I think they were legal documents, if I remember correctly. And I think there's a lot of those use cases in healthcare. But we are very much focused on the. You know, what we're doing today and how we're doing it today, not thinking about what could be. I believe Sarah, you're usually the best student here.

The one that, that actually does her homework and whatnot. and the reason I say that is 'cause our producer, Holly, is fantastic. She sends us all these notes and all these things and Sarah's the great student and she goes out and she does all of her homework and in Drex and I sort of show up and go, oh, there was an email um, and that kind of stuff.

But one of the things she, she asked about was, you know, the stories from this year. I mean, what are the stories from this year around ai?

That was fascinating to me. The other piece that I follow pretty closely is the heavy infrastructure investment, because there is surprisingly. Modest application of ROI so far in the overall like space of ai. Because I mean, people say, why is it delayed? Well, you have the infrastructure investment, but you have to have the value creating applications and the leadership path to get there.

So it's almost like a self-funding tech bubble. You hear this a lot if you listen to Wall Street Journal or other aspects of. All the infrastructure investment's. Awesome. What happens if that ROI doesn't happen fast enough and it implodes in itself. becuase a lot of the capital going into it is the chip sets and the data centers.

Bill Russell: Let's talk about ROI and let's talk about let's talk about the deep fake stuff becuase it's so fun. Yeah it's, it's fun to a point. But, you know, Anahi in one of the panel discussions that we had was talking about this issue of we're hiring people from North Korea. We're hiring people. Who don't exist. And I mean, so this is, this is really a thing. This is like how they're infiltrating now is literally getting a job with the organization. For sure.

Drex DeFord: There's a whole series that I did this year that was called UnFake and a big part of that series was specifically addressing this North Korea fake employee problem.

It's actually an American who has lots of laptops that are running. So that to the hiring authority, it looks like the laptop is in the us but then it actually has some kind of additional software on it that allows. The North Koreans to run it. And the North Koreans aren't necessarily in North Korea, they're in Pakistan or Russia or other countries too.

And at the same time. They're often in there stealing data or doing other things that are bad for the organization. And it's not just like big companies have been through this and have figured out that they've been had but a lot of healthcare systems too have privately told me they've hired these fake workers and then had to sort of sort all this out later.

So it's a real messy situation and there are a lot of folks who are. Are wrapped up in this

Sarah Richardson: digital trafficking.

Drex DeFord: Yeah. Yeah. I mean, in a lot of ways it's like you're trafficking humans. But you know, this is, North Korea was sanctions against them for years and years, or starved for cash, and this is another way they've figured out how to get money into the account.

Bill Russell: Uncle Drex, you always tell the best stories and scare the living daylights out of me. I mean, it's a lot crazy story Health Systems just figured out. They had you know, five, six analysts working for two health systems. And they're like, yes, that

Drex DeFord: that's another whole big thing, right, is that people are double dipping.

I mean, once we had the pandemic and people went home, people are super ingenious and they started to figure out, well, I can work here and I could also take this job over there. AI definitely has helped that because there's. Those super genius people who are working two or three of these jobs have figured out how to use AI to do some portion of those jobs so that they don't they don't show up on the radar at their, at those health systems.

So they're wildly and

Bill Russell: they're using it to get another salary complete.

Drex DeFord: And I think Epic is sometimes the one that sort of figures out like, this person's epic.

Clearly ambient is helping with experience. We've I think we've heard that. I mean, the doctors like it, the doctors who use it really like it. But still we're at like a 60% rate. There's, for whatever reason, there's another adoption. This could be, you know, landing. You know, the moonshot and that kinda stuff getting over the hurdle.

But a lot of 'em are getting sort of stuck at about the 60% rate, but those who are using it love it. So experience, great access, cost, quality. when I'm thinking of ROI, there, there's, I was going down this case this path use cases, ambient listening. The second most common use case I'm hearing is coding and notes.

And when I think about those things, I've heard that none of these have like massive ROIs. They're all like, if they really do hard work, they get it to like, Hey, we know we should be doing this. It has a lot of other benefits, but the hard dollar cost case has been hard to, uh. hard to really put out there and I'm, you know, that's what I'm hearing. I'm curious if you guys are hearing the same thing. It's hard to come up with that, that hard dollar ROI on these things.

But that's more elusive because you had to spend $12 million in either data modernization and or tech debt just to get there. And the first year you've captured, let's just say 10%, like $1.2 million of incremental value. The gap is telling us there's a maturity mismatch versus a technology failure, but that's only gonna be worth that investment if part of the conversation is the long term.

Space that says we're not gonna get back into massive amounts of tech debt. We're not gonna get back into five or six applications that. Collectively do the one we've rationalized. But even we saw app rationalization takes a three or four year timeframe to really prove out its modeling. So you've gotta be really close to your financial team in a way that they're willing to accept.

Drex DeFord: It's interesting that Ambient has kind of created this situation where I think ambient listening gets in the door by saying, we're gonna improve the quality of life for the physician. And it's hard to recruit. It's hard to retain. You can, we've heard physicians in our events say things like you know, you ambient listening extended my career.

I was about ready to retire. Now I'm gonna stay around for a while longer. But it's not just the. You know, the soft ROI, the hard OROI of then now you can gather more data and organizing it and organize it better because you're using ambient listening, which allows you to code better because you've got the data so you can code and defend yourself when the denials come.

And then. You know, technology wise, keep going. You've gotta Exactly. You gotta get outta tech debt.

Bill Russell: I'd like for you guys to think about the most interesting thing that you heard from one of the leaders in our meetings with regard to ai. One of the, I and I like the cont, you know, I like the contrarian kind of things.

If not more work is if you get. And to be honest with you, our organization's designed this way. We have a lot of high level people. We don't have very many, you know, minimum wage kind of, people in our organization we're very top heavy, but we're utilizing AI significantly on, on the automation side and whatnot to handle all that.

All that stuff that would normally require I don't know about you, but you know, when they gave me an admin, I came from outside of healthcare and I came from a consulting background. I didn't have my own admin. I went into healthcare and they're like, you have to have one. You are not allowed to work here without an admin.

Drex DeFord: Is this some of the challenge I think of like when you're starting something new, you think about the best way to assemble the blocks that are in the box that you get, that you have.

But when you're in an organization that's been around for a hundred years. And, you know, your traditions have been full employment and we're not gonna lay anybody off. And, you know, those are sort of like core beliefs. You may not be able to stick to them, but that's really something that you know, you're a community pillar and all of that.

A lot of our health systems who are, this is what we do, this is traditionally what we do, we needed to do more of it versus doing something completely different. It's a mindset and it's definitely a challenge for a lot of the traditional health systems. I think.

Bill Russell: Sarah what's something you heard in one of the meetings from one of the leaders with regard to AI that.

Sarah Richardson: This goes back to the ambient aspect because a lot of that space is how well it's being adopted and the KPIs around it, how people were gonna measure the success of it. But I feel like what worked the best this year were the projects that honored human design and honestly, the emotional outcome. We heard that quite a bit from our physicians that.

And I thought that was really powerful because you realize that empathy and efficiency are not opposite to one another. They're actually one and the same, and you gotta figure out a way that you can actually measure that. But a lot of the physicians will still tell you that there's an emotional aspect to the work that they do, that the only way you can measure it is them telling you, I feel better about this.

That's something to keep in mind. So CFOs aren't gonna love it, but here's the deal. You've got some higher end physicians or any physician that's still doing really well in your organization from a quality perspective, those are gonna be fewer and far between. People aren't going into the PCP space or family practice and the population's aging faster.

Well, not faster than ever, but there's more people going into, you know, Medicare as an example. So yeah. But the,

No, like, sorry. There's, There's like numbers associated with everything we do there. There has to be numbers associated with this. Show us some discipline around this. Like, what are we doing here? And that I

Drex DeFord: think, I think we've been down this road before, right? We, it's hard. Originally we've made a lot of emotional arguments about things around information technology or security or other things.

The emotional argument phase. Is ending soon on a lot of the AI work that folks are trying to do.

What are they doing today and what will they be doing in 2026? The other systems are sort of playing catch up on. And Sarah, I like what you said earlier. I mean, we saw any number of ways that people looked at Ambient and some systems were really proud of the fact that they did a nine month process and looked at all the ambient listening and then others, you know, it's like, yeah, we just picked one.

I mean, and well, how long was that process? I don't know. It was a week, like we just picked one and we went with it and we learned what we learned and we figured, you know, we looked at sort of the makeup and we're like, well, if we need to switch at a future time, we'll do that.

You don't have the most pilots, you have the ones that are really specific, and then you can trust it, you can audit it, you can scale it, and it has a defined timeframe by which to determine if that's even something that makes sense for the organization. But again, you gotta have your governance in place.

You gotta understand your workflows. That goes back to your original statement Bill, of how is it integrating into care? Who's the expert, who's there looking at it and fixing it? I think about that triangle that we use in terms of value for an organization. It's the same time a triangle that's used for change management through Prosci.

And we did a activity. Who's the business? Who's the product, who's the technologist? Any form of that triangle where the value sits in the middle is something all organizations really should think about because each of those three create a counterbalance that allows you to stay true to the purpose by which you decide to do something in the first place.

Bill Russell: and beyond.

I think the ones and direction you could, you can go after this the organizations that are really going to be making progress. First of all they've recognized the data and how important the data is and getting the data right and bringing all that stuff into the right kind of architecture.

I think I would go, I would just go after the money. I would look at it and I would say, look in order to get the CFO on board here, we need to find one or two use cases that just drives a real hard dollar. [00:28:00] ROI. We can showcase it, hopefully as close to that CFO as possible, you know, like maybe within the finance group, maybe within legal and accounting, maybe within obviously coding and rev cycle is a huge case.

Imaging where we're, we can't find enough doctors, but something where I could just show the hard dollar case so that when I, when it comes time to do those other projects that I know. My gut instinct tells me we should be doing this, that I have enough margin built with the CFO that he goes, yeah, hey, it's working over here.

It's working over here. It's delivering this. Yeah, take a little bit of a bet over here. So I think the ones that are focused on money i's, again, cynical and I know it's, it is what it is. Are probably going to get a leg up in the next 12 months I think.

Sarah Richardson: Would you have done ambient if you were still running a major healthcare system?

Bill Russell: Outwardly I'd be talking about those things inwardly. I'd be talking about I, I would've gone with the one with the best coding solution.

the reason is 'cause there's money in coding and there's also money in real time. Alerts back to the doctors while they're seeing the patients, you know, to improve our denials and everything else. Almost time. Yeah. So I would've been saying that from the pulpit. I would've been saying all those things about, you know, making the.

Bringing the joy back into medicine and whatever. But on the backside of the conversation you know, in the sitting with the CFO, I would be saying, look, and we're implementing this coding thing, the CMI would be working on integrating that into the entire workflow. And I'm not saying I want to overcharge or anything to that effect.

I'm saying I wanna get paid for what the work that we're already doing. Yeah. Don't look as good as we're actually capability.

My responsibility is to make any organization profitable in theory. And then how do you get there most effectively? AI may or may not be a tool in your toolbox for that right now, but you better know how it works because you don't wanna be the executive that gets left behind by not being prepared.

That's how we were with telemedicine when COVID hit.

Drex DeFord: The other saying that all of us talk about all the time is the monies and the narrows. And by that I mean the use case for AI doesn't have to solve world hunger. It really couldn't be a whole bunch of these tiny little use cases that can make a big difference.

And a lot of this is the stuff around financials. But it is a priority issue. It is, I mean, it is a prioritization issue, figuring out what are we gonna do? What are we actively not gonna do? How are we gonna. Prioritize these things in cybersecurity, I would say, you know, the bad guys have definitely piled on, right?

They can ask it to do lots of things. Taking apart zero days and figuring out how to weaponizing those zero days way faster than they ever had before. But the good guys are also using it. They're leaning into it pretty hard and they're, you know, using a lot of the tools from their partners. I think that in on the cybersecurity side is the place where they're, they need to put a lot of energy.

Their, the partners have a lot of cool tools, a lot of great things that are coming along. I was just reading a report last night. More than a thousand alerts a day for most health systems. More than half of those who were involved in this survey said that they sort of actively ignored because they didn't have capacity, some of those alerts.

Bill Russell: I think I want every IT person, every single one. If I had 800 people, which I did have 800 people, I would want them all to be telling me what they're doing with ai. Every time I talk to somebody, I'd want them to, you know, Hey, here's how I'm using it. I don't, and I think the other thing we're seeing is that the people that are sort of stuck on the tools that Microsoft gives them, recognize that it's not as good as some of the other tools that are out there.

We're getting rid of our CRM. Yeah, with code that was written from by ai. Alright, so we're getting rid of 12. I'm psyched

Drex DeFord: about this. I think this is the coolest thing ever. Yeah.

Bill Russell: A $1,200 a month bill on the CRM because we wrote our own and people are like, oh my gosh, now you're gonna have to maintain that software.

I'm like, and to be honest with you, I didn't write much of the code, so direction you put in a bug report and I sit there and go, Hey, Drex, put in a bug report. It says this, and it goes, oh yeah, I see that issue. Let me correct it. Yeah, I didn't do it. I mean, that's the level of automation that is available to us these days.

And you

Drex DeFord: see with Chad GPT, when you will ask it something and it will come back and say you know, this, whatever, it'll give you an answer and you'll say, well, a little bit more to the right. And then it'll give you another really good answer. I mean, it sounds like from the coating side, what you are seeing is that when I put in a bug report, it's just taking the bug report and saying, oh, well, it turns out that's correct.

Bill Russell: here's the new code. And that's

Drex DeFord: crazy.

Bill Russell: Made a lot of strides. I see our producer is giving us the acts here. I love hanging out with you guys. I'm so glad that you guys have joined this organization so we can have these conversations.

Not that we didn't before, but it's, it is just, it's fun that we can do it at a drop of a hat and just share what we're learning at the, at the various tables that we're doing the various summits, and we have a great set of summits lined up for next year. I feel like I'm learning more in this role than I did when I was CIO.

It's crazy. It's

Drex DeFord: unbelievable.

Bill Russell: It's

Drex DeFord: like we, you get a master's degree every night at dinner. I mean, you or every summit. So many people with so many stories. Bill, you've told me this before, people come up to you and say things like, well, you're really smart. And Sarah and I have talked about this.

It's not that we're really smart, we've just become really great conduits. Of information from the hundreds of executives that we talk to in any given year, and you're not supposed to that on 26, man, that concert

Sarah Richardson: We do well. We also talk about how important it's to be curious and to be vulnerable and to really listen to what's being said out there. So we absorb all that information. We synthesize it and we put it back out there to our community for the use of good. And so that's the whole purpose by which we decided to serve many versus the few.

And what a ride it's been this year, like hundreds of people, hundreds of hours that we have spent connecting with the HIT community. And you know, they're not alone and they know they have us. It's pretty powerful. Over a

Bill Russell: thousand different people. I've attended our 2 29 events this year, so it's been fun.

Yeah. All well, here's to more of this in the coming year. Thanks. Thanks. Oh, and that's all for now.

Thanks for listening. That's all for now.