Why Refusing to Change Is Riskier Than Embracing Agentic AI | Newsday

Why Refusing to Change Is Riskier Than Embracing Agentic AI | Newsday [00:00:00] Speaker: 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: Hey, it's news day, and I'm joined by my favorite two people in this world. Is that really true? Drex DeFord | 229Project: I Bill Russell: my wife's gonna be really upset. Sarah Richardson: It might be situationally true, but I think your f-- two Bill Russell: See, it's, Sarah Richardson: this hour. Bill Russell: it's his two favorite people on this call. It is Drex DeFord and Sarah Richardson, my, uh, intrepid, uh, team here at, uh, the 229 Project. And, uh, today we are, uh, we're gonna talk about AI. Does that surprise you two? Sarah Richardson: No. Drex DeFord | 229Project: Not at Bill Russell: Um, but I-- Drex DeFord | 229Project: existence right now, yeah. Bill Russell: It is a, a fairly sizable... It doesn't matter what space you're [00:01:00] in, uh, I thought there was a really phenomenal article written, and it was released in, uh, in Becker's Stanford Health Care's application strategy for agentic AI future, uh, written by Heidi Collins, Chief Applications Officer, and Michael Pfeffer, Chief Information and Digital Officer, Stanford Health Care, April 22nd. Um, and I thought it was worth, uh, going through. Here's the setup. So, uh, these are direct quotes. "We are now entering a world where agentic AI, AI that can act as a semi-autonomous coworker, will fundamentally reshape how work gets done in health care." Uh, their vision for the future, "AI agents take on more of the routine, repetitive, and rules-based work. Human analysts and operational leaders focus on higher-order problem-solving, design, and oversight. Uh, the boundary between IT and operations becomes more permeable, especially for lower-risk workflows." Um, let's see. Fear. Uh, "We also surfaced something important, a real and understandable bias [00:02:00] against AI within the team, driven largely by fear about job security. We didn't try to dismiss that fear. Instead, we spent some time unpacking it." By the way, that fear they're talking about was within the IT team. So that was part of-- part of it was, yes, there's fear across the organization, but there's also fear within IT, so that's interesting. Here's their reframe. Uh, "If we don't change, that's the bigger threat to our jobs than if we embrace AI. Demand for digital solutions in health care is effectively insatiable. The constraint is our capacity to deliver." Uh, they, they, uh, line these things out in two parallel tracks using AI to remediate technical debt. Like every large health system, we have accumulated tech debt over time, legacy configurations, workarounds, and workflows that no longer reflect ideal care delivery. Co-designing agentic workflows with operations, we're shifting from a tool request mindset to a problem first mindset. Wow, we've been preaching that for a long time. It's, it's [00:03:00] great to see the- Do them hi-highlight that. And only a couple more quotes here. New analyst role. The analyst of the future is less a ticket taker and more a workflow architect and AI supervisor. Wow. Sarah Richardson: Mm-hmm. Bill Russell: and then, uh, couple more things. How they're measuring... Drex DeFord | 229Project: a while too, right? Stop being Bill Russell: Yeah. Drex DeFord | 229Project: taker and start being a value maker. Yeah. Bill Russell: Uh, well, all right, we're gonna get into the discussion in a minute. I really, I'm, Drex DeFord | 229Project: Okay, Bill Russell: dying to get there. So how, how, how they're measuring success. Historically, much of the conversation has been about tools, adoption rates, number of builds, number of tickets closed. Those still matter, but they're not enough in an, a, in an agentic AI world. Agentic systems are not set and forget. We need ongoing monitoring of performance, safety, and user experience with the ability to adjust quickly as conditions change. Um, I wanna start with that fear, um, on, on actually naming the fear and addressing the fear out loud. That's a really, that's a really [00:04:00] interesting concept. I'd, I'd love for you two to, to weigh in. I've done a lot of talking so far, so I'd, I'd love for you two to weigh in. Sarah, I'd like to start with you on that. Sarah Richardson: Yeah, there's a couple things as you're reading through that that I love about them. They didn't make this an AI project. This is truly an operating model shift. They're not-- I mean, I think about is not delivering tools. They're now co-designing the workflows that are gonna allow them to continuously improve. What people expect them to do, the CI/CD aspects of, of development and other as- and other components. Not being very eloquent. But what I love is in addressing that fear head on, they aren't worried about jobs being replaced. They literally touch on the fact that if they don't evolve, a bigger risk. So now if they don't just build systems, they are workflow designers, and they're supervising the AI, now you can bring the systems thinking together with the capability of not being made irrele-- Like, if you make yourself irrelevant, but you're adapting yourself throughout that [00:05:00] process, then you've gotten way ahead of a curve that people are afraid of. Drex DeFord | 229Project: Mm-hmm. Bill Russell: I, I think I wanna make the person I am today irrelevant, right? Whatever the job is I'm doing today, I'm okay with making that irrelevant because if I do it, I've, I've moved somewhere else anyway Sarah Richardson: Yes. Drex DeFord | 229Project: I think there's this evolution too, just in the department, you know, the information services department in general. You see the connection more and more to the operators on the front end, the work that they're doing. This idea of We've talked about this in the past too, this idea of innovation being a thing where you're buying a product from a partner and implementi- implementing it, and then calling yourself innovative. I don't think those days are going to last too much longer. Innovation isn't just doing the thing, implementing the product. Innovation is changing the whole thing. You gotta change the whole workflow. You gotta pull in the operators from the front line. You gotta pull in the people who are doing billing and reporting on the backside of all of [00:06:00] this. You gotta change everything. That's really innovation. The product is just... I mean, we, we've talked about this before, the product's probably the easy part. It's all the other things that you have to change to really be innovative, and that doesn't stop. It's like a thing that the more that you, the more that you change, the more you improve, the more opportunities you see to improve, and you just, you just keep peddling that bike. Bill Russell: the two things that are converging very rapidly right now are, uh, developers and analysts and really product designers. , They're converging because the, the skill of just writing code, that, that skill's gone. I mean, we can replicate that. That's not to say that the skill of the programmer, 'cause the programmer does a lot of things. They do a lot of things underneath to, to bring data together and to, uh, create structures around it. Even, you know, we, we say that, um, you know, writing code's not that big a deal, but there's, there's all sorts of harnesses and frameworks around that to make sure it's writing good code, and it's testing that code, and [00:07:00] it's following good practices. So the job itself is morphing though, in that it's r- it's rushing towards the, um, you know, towards the elbow, if you will, and the analyst is rushing towards the elbow. And both of them are trying to do the same thing now, which is to create relevance by saying, "Look, I can take your problem and I can, uh, I, I, I can interpret that, and I can create something back for you that actually solves that problem." Um, on this side we have, uh, healthcare IT that has run away from programming for years, that has to develop those skills now. And we have programmers that have stayed in the back room for years 'cause they didn't have to come out. Now they're having to come out of the back room and spend more time at the elbow. Drex DeFord | 229Project: He- here's the other really interesting thing. The people with the actual elbows who are doing the work are also becoming analysts and programmers. They can Bill Russell: Yeah. Drex DeFord | 229Project: I don't need you guys anymore. Like, there's a sum of that going on in our environment right now[00:08:00] Sarah Richardson: As long as you don't isolate yourself by thinking you can do everything by just sitting in front of a terminal. It's almost like a reverse effect of having your basement set of programmers. You hid them out in a dark room with lots of pizza, and they just went and created. If you start to do that with the people who are typically out there and are the voices and the aggregators, then you gotta make sure there's still a balance of the human connection. I do love that they are approaching this as how can AI get rid of our tech debt, automate our testing, and clean up legacy? So a lot of the things you hear are that people can't do AI because they have too much tech debt. Well, let the AI help you with that, and then you lean forward into designing the future with what Drex's favorite thing, fix the workflow first. If you do that, you don't have a bunch of ICs, you have a bunch of ILs. I've been leaning into this idea over the past week, individual leaders that then have all these different components. So your ind- your ICs become your ILs, and when you can literally have a distributed [00:09:00] set of leaders in a systems thinking environment bringing us back together for that human interaction, that is gold in what these teams need today Drex DeFord | 229Project: I was talking about that with somebody in Utah, uh, last week, and idea that, legacy and tech debt and all... There's so many reasons that you can be a victim in all of this, right? I'm afraid of it. I'm afraid of it for, uh, I can just-- we can just collect these things into a bag. There's so many reasons. innovators, though, look at those restrictions and those problems and those challenges, like you said, Sarah, and they challenge accepted. Like, okay, let's figure out how to innovate our way around it. Let's figure out how to, like, make that thing that's a problem just... It's so isolated now, nobody worries about it anymore. is the sort of taking charge of your environment and moving forward. And I think those are the people who are gonna be in business healthcare a year from [00:10:00] now. Bill Russell: it's, it's interesting in my, um... So Claude just did a couple of, of upgrades to their system as, as did ChatGPT. And it's interesting, Claude used to, like, the minute you sort of got an idea, it would say, "Do you wanna build that? Do you wanna build that?" And now it's, it's very interesting 'cause it almost talks you out of building it. It's like, "Hey, look, you guys are spending too much tokens building crap. Like, stop doing that." Drex DeFord | 229Project: Hmm. Bill Russell: Thing that we do to a certain extent saying, "Hey, you know what? You could do this with a little bit of manual here and a little bit of AI here and a little bit of deterministic here, and you could see if this is gonna work. And if it works, heck, then come back and we'll scale this thing up and make it, you know, all-powerful." I don't know why I talked about that. I talked about that just because I, I think there's an opportunity here. Uh, uh, [00:11:00] Sarah, you talked about tech debt, and when I think ab- when I think about the opportunity within IT, it is to start to, to exercise a muscle that we haven't used in a while, which is to say, uh, along the lines of what Drex was saying, step back and say, you know, "Is there a different way to do this than we've done it in the past?" Not just, "Hey, can we automate the review of all these logs?" Maybe there's a, a completely different way of, of looking at, uh, how we, how we design something or go about it. But the, the whole area of tech debt is an area that's a pretty... Uh, I, I don't know if I wanna say low risk, but it's, it's not a high risk, uh, kind of thing that people can sort of get their, their hands dirty and say, "Look, let's, let's dive into this. Let's see if we can solve it." And Drex, we, we talked about this. I think we talked about this on the show. We talked about the solution that was brought forward in the CISO event where they, they took call center calls and, and, uh, put AI [00:12:00] around it, and essentially they're handling their, their call center calls. Drex DeFord | 229Project: That's... that's the problem is that we, we have... We think about tech debt often as we believe that the only way to solve that problem is by taking the old stuff that we have today and replacing it with new stuff that does the same thing. That's tech debt replacement. I have to refresh the network. I have to refresh endpoints. I have to upgrade that particular Bill Russell: see, Drex DeFord | 229Project: I have to whatever, right? And so m- the anchors of my belief are, are, know, like socked into the ground around these particular ideas. You, y- that essentially today's the day you gotta cut those lines. You gotta cut those Bill Russell: do you see Drex DeFord | 229Project: and let yourself go from those anchors Bill Russell: the... Drex DeFord | 229Project: a whole new way of Bill Russell: Yeah. Drex DeFord | 229Project: at the world. Bill Russell: You see that the guy who used to run SpaceX with Elon left... Or, or Tesla. He used to run Tesla with Elon, left and wrote a book. [00:13:00] It's called, like, The Algorithm or something like that. It's five things, and the first one is- Like, get rid of stuff. Like, just stop doing stuff. And he tells a great story where they were, they were having trouble with the heat shield underneath in between the battery and the car. And, uh, he goes, "Man, we, we, we were trouble- we're having trouble to, you know, create the right molding and get it right." And he goes, "We were spending hours and hours." He goes, "We pulled Elon into a couple of these meetings." He goes, "That's really expensive time because he's, you know, trying to solve really big problems, and we couldn't solve it." And he said, you know, "G- go to the source." And so the, the engineer that recommended it, they went to that group and said, "Why'd you recommend that?" And they said, "Oh, we didn't recommend that. That w- that came from this group over here. They, they said we had to do it, and it's a regulatory thing." So they went over to that group and the group said, "Oh, that's not a regulatory thing. That group over there said we had to do it." And they pointed at each other and finally they said, "Do we have to do this?" The answer was no. There, there's no reason to do it. And they're like, "Oh my gosh, we just [00:14:00] spent like six months trying to do something we didn't have to do." And that's a great example of, I think there's... I believe there's a lot of those in health IT Drex DeFord | 229Project: To- Bill Russell: definitely within health care Drex DeFord | 229Project: Sarah, definitely we talk a lot about workflow. We also talk about this five whys question all the time. Why are we doing that? Why did they tell us that we need to do that? Why is there a regulation? Is there really a regulation for that? What, show me where that is. Like that, those are the things, like as you pick at this, you start to understand, man, I am There's a story about ele- about elephants, like when elephants are young and they're sort of like chained to a, you know, a post and they won't walk away. And then as they get bigger, you can almost just turn that into like a kite string and they won't walk away, because in their heads they can't leave that spot. And I feel like there's some of us that are just, we've got kite strings on us. We sh- we ought to be able to just snap these things and [00:15:00] move, but we won't 'cause we've just put ourselves in these prisons. We just can't figure out how to get out of them. Sarah Richardson: Unless you create the space where you still have a human-centered aspect. I mean, it's not AI first. It's like we're gonna use AI for this. Here's what the humans have to own. Here's what happens when more people get involved. creates that less rigidity, though, that... To Drex's point, you cut kite strings when you have the cross-functional product style teams, review. I mean, monitoring just becomes more of a, a safety net aspect of what's occurring, and all of this is so fascinating. It takes all of the stuff we've all historically known and done in our careers. It's layering on. This is everything that MIT is teaching in the class that I'm taking right now, 100% about how you distribute the learning, distribute the, the review of how to implement these things, and you're using AI not just to build your way out of some of your constraints. You're also using it to monitor what the [00:16:00] organization is learning and how it's getting smarter together, some of the tools of watching behaviors of the psychological safety that has to be true for teams to want to pursue this type of work Drex DeFord | 229Project: Do you Bill Russell: All right Drex DeFord | 229Project: are, are le- because we're moving more and more toward the frontline workers and including more people in some of these design conversations, it, there's more people in the room who are willing to ask why 'cause they just don't have the constraint of whatever that problem is that maybe makes it easier for us to move on from those anchors? Bill Russell: , Historically, much of the conversation has been about tools, adoption rates, number of builds, number of tickets closed, et cetera. Um, in the, in, in the AI agentic era, we're gonna be looking at different metrics. Um, let me throw, throw some out. I'd, I'd love for you guys to, uh, comment on. The first one is, uh, time redirected to higher value work per FTE. It's... You know, we see this whole idea that, uh, agents are going to pick up the, the routine and the mundane and [00:17:00] the, uh, the repetitive. Um, I think that's an interesting metric to say, "You know what, Drex? It's not that you're not working 40 hours anymore, it's that you're working on higher order problems." That your time has been re-redirected to, to functions that are of more value. Not only you, but obviously as we put more agents in, uh, more and more people. H-How do you feel about that as a metric, uh, Sarah, Drex, or is there a different metric you would think of? Sarah Richardson: I have a big opinion unless Drex wants to go Drex DeFord | 229Project: No, go ahead. I wanna hear it. Bill Russell: Oh, big opinion. Here we go. Sarah Richardson: Because this is what I'm actually coaching people on right now, and it's if all of your work is now high value, sometimes you need the brain reset of the mundane dumb stuff that lets you just chill out and be mindless. So let's just pretend that you had 30 hours a week of hardcore working and 10 hours of just getting stuff done. Well now, if all 40, we're making that number up for sure, if all of your time is high impact, then you're not allowing your brain and your body to actually reset to [00:18:00] be able to think at that level all of the time. It's more exhausting to be always on versus, "I'm gonna take two hours and just screw around and answer my email today." Sometimes for a lot of us, that was a mental break. Bill Russell: Don't disagree with that, at, at all. I would say that, um, you know, the, the, um, the how do I say this? I'm gonna offend somebody. I'm always offending somebody. Drex DeFord | 229Project: Go ahead, Bill. Bill Russell: Um- No, I mean, it's, look, the, the... Take, take the mental break. Actually take lunch, for heaven's sake, or, uh, you know, take, take a real mental break. Watch YouTube videos. I, I, I really don't care what you do with your downtime. What I care, Sarah, is you're solving the highest order problems for us. Because look, there's enough constraints in a healthcare organization that's gonna slow you down. It's not you that's gonna slow you down. It's organizational change management, it's, uh, regulatory. There's so many things that are gonna slow you down. You're gonna have plenty of, of downtime because if you're working on those higher order problems, you have so much, uh, you have so much minutiae to, to, to wade through to [00:19:00] get to the other side. So I, um, I, I, I'm, uh, mostly I'm arguing because I'm hoping for a world where we never have email again. That's what I'm hoping for. Sarah Richardson: Hey, I'm mostly text, and if I can text and mar- and emojis, even better. Bill Russell: Um, I, Drex DeFord | 229Project: I get Bill Russell: I... Drex DeFord | 229Project: there's a lot of people that are burned out right now, and you think about like, you know, putting all this... There's m- I'm, I'm 10 times more productive, but I have to be really, really engaged all the time. Maybe that makes burnout worse. I think it's, I think maybe it's a, it is a balance thing though. You're 10 times more productive, but maybe you have a little more free time to do the mindless things that Sarah Richardson: But Drex DeFord | 229Project: save you from Sarah Richardson: us. You're wired to just Drex DeFord | 229Project: burnout. Sarah Richardson: and keep being that Drex DeFord | 229Project: fill the glass. Yeah. Sarah Richardson: so it's almost like a deprogramming of a, at least our generation, Bill Russell: You Drex DeFord | 229Project: all Bill Russell: of the, one of the metrics that Stanford or the anchors that Stanford has in here, reduce time to complete a workflow, decreasing error rates, improving throughput. My gosh, Drex, that has to make your heart... Drex DeFord | 229Project: Yeah. Bill Russell: it's lean. Makes your heart warm. Yes. Well, I, I want to close this by hearing what you guys talked about. So you guys both spoke last week. Drex, you were in, uh, Utah, Drex DeFord | 229Project: Utah. Bill Russell: Sarah, you were in Sarah Richardson: [00:20:00] Nashville. Bill Russell: Nashville, Tennessee. Awesome. So, uh, Sarah, what'd you talk about last week? Sarah Richardson: I loved what I talked about last week because every panel was super techy, and then I had, uh, Tiffany, Cross and Maria Sexton and Cinnamon Matthews, and we talked about the aspects of leadership from EQ to the coaching elements as a leader to the transitions in your careers. The higher you move up the food chain, the less you know, the more important it is to be able to distribute the systems thinking and capabilities of your team. Oh, and by the way, here's all the mistakes we made along the way, and here's how we keep adapting and learning from what those feedback- loops are telling us, and really in today's world, how you can not know and yet still feel safe in leading your organization. Bill Russell: I, I love the higher up you go, the less you know, and the more people there are that want to tell you how little you know. It's like it just, they keep telling you. I'm sorry. I don't know why that [00:21:00] strikes me as funny. I, I never was told, um, how little I knew until I became a CIO of a, Sarah Richardson: Yeah. Bill Russell: of a health system. It's like, "You don't understand." It's like, uh, I've studied this for the last 30 years of my life. I have a pretty good idea of what's going on. Anyway, Drex DeFord | 229Project: think that's why you have to be a good learner and a good listener in that job too. Sarah Richardson: be curious. Bill Russell: that's Drex DeFord | 229Project: Yeah. Bill Russell: where the EQ comes in. Drex DeFord | 229Project: Yeah. Bill Russell: Dre-Drex, what did you talk about? Drex DeFord | 229Project: I talked about, I really talked about trust and the challenges that we have around trust right now. The name of the presentation was Sammy the Engineer, His Robot Army, and the Great Trust Recession. And, um, I really just talked about, uh... I gave some, some examples, some stories of things that we just naturally trust in today that have become untrustable. A lot of it comes from UnHack, the podcast, or the UnFake series that I did last spring around, um, know, m- changes in media and the be able- the ability to create deepfakes, and now it's turned into the North Korea, you know, fake employee catfish [00:22:00] problem. And so there's so many things that you can't trust, and a lot of it was just sort of the advice and guidance of the, you know, stay a little paranoid. It's really, it- it, we really live in an odd world right now where we have to be very thoughtful about everything that we see and we read, and, and the video that your mom sent to you this morning, you gotta look at it a couple of times and make sure that it's what you think it is. Bill Russell: So Sarah, Uncle Drex is, um, Sarah Richardson: Heck, Bill Russell: telling those scary stories again Sarah Richardson: I even asked you over the weekend, "Hey, we got an email from you. Is it you?" Because it was not expected. And so Drex DeFord | 229Project: There you go. Sarah Richardson: sending you questions, "Is this really us? Is this really us? Is it really you?" Especially when it goes to my personal email. Drex DeFord | 229Project: Right. Sarah Richardson: uh, I, I mean, trust but verify is like should just be the new credo of all the things that we look at. Even people asking us, do we trust what our partners are telling us? So if you're in a relationship with a vendor that's providing you, quote-unquote, "AI services" or any service, how do you go validate that they know what they're talking about? I mean, people are [00:23:00] asking us for that kind of feedback today. Drex DeFord | 229Project: think Bill Russell: Yeah, it's, it's a, um, it's an interesting world. I, I know that, uh, Drex DeFord | 229Project: the Bill Russell: bookkeeper and myself has some very strict rules as to how communication happens and, um, uh, I would say only really once or twice in the last year did she send me something where I was like, "No, Drex DeFord | 229Project: important Bill Russell: that's..." Whatever. But most of the time she goes, "You didn't send this." I'm like, "I did not send this." And Drex DeFord | 229Project: in Bill Russell: the, nice thing about our bookkeeper is th- they're not even on our domain, so people would have to figure out that person's domain in order to even find them. So it's kind of, uh, kind of fun. And, um, kind of fun, but I'm not, like, throwing the gauntlet out there. I don't want anyone to figure it out, like, uh, come attack us or anything like that, right Drex? Drex DeFord | 229Project: this too is that there's, you have grace for that, right? Like, we're all in a hurry and we all wanna get things done, and we want our teams to move quickly, but when they send us the note, before I click on this, did you really send this to me? Is this really something I'm supposed to do?" And that you have the grace with your teams- [00:24:00] it's okay to take that beat and ask me that question. That's way better than the mess we would've had to clean up if you would've clicked on it. Um, even if the answer is, "Yeah, that actually is from us, it's okay to click on." give them the break. " Bill Russell: Did, you, did you really write this really stupid email?" Uh, yes, I Drex DeFord | 229Project: Right. Bill Russell: did. Drex DeFord | 229Project: me, yeah. Bill Russell: That was me. Sarah Richardson: Because I Bill Russell: there we go. Sarah Richardson: I didn't use AI. That's why it sounds so bad now. Bill Russell: Back to, uh, back to EQ, I guess. Uh, well, we will see each other... When will we see? Oh, well, we'll have another one of these, and then, uh, next week we'll be together in Destin, Florida for another, uh, summit, uh, CIO for me. Sarah, what are you imaging for Drex? Sarah, what are you doing? Sarah Richardson: I have the CIOs, you have the CTOs. Bill Russell: C. That's C. Thank you. I'd be, I'd be in the wrong room. Sarah Richardson: Didn't even need an agent to tell you that. Bill Russell: That is one of my favorite, as you guys know, that's one of my favorite groups to get together. They will completely nerd out. It'll be fun. Drex DeFord | 229Project: Yeah. Yeah. Bill Russell: Thanks everybody for listening. That's all for [00:25:00] now. Speaker: 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.