Elysia Culver (00:00): All of this focus on where the disruption is happening is overlooking where the actual maturity of AI in the rev cycle actually is. Rae Woods (00:09): From Advisory Board, we are bringing you a Radio Advisory, your weekly download on how to untangle healthcare's most pressing challenges. My name is Rachel Woods. You can call me Rae. There's a new buzzword or it's really more of a phrase that's hitting healthcare. I am talking about the increasingly prevailing narrative that there is an AI arms race that's actively unfolding. Based on that rhetoric alone, you'd be forgiven if you assume that health plans everywhere have AI agents that are managing their prior authorizations and reviewing claims and that those AI agents are deep in a battle of the bots with health systems who are also using AI in their case to maximize their approval rates and overturn every denial. (00:55): But let me do a quick reality check. This battle of the bots assumes a level of AI maturity that the majority of healthcare businesses just don't have yet, and that assumption can result in healthcare leaders making hasty investments in an effort to catch up to their competitors, resulting in unreliable ROI and a distinct lack of revenue cycle optimization. (01:19): So to help us understand where AI can actually drive the most value in revenue cycle, I've invited two Advisory Board experts, ISIS Monteiro and Elysia Culver. Isis, Elysia, welcome back to Radio Advisory. Elysia Culver (01:34): Thank you. Isis Monteiro (01:35): Thanks, Rae. Rae Woods (01:38): We are talking about a pretty big topic that I'm feeling is in the zeitgeist, at least among healthcare leaders. And it's this narrative about the AI arms race that's happening in revenue cycle. I see this at conferences, in conversations with healthcare leaders. I see it in headlines. How often is this AI arms raised coming up on your calls with CFOs and with rev cycle leaders? Elysia Culver (02:06): This is probably coming up in every single conversation that we're having right now with leaders. We've had I'd say 20 or so calls with rev cycle leaders and CFOs and a lot of the time the conversation actually comes up unprompted. Rae Woods (02:20): Oh, wow. Elysia Culver (02:21): And I think it's because these leaders believe that payer pressure is the number one threat to their margins right now. Rae Woods (02:28): It's good to know that it's not just me, that this is coming up everywhere. And you already named why that's happening. If you're a provider, you're probably feeling like, wow, plans are rapidly investing in AI. Providers then feel their own pressure to invest in their own AI capabilities to match what their competitor is doing. That then makes plans feel like they need to invest more and more and on and on and on into perpetuity. (02:55): My question is, is that a bad thing? Is this so called race something that's driving healthcare organizations towards something better, more innovation, better clinical and financial outcomes? Elysia Culver (03:07): I think you could see it being that way, but I think the long story short is that it's a no. And I think it's because this competition is built on the assumptions of what others are doing. And if you're making investment decisions on what you think others are doing, not what they're actually doing, that's not going to lead towards better clinical and financial outcomes. Providers think other providers are ahead in their use of AI. Payers think providers are ahead, but what we have found in our calls is that nobody is actually ahead. Rae Woods (03:39): Oh, really? Elysia Culver (03:40): There is a ton more nuance to this conversation than people think. And we are seeing actual AI investment in the revenue cycle, but it's not at the scale that the media is portraying. Rae Woods (03:53): So less arms race, more fear of missing out? Elysia Culver (03:57): Yeah, and I think there's a couple other things too. I think the bigger one is that there is this conflation of automation and AI, which makes activity feel more transformative than it actually is. Rae Woods (04:09): I actually find that that's happening in 2026 more than it was happening in 2023. We were initially spending a lot of time on AI kind of defining what it meant. And I find myself three years post-generative AI coming back to let's do some definitions here. Elysia Culver (04:25): Yeah. And then I think secondarily that there's a lot of buzz around the new generative capabilities of AI. I think there's a lot of focus right now on this influence of ambient listening on revenue cycle and this potential disruption is actually being mistaken for maturity. So really, I think nobody is winning here. Isis Monteiro (04:45): I think it's something like $200 billion per year, and that's a conservative estimate, that's wasted in the back and forth between payers and providers. So just administrative costs related to financial transactions, that has nothing to do with actual care delivery. In theory, if payers and providers adopt these tools that reduce these unnecessary costs and spending, those savings could theoretically get passed on to patients maybe eventually, but I think it's more likely, at least in the short term, that those savings are probably going to get reinvested into more technology to amp up this arms race or be used to improve the bottom line. (05:23): The point being here, at minimum, it's safe to say that patients are caught in the middle and they're not immediately benefiting from this. I think the more clear winner, if we want to assign that label to someone, are probably the vendors that are selling to both sides and capitalizing on the urgency that both parties are feeling. Rae Woods (05:43): Which is one of the reasons why in our first episode of 2026, when we talk about what are the core things that executives and CEOs in particular need to know about healthcare, we actually describe the power of vendors as being something that every senior executive needs to watch. Elysia Culver (06:00): I actually don't know if I even see them being a winner or loser even in this situation because it's been the status quo for years that they've been selling to both sides. But I think really the question is how are they going to be the one setting the pace and change of adoption, especially if a lot of leaders are wanting to kind of wait for their core platforms to introduce new solutions to them. That I think is the bigger longer term thing to be looking at is maybe not who's the winner or loser, but how are they maybe going to be setting the pace and change of adoption? Rae Woods (06:31): So I'm already coming to a takeaway in this conversation and that is that the entire industry's framing around the arms race might not be entirely accurate and it also might not be helpful. It might not be accurate because again, it's more FOMO than actuality. And it might not be helpful because then when we talk about an arms race, we have to start talking about winners and losers, and we should actually be talking about how to make the system better. (06:58): But I want to empathize with where leaders are actually coming from and what's fueling the sense of that arms race is ultimately margin pressure. From the provider side, they're not just feeling margin pressure, they're feeling payer pressure, right? Their reimbursement is getting tighter, they're seeing more denials. So they are looking to their rev cycle, frankly appropriately, as a line of defense against margin erosion. But I'm already starting to hear you say, "Hey, AI is not going to be able to deliver on the transformational gains that we hope for, at least not in the near term." So how do you want leaders to be thinking about rev cycle optimization? Isis Monteiro (07:34): I think that, and this is consistent with what we've been saying since before this AI boom, which is don't chase products, chase problems. Take the time to identify the source of your revenue leaks and plug solutions into those rather than going after the flashiest technology because you think that's what your peers are already doing. (07:54): Also, this conversation is around revenue cycle management, but to your point, Rae, revenue cycle optimization is not the only thing that providers should be paying attention to right now. It's on component of what we think should be a broader, more comprehensive approach to margin managements that also elevates workforce productivity and stability, clinical and operational efficiency, and strategic growth and access. Rae Woods (08:19): It seems like the best thing that we can do is to take your research and actually challenge some assumptions that are underlying health systems AI and rev cycle strategies. I am also cognizant, we talked about AI being a buzzword, that rev cycle is kind of a buzzword. What do we mean when we talk about rev cycle? What are the functions? What are we actually talking about? Isis Monteiro (08:43): So we're talking about the entirety of the rev cycle, which includes front end, mid-cycle, and back end. Front end is what we think about and often sits within patient access. That's scheduling and pre-registration and then registration and check-in. Mid-cycle is charge capture, utilization review, and coding and clinical documentation integrity. And then back end is claim submission, payment and remittance processing, denials management, and patient collection. Those are the interactions with the payers. Rae Woods (09:13): And where can AI meaningfully fit into those phases? And where are we maybe seeing more AI certainty versus more of that AI hype that we were talking about? Elysia Culver (09:24): Yeah. So I want to actually reiterate that there's a lot of nuance to this conversation and answering that question is actually a little bit more complicated than it appears on the surface level. Rae Woods (09:34): As are all things in healthcare. Elysia Culver (09:37): So reiterating the point again that there's a lot of conflation right now between AI and automation and people are mistaking disruption for maturity. Rae Woods (09:44): Can we actually pause on that? Why is it important to name the difference between AI and automation? We're not just being nitpicky about language here, right? Elysia Culver (09:51): Yeah, no, this is an important question. And I think the biggest difference is because we are seeing new capabilities with AI that we haven't seen with automation alone in the past. When you think about these generative AI models, these large language models, there's theoretical possibilities with this technology that are creating opportunities in the revenue cycle that we haven't seen in the past. That's the main reason why that differentiation matters. Rae Woods (10:21): And you said we're also conflating disruption and maturity. Elysia Culver (10:25): We are. So because of those two reasons, I think it's actually more beneficial to look at not AI necessarily across the different parts of the rev cycle, but where is AI being most used today? Where is it emerging or unevenly adopted? And where is it limited and why? Now we do have a field guide coming out that will map things to the actual rev cycle, but for this conversation, I think it's helpful to look at it that way. Rae Woods (10:54): So let's talk practically then. Where is AI not being used today? Where is it just still limited and we're really talking about FOMO? Elysia Culver (11:05): I think this is really on things that are typically patient-facing or complex. So if you think about the patient-facing financial conversations, utilization, management, review, final payment decisions, things like that, those are where we're not seeing AI being used. (11:24): Now, in terms of the mature AI use cases, the focus in this category is how are these tools being used to do things like summarizing, prioritizing, helping staff work top of license? And I think we've historically seen the more mature use cases in things like prior authorization, AI-assisted denial management, things like that. Now that's not to say AI is being used perfectly here, but it is where we've seen those mature use cases. Rae Woods (11:54): And I'll say that is where a lot of the conversations, at least I'm having, again, at the conferences out in the market, they are perhaps talking about some of those mature use cases. Elysia Culver (12:01): Yes. Yeah. And now for the emerging use cases category, this is the category I would say you see a lot of that theoretical upside, but it's not where things are working at scale. It's also the area where, again, we're seeing the most hype. So I think solutions that fall into this category are things like claims submission risk prediction and coding and charge capture. And it's here where we're also seeing that hype around things like documentation and coding and CDI, again, to the point because we have these new generative capabilities made available by large language models. (12:39): Also with these kind of new capabilities that AI is bringing, I think we're hearing that it's becoming more economically viable to pursue new pools of denials that weren't previously valuable to pursue prior. So that's kind of something of interest in this area as well. Rae Woods (12:57): Mature area, but let's go further and go deeper. Elysia Culver (12:59): Yeah. And despite the theoretical upside that we see in this category, this is where another big assumption comes into play and that's that an AI investment in these areas is the most effective way to defend revenue. Isis Monteiro (13:16): Here are a couple of examples that we've heard across some of the research interviews that we've been doing about how organizations are currently deploying AI and other capabilities in order to smooth out the rev cycle. So in the front end, we've heard of organizations combining technology investments with front desk staff education and also creating bonus incentive structures to increase point of service collections. (13:40): In mid-cycle, we're talking about folks using AI to minimize response times for additional documentation requests from payers, seeing results as much as minimizing backlogs from two and a half weeks to same-day turnarounds to improve cash on hand. (13:55): And in the backend, to Elysia's point, one of the more advanced use cases that we've heard is using AI to pull from medical literature, clinical documentation, and payer policy to draft these really persuasive denials appeals letters that have a higher likelihood of getting overturned. Rae Woods (14:14): So what I'm hearing from the examples that Isis just gave is that AI does have the potential to reduce work, prevent avoidable revenue loss. That's promising. That's exactly what health systems are looking for. They're looking to find opportunities to have an outsized impact on protecting their margin. But at the same time, we're saying that there's a lot of variation in how mature these examples actually are. What's your take? Elysia Culver (14:42): Yeah. The assumption is that AI will improve documentation and coding and that will lead to revenue growth. And what we're saying is that we shouldn't look at it as revenue growth, one, we should look at it as cost avoidance. And two, that take I think is overly optimistic because few organizations have actually achieved substantial revenue lift. Rae Woods (15:06): So there are examples of supporting the rev cycle, but it's not actually translating into meaningful revenue growth. Elysia Culver (15:13): Yeah, and I think that's because ROI is really hard to prove. Part of it, I think the reason that's the case is we're really early in all of this. Rae Woods (15:22): Yeah, I was going to say, it's not doing this yet, maybe should have been at the end of my sentence there. Elysia Culver (15:27): Yeah. Yeah, it's not doing this yet. Part of it is that the conversation is really early. And then another big part of it is that the benefits that people are wanting to track, so avoidable denials, reduced work, et cetera, are meaningful, but they're difficult to isolate and quantify. (15:44): And then also, the even bigger thing is that ambient listening is actually still being positioned as upstream clinical infrastructure rather than kind of a revenue cycle owned capability. So the goalposts for what you're tracking for ROI have shifted where before, you're looking at are you saving time in the clinical visit by having this thing write notes for you? Now we're like, "Oh, is this going to help revenue cycle and better code and lead to benefits there?" (16:14): So again, the conversation is shifting and I think long story short, or I guess long story long, all of this focus on where the disruption is happening is overlooking where the actual maturity of AI and the rev cycle actually is. Rae Woods (17:32): I'm doing something that's very unfamiliar and uncomfortable for me as a person and as a podcast host, which is teetering between optimism and pessimism. And on the one hand, I am hearing that for real revenue growth, AI in the rev cycle, probably not there yet. But I'm also hearing some success stories, right? Isis, you named them. So the more specific question I want to ask is what is the real near-term opportunity for AI in the rev cycle? Elysia Culver (18:00): I think the real value of AI in rev cycle is lowering costs by improving administrative burden and using it to improve operational efficiency. The examples we called out earlier related back to prior auth support, real-time eligibility check, denials management, et cetera. And there's so much, I think, low hanging fruit that providers shouldn't be overlooking I would say kind of in that front end of the revenue cycle. Isis Monteiro (18:30): I do just want to emphasize, especially with the front and the back end, which do have more to do with those sensitive financial conversations that Elysia had pointed to earlier. This looks like educating your front desk staff about how to ask patients for money to improve point of service collections or educating patients about how to retain their Medicaid coverage eligibility and connecting them to community or financial counseling resources to mitigate coverage churn or answering difficult questions about a bill that a patient gets. There's still this human element that can't be entirely replaced by. Rae Woods (19:08): That's what I was just going to say, Isis, because I totally hear you on we love to start with a low-hanging fruit anywhere, we want to start with what are the easiest opportunities. And again, the opportunities are less on revenue growth and more on preventing margin erosion. But the example you just gave was not a tech example, certainly not an artificial intelligence example, right? Educating front desk staff? Isis Monteiro (19:28): Yes, I think there is this education and this trust building component that absolutely can't be ignored when you're implementing this technology. We're also not seeing organizations outright replace their revenue cycle staff with AI because there's still this understanding that a lot of these tools still require human oversight and a human to act as the final decision maker. Rae Woods (19:50): Especially because there is clearly such variability in the maturity if we're looking at AI tools at different stages of the lifecycle. Isis Monteiro (19:58): Yes. What they are doing are leveraging these tools, again, to augment workforce capability and productivity, especially in light of projected staffing and expertise shortages, which are particularly salient in the rev cycle because of mass retirements and high turnover rates. So it alleviates the burden, but it does not replace the workers in the rev cycle. Rae Woods (20:20): But is it something that can help then stabilize the workforce? We think about financial goals. One of them is reducing costs. A way to reduce cost is by reducing headcount. I hear you on, we're not at robots replacing humans yet in rev cycle, but is there an argument to be made that strong, mature AI tools can help stabilize the workforce? Isis Monteiro (20:40): Yes. And lowers cost to collect, makes it cheaper for providers to get reimbursed for the services that they've already delivered. Elysia Culver (20:50): Yeah, I feel like this is something that has come up in honestly all of our AI conversation is this idea that AI is going to take over jobs. But even in revenue cycle in particular, it's the same story that you hear everywhere else where they're operating with less staff. And the whole point is not to replace staff because they don't have staff. The whole point is how do you help them work better and do their job more efficiently with the staff that you have? Rae Woods (21:17): It kind of connects us back to when we were talking about FOMO, right? There is a lot of hype around where we can use tech and AI to fundamentally replace human roles. There's whole conversations about where that is appropriate, where it isn't. And what I'm hearing from you here is this is perhaps another example of hype. Isis Monteiro (21:36): Yeah. Rae Woods (21:38): I appreciate that we are having a real talk conversation around the state and the maturity of AI in the rev cycle today and the opportunities that could come next. We said at the start of this conversation that we wanted to understand the assumptions that are underlying all the hype that's out there. What are those assumptions? Elysia Culver (21:58): I personally think it boils down into two large assumptions that providers are making in this space and the first one being that payers are the biggest threat to their margins. And this is probably a little bit true, especially as it relates to rev cycle. But again, we talked about how revenue cycle is just one piece of the larger cost management structure that leaders should be looking at. Isis Monteiro (22:25): Yeah. I think the second big assumption is AI and automation are not the same thing and that distinction does matter because it drives a lot of mistimed investments. Rae Woods (22:35): So then what should leaders, our listeners do differently to drive real revenue cycle optimization that is meaningful and realistic today? Isis Monteiro (22:47): Yeah. I would actually take it up a level from just thinking about revenue cycle optimization to thinking about margin management as a part of a broader comprehensive framework that, again, takes into account clinical operational efficiency, other cost-cutting measures, and workforce productivity and stabilization. Then the second thing that I'll reiterate is just chasing problems and not products. Take stock of all of the issues that are leading to revenue leaks and map solutions onto those problems. Elysia Culver (23:17): Isis just named two things leaders should stop doing, but I'm going to actually add one more to that. And I think it's to stop chasing AI because you feel behind or because vendors frame it as transformational, just because that's not a good reason to invest. Rae Woods (23:32): Essential word being you feel behind, which as we've named over the course of this conversation, might be more feeling than actuality. Elysia Culver (23:40): Yes. Rae Woods (23:41): So then what do you want our listeners to start doing? Isis Monteiro (23:45): Refocus your attention not on revenue lift, but on margin protection, scaling the products that work, and financial resilience. Rae Woods (23:55): Well, Isis, Elysia, appreciate the real talk as always. Thanks for coming on Radio Advisory. Elysia Culver (24:01): Thanks so much for having us. Isis Monteiro (24:03): Thanks, Rae. Rae Woods (24:08): Here's the bottom line. I don't want you to get caught up in the hype around the AI arms race, especially since as we've learned, progress here is, at best, uneven, maybe even a little bit slow. So what I want you to do is to make smarter decisions. That might mean optimizing your revenue cycle, it might mean focusing on margin protection more broadly. But the stakes are too high to make misplaced investments. And remember, as always, we're here to help. (24:56): New episodes drop every Tuesday. If you like Radio Advisory, please share it with your networks. Subscribe wherever you get your podcasts and leave a rating and a review. Radio Advisory is a production of Advisory Board. This episode was produced by me, Rae Woods, as well as Abby Burns, Chloe Bakst, and Atticus Raasch. The episode was edited by Katy Anderson, with technical support provided by Dan Tayag, Chris Phelps, and Joe Shrum. Additional support was provided by Leanne Elston and Erin Collins. We'll see you next week.