Winning the reimbursement battle: Where AI can help and what still needs a human

Marla: [00:00:00] Welcome everyone, and thank you so much for joining us for Winning the Reimbursement Battle: What AI Catches and What Still Needs a Human. I'm Marla Raineri, Head of Clinical Innovation and Clinical Strategy at Prompt Health, and I'll be moderating today's discussion. And when we say winning the reimbursement battle, we're not talking about fighting with payers.

We're talking about something that's within our control today, and that is how do you maximize every dollar you've earned while building a revenue cycle that can scale? And every year, collecting payment for the money you're already owed has become more complex. Uh, we're seeing this more and more. Payer requirements continue to evolve.

Administrative work continues growing. Margins are tightening, and billing personnel is really hard to find. So at the same time, we have demand of patients, so practices are continuing to grow, and they're opening up new locations, expanding their services, seeing more patients, but they're trying to do it without eroding margins.

And that's really important because once we wanna keep seeing more patients, we wanna do it in the best way possible. So that's really changing how leading organizations are starting to think about revenue cycle. It's not measured by how-- it's not measured by how many people you have. It's not measured by what's going on in terms of that.

It's measured by how intelligently your operation works. And the organizations that are outperforming today are the ones [00:01:30] building revenue cycle operations that combine experienced billing leaders, efficient workflows, automation, and AI, each doing the work that they're best suited to do. So when people spend less time chasing preventable issues and more time on the complex problems that actually require their expertise, all while reducing compliance risk, that is where we really start winning.

And that's what today's discussion is gonna be centered around, building the modern revenue cycle and what that looks like. We'll explore where AI creates meaningful value, where experienced judgment still drives outcomes, and how organizations can build an operation that maximizes payments while creating the efficiency needed to scale confidently.

And we've really assembled a fantastic group of panelists today. We've got enterprise revenue cycle leadership, we have AI innovators, and customers who have experienced this transformation firsthand, and we're all really excited to dive in and help all of you. So before we jump in, just a few housekeeping items.

We really want today to be conversational and interactive. So there's a chat on the side, please feel free, start introducing yourself, tell us where you're from, how many clinics, um, talk to everybody in the group. And if you actually have a question, we want you to use that Q&A button at the bottom of the screen.

That's really important, because we do wanna get to everybody's questions, [00:03:00] and we'll do that at the end. But if we don't have enough time, we will be able to email you later and answer it afterwards if you use that Q&A button at the bottom. And then finally, we are, um, recording this whole session, so you will have access to it after the, the webinar, and you can share it with your colleagues, and you can go back and review.

So as I said, we wanna get all of your feedback and hear from you, so we're gonna start with two polls. The first one, you're gonna see it pop on your screen, and we wanna know from you what is the biggest RCM challenge your practice is facing today? Is it preventable denials, slower payer payments, AR aging, staff or billing capacity, lack of visibility, or prior auth or eligibility issues?

So if everybody could fill that out, and then you will be able to see results in just a minute. And I love seeing all of the messages in the chat. Keep them coming in. We love to see who's here with us today. All right, great. Those results should pop right up Okay. Prior authorization, that's a big one on here.

All right. Great. And then I'm gonna put one more poll up here for today, and we're gonna ask you how you describe your current use of automation or AI in billing. [00:04:30] So are you not really using it, exploring it but haven't implemented it yet, using basic automation, using AI limited, or really doing it through all your billing systems, um, or actively using it in your RCM?

Okay, great. Gonna have people fill that out, and we'll get to see those results as well

All right. Those should pop right up in one minute for us Okay. So we've got a few people actively using it. That's great. Um, and most majority are exploring it but haven't implemented anything, and then all across the board for the rest. So perfect. Love to see, uh, people starting to experiencing it or at least being interested in how to utilize and do the modern RCM cycle as we speak.

Okay. So we're gonna introduce our panelists, 'cause as I said, we've got some really great people for all of you, and we do want you to be able to ask them questions and hear from them. So we'll start with Brittany Amaral. Brittany, such a pleasure to have you today.

Brittany: Thank you. I'm so excited to be here.

Marla: And Brittany is the revenue cycle manager for Highbar Physical Therapy, which is an outpatient organization with more than 60 clinics across Massachusetts and Rhode Island, and she has 20 years of experience in outpatient PT and oversees the organization's revenue cycle from payer relations and [00:06:00] authorization workflows to billing operations, compliance and analytics.

And her team implemented Prompt in March of 2022. And since then, Brittany has helped lead the evolution of Highbar's revenue cycle by leveraging AI and automation to streamline manual workflows and free her team up to focus on higher value work. Thanks, Brittany.

Brittany: Thank you.

Marla: We also have Elizabeth Navarro.

Elizabeth, such a pleasure to have you.

Elizabeth: Thanks for having me. Great to be here.

Marla: And Elizabeth is the director of accounts receivable at SportsCare Physical Therapy, which is also an outpatient rehab organization with more than 70 locations in New Jersey, New York and Florida. And she's been a dedicated member of the team for more than 22 years and has over 30 years of experience in revenue cycle management.

She has deep expertise in billing operations, payer management, process optimization, and is passionate about developing high performing teams while leveraging technology to create more connected, efficient, and patient-centered revenue cycle operations. And her group implemented PromptHealth in 2026, uh, March of 2026, pretty new, and since then she's been helping lead the organization's modernization of revenue cycle workflows while supporting its continued growth.

Thank you, Elizabeth.

Elizabeth: Thank you.

Marla: We also have Dana DeYoung. Dana, pleasure to have you as well.

Dana: Thank you. I'm super excited about this topic. Very exciting. [00:07:30]

Marla: And Dana began her healthcare career as a psychologist before spending the past 26 years in enterprise revenue cycle leadership, and she recently joined PromptHealth as our director of enterprise billing success.

She left a leadership role with one of the country's largest outpatient PT groups, um, to join Prompt because she believed in the company's vision for transforming revenue cycle technology, and she now helps practices bridge that gap between technology operations and sustainable financial growth. Thank you, Dana.

And we also have final panelist today is Pedro Teixeira. Thank you Pedro for coming.

Pedro: Happy to be here. Hi everybody

Marla: And Pedro is the vice president of AI engineering at Prompt Health, where he leads the company's AI strategy and development across clinical and revenue cycle workflows. He's a physician scientist by training.

He also earned his MD, PhD, and master's in biomechanical informatics from Vanderbilt, and went on to completing biochemical sciences and computer science at Harvard University. So very well accomplished, to say the least. Um, and he previously co-founded PredictionHealth, which is an AI company focused on clinical intelligence, which was acquired by Prompt and embedded within our system.

Um, and he works, centers, his whole work centers around using AI to eliminate administrative burden while performing clinicians and revenue cycle teams, not replacing them. Well, thank you, guys. It really is great [00:09:00] to have operational leadership, technology innovation, and real world customer experience talking today and represented.

So before we get into the technology part, I do wanna start with understanding what's changed in the last few years and why we're where we are at this moment. So Dana, what, from your perspective, has actually changed in the payer behavior over the last few months that's making payment collections more difficult for practices?

Dana: Well, that's such a great, um, topic to start off on and there was-- if anyone has the chance, there's this really great article in HFMA's, um, magazine, and it was called Battle of the Bots, and it actually addresses this exact same question and, uh, topic, and it's so enlightening to see. But I'll spoiler alert, has to do with Battle of the Bots.

I th- I'm sure everybody can, um, figure out what that's gonna be. But I think for me, really the first payer behavior has really been how aggressive payers have been with their denials. It, it's been incredible, but and I think what makes the difference is historically denials really weren't that hard. It was like maybe missing an ID number or just missing some information.

And I think now we're looking at really nuanced issues that are a lot more difficult to figure out, such as medical necessity or documentation specificity. And there's also this new thing, it's called, um, precision denials, and it's where only one [00:10:30] small little piece of the claim is being denied. So now what it's doing is it's taking more people, more time, trying to figure out exactly why those claims are being denied, 'cause it's not the whole claim.

And then we have to, of, of course, solve. We don't want that to happen again. So it's a little bit more in depth. And I think the next one that I would say too, prior authorization. So I was kind of excited to see that's a pain point with-- not that I'm excited about the pain point, but I think everybody shares the pain point about prior authorizations.

And this one became a lot more dyni- dynamic over the last couple years because it- It stopped being, "I just need an authorization number." Now it's turning into, "I need an authorization, increased number of requests for those, um, different services, mid-course authorization changes." So authorization management now turned into an entire workflow needing an entire team instead of just a one-time administrative task.

And then the last one is policy volatility. And what I mean by that is just how quickly all of the payer rules are changing, and we don't have the staff. The team doesn't have the staff. So it's really requiring someone to comb through all of those changes to make sure that we are aware to prevent those denials.

Marla: Absolutely. That is a great way to sum it all up. Um, and what would you say are you seeing that are the greatest source of payment leakage today?

Dana: So there's, there's always... [00:12:00] There's gonna be a couple of those. I- one that's really not talked a whole lot about that I think it's as important is paym- um, patient payments, over-the-counter collections.

Mm. And I think this really started when we started to shift the payment model where we went to the high deductible plans. So now we have a lot of that payment that we are expecting is coming all from the patient. And that responsibility is a lot of a burden on the payment, on the patient. But what happens too is that if we don't collect those payments upfront, there's a lot of studies out there, but now we're hovering around, like, that 30 to 35%.

Once that patient leaves the practice, we're not gonna see that money. And as that time gets further and further down the road, it drops even more dramatically. For me, that is something that I look at every single month is looking at where my patient collections are. And then the other one is, this is a reoccurring theme, is prior authorizations.

Um, you know, like I said, it is now a workflow, and I think the problem is, is that unfortunately we don't capture upfront. A lot of systems don't capture that, so now we're, we're chasing those authorizations after the fact. And unfortunately, a lot of payers don't do retros.

Marla: Yes, and as you said, that's all money that is already our money.

So we want to collect that. We don't want it to let pass, and the longer it goes, the further it doesn't come to us. Um, so would love to know from your perspective, Brittany, with all of [00:13:30] those changes that have been happening over the past few years and before you really modernized your RCM, where was your team spending a lot of time manually that shouldn't have required experienced billing professionals spending their time?

Brittany: Great question, Marla. And I'm gonna piggyback on what Dana said a little bit in regards to authorizations. Having been in the PT world for so long, I've seen a lot of change. Good, bad, ugly, right? Ugly to me is the unnecessary admin burden that so many payers have layered on over the years. At Highbar specifically, we spent an enormous amount of time copying a PT's note, copying what the patient has sent, and then putting it into a portal just to continuously check that portal stat- for status over and over again.

So for us, we want our staff doing the hard work and focusing on patient care. Copying and pasting information into a portal is not rocket science, right? It's not something that we need a human to do. So we really want our team to focus on, you know, worrying about getting that patient the authorization they need, turning around any medically necessary denials.

That's what requires a human

Marla: Perfect. That's really good to hear. I mean, again, and as you scale with more clinics, you don't wanna just keep adding people to input information in. So Elizabeth, I'll pass that question along to you. What was the moment [00:15:00] that you realized your billing operations weren't going to scale the way it was set up, um, and what really stirred that, that realization?

Elizabeth: Well, that moment we realized our, our billing processes could not scale was when the growth, uh, started creating more manual work, um, instead of greater efficiency. We were relying on different systems, spreadsheets, uh, workarounds, repeated follow-up, uh, to track claims, uh, paym- uh, payments, denials, outstanding balances.

Um, as volume increased, it became harder to maintain clear visibility and accountability across the revenue cycle. I knew we needed a platform that could connect the, our workflows, reduce the manual touchpoints, and help us identify issues before they affected reimbursement. That realization led us to look for a more modern solution that could support both our team and the continued growth of the, the company

Marla: And you guys are both really large organizations, um- Yeah

you know, 60, 70 clinics. So I think this is fantastic 'cause I'm sure we have people from all different, from 1 clinic to 5 to 10 on the call, and it's showing them, well, do it earlier because when you get bigger and you start adding more people- Yes ... doing that busy work, it doesn't pay. You're just paying to do the more work.

So what I hear from you [00:16:30] guys is that adding more people isn't the answer. Sure. Modernizing the revenue cycle and utilizing thoughtful AI with automation may be more of the answer. So with that, um, there is a lot of buzz, though, around the terms automation, AI- Mm-hmm ... agentic AI, and it's everywhere in marketing.

Could be really hard to listen to the market and to know what to trust, to know what it means, and to know where it fits into your own company. So Pedro, can you help everyone understand the difference, and more importantly, how you determine which technology is best suited for different parts of the revenue cycle?

Pedro: So happy to do so, and I like to give folks, um, tools, you know, rules of thumb, things that they can apply to understand, like when does it make sense, when it doesn't make sense. So starting with AI, um, artificial intelligence. So it should be doing something that is kind of hard. Uh, so something where there's a clear rule.

So, you know, if I have a denial code that says I needed to attach the medical records, then you don't need... You... It's an if/then rule. If this, then we need to resubmit with that. And, uh, but if you have something like, oh, something needs to be phrased or talked about in the note, that is now something that's a little bit harder.

There's so many different ways that you might say it. There can be different words that mean the same thing. You know, did you actually describe the medical necessity right for this? That's a harder problem. That's not an if this, then that, right? It's if something like you have to interpret, in this case, the text of the note.

Um, maybe [00:18:00] interpret what was discussed in a visit. When you're going to that point where the input data is messier, usually something that people are writing or saying, that's a good use case for AI. Uh, and then that kind of sets us up to talk about the agentic stuff because agentic, uh, well, agent- agentic, agency.

If I have agency, I get to decide what I'm going to go do next. And again, if it's if this, then that, you don't want something that's gonna have agency 'cause sometimes it might decide to do something differently, and then you're getting a bad outcome, right? So the AI is a very powerful tool. It's kind of like having the sledgehammer and, uh, sometimes you don't want to use the sledgehammer.

You want the little, you know, very precise, you know, hammer. And so combining automations with AI so that if it is a complicated problem where it's gonna change a lot and you don't have a simple rule, then you should be using AI and designing the system to, you know, benefit from the strengths of AI- But then also keeping it well contained and monitoring, uh, it to make sure that you're getting good outcomes with things.

So that's a way to think about AI. That's a way to think through when you actually need something agentic and maybe when you don't need something agentic. And I hope that helps folks kind of parse through the noise, and hopefully we do a reasonably good job of accurately describing when it's an automation and when it's more AI, uh, to be helpful, uh, to the industry.

Marla: It's great, and I, I think you said that right. The, the goal is to apply the right intelligence to the right work because if you have the wrong tool and it's way bigger, like agentic to something that should use automation- Mm-hmm ... then you actually are more at risk for compliance issues and it thinking on its own and delivering something different than an if then statement.[00:19:30]

So that-

Pedro: Exactly ...

Marla: really helpful framing that up. Um, and I- what really matters most is not just the technology, but it matters if it proves outcomes. So let's make this really practical and find out a little bit more, um, how do we produce the outcomes we want? And Dana, I'm gonna ask you, when you evaluate the health of an RCM operation, what are the KPIs that every owner should understand and be evaluating to know their standard and baseline before making changes and then seeing if the outcomes are there?

Dana: This is truly one of my most favorite discussion points. But first I would like to just reframe that question just a little bit, 'cause in my mind the question should be: what do I evaluate the org- how do I evaluate the organization from a ca- a RCM perspective? And the reason why I say that is it's really, the, the RCM is really just working in tandem with the organization.

It's not RCM specific. The RCM is the outcome, and that's what I love because there's so many pieces and parts. And with that being said, I also wanna say too, the, the KPIs that I look at, they're never in a silo. It is always in tandem. I wanna paint the most pac- most accurate, vivid picture that I can. So I always look at all of the KPIs together.

But the first KPI that I look at, and I'm not gonna lie, I get a little bit excited when they come out at the [00:21:00] beginning of every month, is your net collection rate. I'm, I'm that person that gets excited by numbers. Net collection rate, and then your either payment per visit or some people refer to it as your net revenue per visit.

And the reason why is because it really touches on many different aspects of revenue cycle. For example, contracting, coding, payer mix, units per visit, and adjustments. Everybody loves to see where those adjustments... But when you put it all together, that's why I really like to look at those KPIs. And through those spec- specific KPIs, you're really measuring the financial effectiveness and identifying revenue leakage, but it also shows how well your billing and your collection teams are working together.

And the second KPI that I don't think gets a lot of traction, um, and I did mention it before, I still look at every single month. I like to look at what are our patient collections, um, because I truly think this is one of the very, the one of the quickest way that we can have a positive impact on our cash flow.

So making sure how we're, what kind of job are we doing at collecting our patient balances? The other KPIs, and this is just a highlight, I'm sure everyone looks at these as well, but I want to make sure I'm looking at our denial rate and the reasons why those denials are coming in. My clean claim rate, because that can point to a workflow issue.

Days in AR, and then the last one is the aging bucket. I like to look and see where each one of my-- where all my dollars [00:22:30] are. But I specifically look at 90, 180, and 365, because to me, those are really indicative of how... is that money gonna come in or not? What kind of workflow issue do I have? And then I break it down by, by payer

Marla: That's great.

And I see some people in the chat asking maybe you put those three-- those few things in there so they can see those too. Absolutely. That was a great recap. Yeah. Um, and Pedro, when the system is built holistically to combine automations and AI, like you were talking about earlier- Mm-hmm ... the right tool for the right problem, um, where do you see the biggest lift?

Out of all those things that Dana described as the ones to-- metrics to look at, where do you see the biggest lift and the most bang for your buck and outcomes when you add these right tools, such as what Prompt's doing with AI and automation? You

Pedro: know, so, um, I mean, there's obviously very different experiences for folks.

Um, so there's lots of different parts that you described, and different folks have challenges with different ones. But the, the holistic aspect I think is really key, and I think it's helpful for me to also highlight that holistic means not just like when you're doing the RCM, you know, part of it, right?

Um, that goes all the way back to when the person first came in the door, right? When they provided their insurance card, and did the information even get made-- you know, put in there accurately. And if you create like a funnel problem, a small issue at the very, very top of that funnel can have a huge [00:24:00] impact down, uh, below.

So getting that information collected, having-- ideally using OCR, right? Accurately pulling information so you have it, so then you have the elig-eligibility verification check done, right, is a, is a huge part of that. Also, again, uh, that, that focus at the very, very top of the, the funnel also includes the originally, uh, the original visit and then the documentation of that visit.

And so for focusing in on AI, for example, then one of the great things that it can do, uh, is, uh, for example, Sidekick. Sidekick looking through and designed to pick up from the conversation that things are documented thoroughly, that you're mentioning things like, you know, is there a skill need here? Um, and things like goals, right, which we could dive into, um, as well, are things where you have to write it the correct way.

And there are certain rules of thumb or guidelines, descriptions in the compliance, you know, documents of course, of like what you're supposed to do, but it's not always an obvious rule of if this number, then do that. And so I would f- you know, if you wanna look for maximum impact, maximum pickup, I think that top of the funnel part makes just s-such a huge difference, and it's easily-- easy sometimes to underappreciate, although I'm sure this group very much appreciates 'cause you often deal with the consequences of things at the top of the funnel not being necessarily well done and fully buttoned up for you.

So that, that's where you get, like, really huge leverage. And when you're thinking about the AI, again, think about those tricky things. You know, someone has a card, and it might be all smudged or whatever, and then a human keying it in versus the AI accurately pulling it out to get you the metrics or interpreting the [00:25:30] conversation or just making sure that the note, even if you're not using Sidekick, you can process the text that therapists are writing to make sure that those things that are important for compliance are also foundational for RCM.

Marla: And that holistic view, like you said, is the second that they're scheduling and the AI online scheduling that Prompt's utilizing and the way the information is OCR is reading it and putting it into the platform, and that feeds the next step and the next step, is why one of the reasons that Prompt has the lowest, uh, denial rate in the system.

Most people, that's their top, but once you start utilizing Prompt, that denial rate goes down significantly. Um, and authorization rate. So I'd love to hear from you guys. I'll start with Brittany. After switching to Prompt, how did your metrics change and what measurable improvements stood out the most, um, with all of the AI automation and workflow changes that you've made with technology?

Brittany: Yeah. Honestly, that's, that's great, and it's so hard to think of which just one of the biggest impacts, right? Because RCM feels all of the effects, not just, you know, certain things. It's really full circle. So one of the biggest, I guess, would- I would say is giving the providers the ability to submit claims electronically at checkout.

That was a huge impact that Prompt allowed our team to do. We worked hard to get our payer rules to flag a PT, either to correct something or if something needed to be updated before [00:27:00] they even checked out and submitted that claim. So that means our billing staff, it wasn't something they had to go back and fix later.

And then adding Prompt's level of failed edits and rejections, that allowed my team to just look at the claims that they needed to look at. That allowed us to move three full, uh, full-time staff to... from claims review to a more rewarding high-value role within RCM. It also changed our days to submission, which for me is huge- Mm-hmm

and our clean claim rate. So our days to submission went from 7.4 days, days, that's crazy, to 1.2 across all brands. That's like a full week. It's amazing. And what that means is the payers get our claims faster and we get paid faster. So hurrah. It's amazing. And then just doing our payer rules alone, it improved our clean claim rate by 8%.

That's huge.

Elizabeth: Um.

Marla: That's huge. That, that's fantastic. And as you said, you were able to take those teammates and put them in higher value tasks and repurpose them, and I'm sure they found that a lot more rewarding as well.

Brittany: Oh, yes. They were very happy with me.

Marla: And Elizabeth, what about from your perspective? I know that you guys are still newly in, about four or five months in, but which KPI do you feel changed most noticeably once billing [00:28:30] performance became more visible and AI started catching issues earlier?

Elizabeth: So like Brittany mentioned, um, for us, uh, the KPI that, um, was the most noticeable from the beginning was, uh, clean claim rates. We went from an 81% on a best day, which wasn't many best days, uh, to 90% with Prompt. Because of the great visibility into the billing, uh, performance, combined with Prompt's ability to identify issues earlier, um, it helped us correct problems before claims reached the payer.

Like Brittany said, um, having the therapists, uh, submit claims from checkout, that was a game changer. That reduced preventable denials, limited rework, because I detest rework, and helped claims move towards reimbursement faster. Um, the improve- the improvement also gave our team clear accountability and better insight into which workflows needed attention and, like Brittany said, moving a staff member to a, a better, um, more rewarding, more, um, uh, position

Marla: Yep, that's... Thank you. That, that's great. And, you know, and like you said, you're, you're still, you're still seeing that shift- Still

Elizabeth: very new ...

Marla: awesome 'cause you guys are still working through all of that. Um, and love that you are catching those issues before they happen and preventing the problems [00:30:00] to the money that is already owed you.

Elizabeth: I think what-- if I can add one thing, is authorization. That has been the biggest reward because, um, with many systems, you don't get to receive, um, denials, or you get to catch them in a short am- a short amount of, uh, timeframe. Right now, claims are not going out until the authorizations are received, and it's in the system, and we get to have that opportunity to make sure that everything goes out clean.

And it's, it's amazing. It's amazing.

Marla: That, that is. And Pedro, I wanna know from your mind, um, how is that happening? So what is doing that prevention? What kind of patterns is a model, an AI model, actually good at spotting- Mm-hmm ... before a claim goes out? And why is that helping practices so much? If you can explain between the types of AI and what is going on there.

Pedro: Yeah. So there's-- I also have to give credit to some of the automations, and that's something that Prompt, of course, has been doing for a good while. And once you find certain patterns, which I'll not nerd out too much, but you can use AI to scan really messy data, including structured data, and to find, like, "Oh, this is really influencing the outcome."

Like, we have a, of course, a no-show and drop-off prediction model about scheduling and whether somebody shows for appointment, and that's actually structured data. You can try to predict or forecast, uh, predicting and forecasting, of course, denials. And doing the data analysis there, also really, really impactful.

But then one of the nice things that can happen is that, [00:31:30] yes, you can still run the AI models to try to predict and forecast, but you can also make it very actionable. And so, uh, when people are putting things in, and this is something that each site is going to be partially responsible for, having good automatic rules, good automations built in for your payers, appropriately checking the CPT codes, uh, checking the modifiers, and leveraging those tools is very, very impactful, but something that does require some tending to keep up to date.

And so those patterns can be highlighted of things that should be automations, and I know we provide some there. But then also, you really wanna make sure to keep those up to date. That is frequently a thing where we see it slip through, and you do ideally want to get that at the earliest point. Ideally, the system telling you in the moment, maybe when that code is getting put in incorrectly for a given payer in there.

And that is, though, to be fair, right, an automation. Then specifically on the AI side, I didn't really get to talk about it as much there, but also a frequent issue that we see is the goals. The way that the goals are phrased, now again, a reminder, this is something that's not as easy to put a rule because you can't just look for like five words that if you have those five words, then everything's fine, right?

It's a little bit messier because the way that you phrase it to have, for example, a SMART goal, right? Measurable, attainable, et cetera, timely. Um, folks can phrase that a lot of different ways. And very frequently, the way that the goals are stated can potentially undermine that the work is skilled, right?

Uh, or that you've already achieved things. If you've already achieved all your goals and you're still seeing the patient, you know, now you can end up in sort of a different problem. So again, you have automations, things like the rules that you [00:33:00] can leverage, very, very impactful for folks. You have to keep those up to date.

And then again, looking at another example from the actual text of the note, the goals really do have to be phrased well. I'm sure folks have seen stuff where you get denied because they went and they're like, "Oh, you've achieved everything here. You know, you shouldn't be seeing, seeing the patient anymore."

But that becomes, again, a very messy thing that actually does merit an AI.

Marla: And I love how you mentioned that automation is something that we've had in the platform as a staple for quite a long time, and now we've layered on the AI to catch those higher thought-provoking, um, pieces. And Elizabeth, you had mentioned that your denials changed and you had much cleaner rates and so would love to know what kinds of denials you see now, uh, that or that you see that are being prevented and being caught earlier than before with all of these automations and AI that's occurring.

Elizabeth: Um, so the type of denials that we're seeing, uh, now is, um, that we're preventing, um, would be authorization, number one. That's number one. Um, we are seeing, um, less and less of the clinical medical necessity. Um, I believe with Psychic, um, it has helped the clinicians with their documentation, so it is helping.

Um, catching, uh, benefit level, um, processes, um, their, their, um, max benefits. Um, great, great tool is also [00:34:30] with, um, the KX modifier for, uh, the p- the one one nines for Medicare and any other of those, um, insurances. Um, that's, that's been the one of the biggest.

Marla: That's, that's great. And, and like we said, it, I think when we did the survey at first, that authorization denial was the number one of this crowd, so it's good to show that there's a solution out there for that.

Um, and, and end in sight, right? Um, and Brittany, not everybody trusts AI right off the bat, so was there a period of trusting automation and AI, um, before using it with your own eyes? And what got you comfortable with that, and how did you make that transition?

Brittany: Oh, yes, 100%. It was... I originally started using AI with trust issues, so it took a while.

But, um, I have a lot less gray hair now that the processes are going through, because it, it was pretty tough, but it's so rewarding. So for me, you have to think of training automation the same way you would train a new hire. You give it the training, and then you're double-checking the work for a while.

There is a period where you're essentially doing double the work to make sure accuracy holds up. But the sooner you do it and rip off the Band-Aid, it's gonna be able to identify different scenarios, identify things that you need to change immediately to start trusting that process. [00:36:00] A good example of trusting and a workflow that for me worked very well was insurance verification.

For some payers, it's monotonous for a person to sit there and verify and input data and do all of that, but automating it for certain payers has made such a big impact for our organization. And to get there, we did have staff running the same verifications manually alongside automation for about two full weeks just to confirm everything is the exact way, one, our organization wants it, and two, the way Prompt can, can really understand and put it back into that, you know, there's boxes are being checked, documentation is being put in.

But honestly, it's so rewarding. Like I said, those first couple weeks everybody will probably have a lot more gray hair. But in the end, it is such a rewarding piece that I can't speak to enough.

Marla: That's, that, that is so nice to hear, and I, I love that, you know, you said there are trust issues in the beginning, and you got past them, and the quicker you rip the Band-Aid off, the better.

Um, but of course, technology isn't always meant to replace every part of the revenue cycle, and we do want to talk about that as well, 'cause some work still requires experience, relationships, and judgment, um, and that billing expertise. So, um, Dana, would love to hear from you a little bit about once prevention has [00:37:30] done its job, what's left that still requires a phone call, a judgment call, or a relationship with a payer, um, and where do you feel that human intervention is best utilized?

Dana: I, I'm so glad that we're addressing this because there are some pieces that we really need to make sure that we're paying attention to. Um, I think, and I, and I've, have some examples, I think denials are a b- are a big one as well. We can automate quite a bit, but even though not everyone's gonna agree, deni- Denials are not always a mistake, even though we would wish they were.

Um, some denials are accurate, and even with the best prevention tools, there's always gonna be legitimate payer disagreements, whether it's due to medical necessity, coverage interpretation, policy rules. In these situations, they require clinical and revenue cycle expertise in order to determine why and how we can actually fix that.

And what it does is it requires a team to actually work together and say, "Eh, is that workflow actually right? Is it accurate? Is there something we need to do to tweak that?" And that really takes someone in our expertise to be able to determine that. And I, uh, and I think the other thing are the complex appeals.

Um, those are absolutely probably the-- nobody likes to do them, but they're required. Um, but I think the flip side to my first point is not every denial should be written off or [00:39:00] automatically resubmitted, and it takes a person to understand what type of appeal. Certain payers require certain information.

They like to see it certain ways, and we're talking a little bit more of our more abstract payers. But we have to really be able to sift through clinical documentation, payer policies, contract language, and then what kind of treatment did the patient have? That's gonna vary from, from, um, from patient to patient, and we have to be able to understand that.

And I think probably one of my favorites, relationships really matter still. Payer relationships, I couldn't, I couldn't talk about this anymore. Um, e-- and from a contracting perspective, that's probably the number one thought. But also, and I have a great example of this, um, in a previous experience, and I'm sure everyone can probably figure out who, but we had a payer that denied over three thousand claims.

It was a system issue. And the response back was, "You all need to reprocess those." And the answer was absolutely not. We used our pay-- um, payer relationship, explained to them this is their issue. We worked through that. So now we had them reprocess those three thousand claims, so we were able to alleviate that work from our team, and then we were able to get our cash quicker.

Wasn't our issue, but I think if we didn't have that relationship, we wouldn't have been able to have those conversations. And then I think the other piece, let's talk a little bit back about the [00:40:30] patient. Um, our patients, no matter what we do in our workflows, they don't know what that workflow is. We still have to be able to have those conversations with that is a human Humans wanna talk to a human.

They wanna be able to understand. In that patient experience, we need to be able to combine high tech, which Prompt is amazing at, but they need that high touch. And when you combine the two, that's where the magic happens with your patients. So if they have a question about their bill, they wanna talk to someone.

They wanna have that connection. And then the other piece of that too is what if we have to have an appeal? We wanna have a good relationship with those patients so that they, we, they can work with us on those appeal for our therapists to get paid for all of that incredible work that they did. And then I think the last one that I would probably bring up, and it's probably the least favorite of everyone on the call, is compliance audit and revenue integrity.

Ugh. With all of the, all of the processes that we have in place, and I think we're really good, we're getting really good at being able to, um, identify unusual billing patterns or potential compliance risks. But what we need in addition to that is professional judgment. Regular account audits are absolutely essential, um, especially when you're working with regulatory guidelines.

In working through these audits, and I think this is another key piece, is that we need to be able to provide feedback [00:42:00] back to the clinician and operations. And that is on our team to be able to facilitate those conversations, 'cause that's only going to drive a better experience not only for our patients, for our team members that we work with, and it's also going to, um, have a positive impact on our finances

Marla: Wow, I couldn't have- Good

couldn't have heard that any better. I think that was so perfect- Mm-hmm ... to hear and to understand the insights of where that human expertise fits in. And Pedro, would love to hear from your perspective what we've done in Prompt to allow those human expertise to do that part of the job and filter onto what is that piece and what is not.

Pedro: Yeah. So something that's kind of, um, different about doing AI things is that, um, even when you figure out what you want to build, there's usually stages that you have to go through. So initially, you know, to, to Brittany's point earlier, right, those first two weeks, like you're gonna have very careful oversight.

There's this whole oversight period before a feature even gets launched, right, where you're, you're on top of it, and ideally, it's great to have somebody in the workflow. So, you know, you had Clinical Sidekick just as a historical example where folks can think back through, through that, right? It's, it's great now.

You click it, it runs, it does this, this task for you. But from the early days, there was a lot of more human involvement and it presenting the drafts and humans accepting and us learning from those flows. So similarly, because payers are so different for each one of them, because there are different use cases, because they change things [00:43:30] over time, it's really great when the AI is still learning, as we're still tuning in, and as we're learning each new payer with their new rules and things that they require you to do, having a human in the loop with the person, with the AI being the, their sidekick, being a part of them, is a really, really critical step.

So it's in alpha, uh, right? But Billing Sidekick is something that, that, that is where it is right now. It's helping in these portals to get the content filled in. Um, and again, it's in an alpha. If people are interested, you can reach out and be put on the wait list if you reach out through support, but again, early stage.

Um, and that's in workflow, but that's part of the process, so that it can go off and do more and more of the work by itself and actually merit the trust and, and infrastructure in place so that it can be more agentic. And so, and when you're thinking like the humans in the loop, that's a really, really great thing to do.

But at some point, if you need a human judgment, if it's a more complicated thing, right, then it's great to have the ability to lean on that person with that judgment and with that experience and to focus their time on things that are much harder to automate, which I think folks also enjoy doing less repetitive work.

Marla: Absolutely. And Brittney, you brought that up earlier about how you repurpose your staff to do higher value work. So for you, um, how did the automation and AI change your billing staff and what they do day to day and how much they do of it? Where do you see that happening and what are the changes there?

Brittany: Yeah. Um, I guess I would say it really, it's really both. The biggest shift for us was the [00:45:00] kind of work that the team does. So before, it was very manual. Mm-hmm. Reviewing every claim, chasing down authorizations, copying and pasting in between systems. It was all manual. Nothing great about it. So now that automation handles the repetitive rules-based work, my team spends their times on the things that actually needs them to focus on.

Digging into why a claim got flagged, pushing back on denials, being that liaison for when an insurance pushes back on treatment, and that for me is huge because that gave us the ability to shift from being reactive to proactive. And it allows us, just like Dana said, to go and build relationships with payers because it does matter.

It is such a huge impact to an organization when they have someone at the payer level that they can trust and reach out to, and that's something that, again, automation can't place. So it's, I guess I would say it's not that they're doing less or it's not that, you know, it's taking away from them, it's more it's leaving what they need to focus on and it's leaving the higher value work that is allowing them to use their judgment versus using their patience.

Marla: Perfect. Perfect. I, I love that explanation and, um, and if, uh, it'd be great if, uh, you can, Elizabeth, go on and tell us a little bit of a situation that a claim actually you're glad that a person handled it, just like Brittney explained or Dana, but [00:46:30] if you have another example, that would be fantastic.

Elizabeth: Um, for me, it's more like, um, I'm always proud when a team member handles situations involving patients who are confused, worried, or frustrated about bills.

Um, a system can identify the balance and explain the transaction, but it cannot fully understand the patient's circumstances or how the situation's affecting them. A compassionate team member can listen, explain the insurance outcome in simple terms, um, review available options, and help the patient feel respected rather than processed.

Um, technology should support that interaction, not replace it. Um, the human connection is es- especially important when discussing financial responsibility 'cause that's, you know, that's something that patients are very confused about because trust, empathy, and clear communication can completely change the patient's total experience So it's where that is great for us

Marla: I, I like that emphasis on trust, empathy, and patient communication.

Um, and I've seen a lot in the chat, and I know we've got com- questions coming up in a few minutes, but a lot of people have asked about what's coming and what's next. And that brings us to Pedro. Um, really, what is the future and what is happening right now at Prompt? I know you mentioned one or two things, but tell us about what practices are preparing for next, and what's most exciting to [00:48:00] you that we are building and releasing in revenue cycle.

If you can give us any sneak peeks.

Pedro: Uh, tiny sneak peeks. Um, so I'll start off tactical, and then I'll go to the strategic part. Uh, I, I, uh, I'm especially excited about some cool stuff there. But to start off, um, there is some very exciting stuff that's, like, in the alpha stage. Again, these are... So some of these are alpha.

They're, they're early. This is what's coming. Uh, but benefits verification is a really great area of a thing that's very repetitive, where you have to go to various different places. You know, you may have to go to a portal, right? You may be able to hit a, an endpoint. Um, and so folks are gonna start getting to see that for various payers right within the Prompt system, uh, in a way that they can have a lot of control and visibility for how it works and be able to flip it on, uh, for different payers.

And we're very excited, uh, about that feature. It's, again, at the top of that funnel where you have such high leverage impact on so many things that happen later on. Uh, additionally, um, you have, uh, Billing Sidekick, which folks may have heard a- about from the V10 webinar. Again, it's in alpha. Uh, so, uh, you, uh, can reach out to support and be asked to put on the wait list, uh, for it if you're interested.

It doesn't guarantee, but, but then, of course, those folks will get to hopefully play with things sooner, uh, and teach us about how we can improve it sooner. But that's now in workflow, and so whether you're in Prompt or you're in a portal, that's an opportunity to get a faster, um, gathering of information with claim statuses.

There's, again, sometimes verifications you have to go do in a, uh, portal as well And then we're gonna have, uh, a lot more stuff coming, uh, in the, [00:49:30] the future, um, related to, uh, then expanding that to things like prior authorization, uh, to-- and making our, uh, automations that we mentioned earlier, uh, automations with rule actionable logic incorporated to it that will augment, uh, denials.

That's also something folks will be seeing in the near future. And again, automations and AI work very much hand-in-hand. You, you know, may be able to trigger, uh, AI steps with, uh, automations, which will be really, really great in the case that, you know, you have to call and navigate an IVR. That's a messy process, right?

That that's really, really great for an AI agent and actually merits it deciding at each point, what do I do next versus waiting on a human to somehow get looped into the, the IVR navigation process. Um, so those are some tactical items. I hope that's a good enough sneak peek. Uh, strategically, you know, when you ask about how excite-- or what I'm most excited about, um, I, you know, would-- PredictionHealth team joined the Prompt team in part because we had so many folks asking us to build things that would be seamlessly integrated into their EMR workflows.

And so when you ask, you know, what's really most exciting, I think it's getting to really expand beyond the focus that we have now. And I like to joke everybody deserves a sidekick. Everybody deserves a sidekick in anything across the entire experience, right? Maybe preventing that denial by being incorporated into the clinical sidekick conversation and perhaps a little bit of a nudge, maybe a little reminder there that then enables you to have the data necessary to do the repeat prior authorization, right?

There's so [00:51:00] many really exciting things that we can do by thoughtfully combining across all of our different services, all of the different parts of interactions that people have, uh, with Prompt, um, that I'm just super, super excited about. And I think that because we get to build holistically into one system, uh, we get to do some really, really special things that hopefully will be really magical, uh, easy experiences, uh, for all of the wonderful folks that use the system and, of course, the patients that they take care of.

Marla: Lots coming, lots to be excited about, uh, and continuing to help all of our, our customers and the industry in general, 'cause I do feel like Prompt pushes everybody forward, which I love. Um, but I also wa- wanna know, sometimes prior to, uh, Dana, especially you, prior to coming to Prompt, you have seen the market where there are times that just jumping right in and automation, um, AI wasn't necessarily the best.

It looked good on paper, but it didn't work in practice. So can you give us an example of that in your prior expe- experience, um, before you joined?

Dana: Absolutely. And I think this really ties in what everyone has talked about today. Um, couple of examples is automating a broken process. Um, you need to make sure that it is working accurately before you go down that rabbit hole.

I think the next one is probably really, really important is skipping the pilot phase. Some people get... and, and I [00:52:30] think this is very contagious. You get very excited. You wanna start it. But don't skip the pilot phase. Whether you're implementing a new software solution, maybe a new workflow automation, or integrating a third-party application in your EMR, I can't tell you how important it is to make sure that you take a step back, pilot each one of those processes, and just ensure that we can validate those outcomes and uncover any unintended consequences, because I guarantee you it will not...

Nothing goes as smoothly as you hope, right? Um, and then I think the last one I would mention, one of the pitfalls, we always talk about how fast. We always wanna go faster. We wanna process more claims. One item that I would say is that, you know, we have to be responsible for our company. We have to be, make sure that we're really great operational stewards.

But, you know, our mission and our goal is we have to make sure we're accurate in reimbursement, compliance, and we have sustainable financial results. And I really like to think of speed as a metric because the outcomes are our mission

Marla: Thank you. Wow, that is a great lasting impression and something for everyone to take away to be thoughtful about.

And I think that is one of the ways our company works, is being thoughtful and piloting almost in our own. We have our own billing cycle revenue services where we get to actually try our own automation and [00:54:00] AI out before releasing it to customers. So exactly what you just said, I think, is how we've built our models and work internally, so whatever we're putting out is really useful and thoughtful, and not putting anybody at a compliance risk or a risk that you need to then step backwards from.

Um, and Brittany, Elizabeth, one of you, I'll let one of you take this question. Just looking back, what advice would you give a practice that's just beginning this journey? From our first poll question, we had a lot of people say they're just starting to explore. What advice would you give a practice like that?

Brittany: Elizabeth, you want me to take it?

Elizabeth: You can. Go ahead.

Brittany: So for AI specifically or automation, my advice would be to start small and be patient with the process. Don't expect to flip a switch and have it be perfect day one. I would say pick a workflow that's really bogging down your team and talk to your team and s- figure out, you know, what is it.

What are you doing that you think doesn't require you to use your brain or to... That, you know, is just a push of a button or it's very repetitive, and I would start there. Like insurance verification or checking authorization status, because that's what's really gonna help drive the focus into where you can find those pitfalls of what can be automated.

And I would say, too, one thing I highly, highly encourage is having real conversations with your [00:55:30] team about why adding automation is important, because we never want them to feel like, oh, we're bringing in automation and we're gonna start to replace our employees, because that is never the goal. We want to replace things that they don't need to have, that they don't need to focus on.

They don't need to use their brain. They're not gonna, that's not gonna help them develop, right? We want to give them the tools and the space to be able to develop that. So we want to make sure that, you know, your employees and our employees can focus on things that's actually gonna be rewarding for them and to the overall organization

Marla: That's great.

And like you said, there's- no one is- there's no one single operating model. Everyone's a little bit different. So just pick one thing and start to do that. Um, and I am gonna put a final poll up for everybody that's gonna ask if you wanna learn more about Prompt or about SportsCare or Highbar, please feel free to fill that out.

And since there is not one operating model for practices that wanna fully outsource their billing, Prompt does do RCM services and do the billing for you. Um, or if you wanna try to bring billing in-house, when it was outsourced, Prompt also offers business coaching services to help you do that as well.

So the goal is to make sure you build the revenue cycle model that's right for your organization and to sport- support it with technology and expertise that you can trust. Um, so that's what we're here for. And Brittany and Elizabeth, thank you so [00:57:00] much for all of your input. Um, and I know a lot of people have questions for all of you, so we're not gonna get...

Oof, I don't know if we're gonna get to all of these right now. Um, yeah, let's see. "Can we talk about posting, how AI can help with that?" Maybe, maybe we can do that one quickly, um, from Hillary.

So Pedro, talk about what you were saying- Yeah ... and how AI can help with that.

Pedro: That is definitely a repetitive task. Um, now upstream always, of course, important to remember getting your electronic, uh, connection set up because you can sort of prevent the manual steps. So again, trying to think holistically all the way at the beginning.

So I just have to remind folks on that one, because sometimes you can just set that up and then don't have to worry about the back end part of it. But then if it is coming in in a way that you have to extract the data, that is definitely something we're, we're looking into. It is a impactful thing for us to, of course, have internally.

Um, but you know, a practical reminder you can apply today, and then, yes, that is definitely something that can be helpful or that can be assisted, uh, with automations and

Marla: AI. Okay. Perfect. Um, and Alicia asked, when Pedro, you were talking about some of the new features, she wants to know what's available right now where you said to reach out to beta or your CSMs.

Um, can you go ahead and give us some of those that are available in beta that somebody can test and pilot right now?

Pedro: Yes. So, um, if you specifically want AI, AI features, uh, [00:58:30] Insight, uh, is very, very powerful, has the compliance models built in, the CPT Insights model. Of course, getting that i-right is helps prevent it from ever being sort of a denials or RCM issue in the first place.

It gives you a lot of visibility into, um, the, the data as well and how you're doing. So that's a visibility tool. Um, Sidekick could be very, very helpful again for preventing. It is a very, very complicated fun, uh, fun to build, uh, it was a fun-to-build adventure, uh, tool as well. Uh, more on the RCM side for things that are available right now.

Again, the automations. Folks very frequently underutilize the automations. Um, so that's, again, a thing that's available right now. Sans your versions, the denial automations that I mentioned, that is a, is something that's coming, uh, further i-in, in the future. Um, and let's see, other things available today.

Um-

Uh, some of the other things I mentioned are just alpha. So, so again, if you really wanna be part of that early exploration, early feedback, you can definitely let me know about that as well. The billing sidekick, again, is an alpha feature, uh, is on the, uh, at that stage. But that is something that we've seen really great and increasing usage of as well.

Marla: Perfect. And then this was a good point. It says, "Is there any ways prompts can help those in RCM on this call or other interested to connect with one another on RCM issues?" Please put your contact right in that chat, uh, for each other. If you guys wanna connect on your own, go ahead and put that in that chat.

And if you did indicate that you wanted to connect with Brittany or you wanted to connect with Elizabeth or Dana, um, we will reach out to you for sure. And we're not gonna get [01:00:00] to all the rest of these questions, but we will answer them, any that was in the Q&A button, and we will email you and answer those questions.

But in wrapping up, if there's one message to take away from today's conversation, it is that the best revenue cycle operations are built on a simple idea. Put every part of the process to work on the p- on the problem it's best equipped to solve. Automation handles the repetitive work. AI surfaces patterns and opportunities that would otherwise be missed.

And experienced billing professionals apply the judgment and relationships that turn those insights into results. And when those pieces work together, practices don't just become more efficient, they maximize every payment they've earned, they cut unnecessary administrative work, and they build an operation that's ready to grow without adding complexity or risk.

So thank you so much to all of you. I love the engagement in the chat. Thank you to our panelists, Brittany, Elizabeth, Dana- Yeah ... Pedro. Appreciate your expertise so much. Um- Thank you ... and thank you. Yes. Thank you, guys. We really look forward to continuing conversations and to many more webinars for you to join us.

So until next time, thank you.

Brittany: Bye. Thank

Dana: you. Bye, guys.

‍