AI in Healthcare Revenue Management: From Automation to Real Results in the RCM Cycle
Artificial Intelligence is changing the game in healthcare revenue management, and we’re not talking about minor tweaks.
For a long time, automation in the industry was already part of the daily routine for RCM teams, with auto-adjudication systems and standardized workflows that helped organize the day-to-day chaos. These tools had their value, of course, but they operated within well-defined boundaries: they followed rules, executed repetitive steps, and freed up teams to focus on tasks that required more human judgment. It was a real step forward, but still pretty linear.
What’s happening now is different. The new wave of Artificial Intelligence didn’t just show up to do the same things faster. It arrived to reason about data, anticipate problems, and change the way teams make decisions before they even act. This completely shifts the work dynamic within healthcare organizations, especially when it comes to Revenue Cycle Management — a process that has always been complex, time-consuming, and full of failure points.
That was exactly the topic three industry experts discussed during the BHB VALUE Conference, held in March 2026. On the panel, Becky Carlson, Head of RCM at Joyful Health, Emma Sugerman, Co-founder and COO of Mavida Health, and Arnav Simha, Lead PM at Nirvana, brought complementary perspectives — and sometimes refreshingly honest ones — about where AI is delivering real results and where expectations are still out of sync with reality. The conversation was rich, practical, and free of the excessive optimism that usually comes with tech discussions in healthcare.
What becomes obvious right away is that there’s a huge difference between automating tasks and using AI to think alongside your team. That distinction, seemingly simple, is what separates organizations that are seeing results from those still waiting for the magic to happen. 🎯
When AI Stops Being a Tool and Becomes a Partner
During the panel, one of the most striking points was precisely this conceptual shift: stop viewing Artificial Intelligence as a set of tools that execute commands and start treating it as an intelligence layer that processes context, identifies patterns, and suggests paths based on real data. Becky Carlson was direct when talking about the experience at Joyful Health, highlighting that the real differentiator of today’s AI lies in its ability to reason about data, not just process it. According to her, it’s now possible to predict whether a claim will be paid before even taking any action on it. This allows teams to adjust their priority order, approach, and even the actions they take in each case. It’s no longer about simply executing activities — it’s about changing the entire work framework.
Emma Sugerman, from Mavida Health’s perspective, brought an equally relevant take on how intelligent automation impacts smaller teams. In organizations with leaner resources, every wrong decision in the revenue cycle carries a high cost. She pointed out that current tools are not only smarter but also easier to use, which is critical because there’s still an essential human component in this work. Before the partnership with Nirvana, Mavida already had tools and information available, but that data wasn’t actionable or easy to interpret. With the same operational team, Mavida was able to accomplish significantly more after implementation, and that translated directly into margin improvement and operational efficiency.
Arnav Simha rounded out this view with a more technical angle, explaining that there are two important fronts of technological advancement. The first is that AI-focused companies like Nirvana can stay agile and adopt the latest technologies quickly. The second, often overlooked, is that EHR and practice management systems themselves have also evolved. Five years ago, these systems were extremely difficult to integrate. Even if a phenomenal AI product existed on the market, it was virtually impossible to connect it in a functional way to the provider’s environment. Now, he described the landscape as a rising tide that lifts all boats. The improvement in EHR technology allows companies like Nirvana to integrate AI agents in genuinely useful ways — both for collecting information more efficiently and for delivering it back to the right place at the right time.
AI Isn’t a Magic Wand: What Needs to Be in Place Before ROI
One of the most relevant themes of the panel was the discussion about where organizations are overestimating AI’s impact on the revenue cycle and what needs to be in place before it actually delivers a return on investment.
Receive the best innovation content in your email.
All the news, tips, trends, and resources you're looking for, delivered to your inbox.
By subscribing to the newsletter, you agree to receive communications from Método Viral. We are committed to always protecting and respecting your privacy.
Emma Sugerman was the first to raise the issue of change management. According to her, it doesn’t matter if you have the most advanced tool if your team resists using it. There’s often a natural resistance to new technologies, and that means you need to invest in training and in helping people understand how the tool works and why it matters. If that resistance persists, the expected impacts and improvements simply don’t materialize.
Becky Carlson was even more direct in addressing what might be the most common mistake in the industry. According to her, some people see AI as a magic wand that’s going to save RCM. The reality, though, is quite different: if your RCM processes are broken and you introduce AI, what happens is you’re going to create chaos faster. The essential thing is to fix the fundamental problems first. AI should be used as a workforce enabler — a force multiplier that increases the amount of work each team member can accomplish. But it still requires that processes are clean and RCM operations are solid before being introduced.
Simha reinforced this view with a practical phased approach. For him, the right path is to crawl, then walk, and only then run. In the initial phase, the organization should look at the volume of information it already processes — claims, checks, portal interactions — and identify the biggest problem areas. In healthcare, the information usually exists. The problem is that it takes too long to access or requires excessive manual work. So the ideal starting point is the processes that consume the most time, require phone calls, or produce unreliable information. That might represent only five to ten percent of total volume, but those are the processes that burden the team the most. If AI solves those problems reliably, then it’s possible to expand from there.
Claims Prioritization: Where AI Delivers the Most
If there’s one area within Revenue Cycle Management where Artificial Intelligence is generating measurable impact consistently, it’s claims prioritization. Becky Carlson explained that traditionally, RCM has always prioritized claims based on aging buckets: zero to 30 days, 31 to 60, 61 to 90, and so on. Everyone at conferences talks about this. The problem is that the likelihood of collection in each bucket is different, and the level of effort required varies enormously.
RCM produces massive amounts of data, but few organizations use that data as a feedback mechanism for future decisions. By leveraging AI, it’s possible to identify patterns within claims: which ones have a high probability of successful resolution and which ones, unfortunately, likely won’t be paid. This allows organizations to make faster decisions about where to focus their time. When you’re dealing with so many claims and so much data, that kind of prioritization becomes incredibly valuable.
Carlson also warned about the risks of maintaining prioritization exclusively by aging. If, for example, there’s a credentialing issue with a payer affecting the organization, the team can spend a lot of time navigating through the noise instead of solving the root problem. Once the credentialing issue is fixed, many of those claims can be reworked automatically. But when the focus is exclusively on aging, the team may end up working on easier claims first while more complex issues keep aging or fall through the cracks. Considering the payer mix and prioritizing those that represent the largest share of revenue can be a much more strategic approach.
Eligibility and Scale: Mavida Health’s Case with Nirvana
Emma Sugerman shared details about how implementing AI focused on eligibility at Mavida Health transformed the company’s operations. Before working with Nirvana, Mavida already had a tool, but it wasn’t getting the team where it needed to go. One of the biggest problems was that the information coming back from eligibility checks wasn’t digestible. It wasn’t actionable for the team or for Emma herself.
The result was that the team tried to translate information for patients, but the data wasn’t clear enough for anyone to act with confidence. Once smarter tools were implemented, the difference was immediate. The information became easier to understand, easier to communicate, and easier to use. That allows each team member to accomplish much more.
But not everything improved overnight. Sugerman was transparent about the persistent challenges. She brought up the concept that many in the industry call garbage in, garbage out. If the underlying information isn’t complete or standardized, the tools can only go so far. For example, it was possible to verify that a patient’s plan was valid and active, but answering bigger questions — like whether the care is in-network and what the patient’s financial responsibility will be — is still challenging. The worst experience for a patient is receiving care and then getting hit with a surprise bill. So there’s still a lot of work to be done across the entire industry to improve the accuracy and clarity of this information.
Smart Automation and Operational Improvement in Practice
It’s one thing to talk about automation in abstract terms. It’s another to understand how it translates into real routines within RCM teams. The panel made it clear that the organizations finding the most success are the ones that implemented AI incrementally — first mapping the most critical bottlenecks in their revenue cycle and then applying specific solutions to each pain point. This is different from adopting a generic platform and hoping it fixes everything at once. A surgical approach, combined with clear tracking metrics, is what transforms AI from an expensive experiment into a real lever for operational improvement.
Becky Carlson was especially emphatic in highlighting that well-applied automation frees RCM professionals for functions that require skills no AI can replace: payer negotiations, complex case resolution, relationship management, and decision-making in ambiguous situations. When repetitive, low-value tasks are absorbed by the system, the human team gains room to work on problems that truly need contextual judgment. This repositioning of teams is, in her view, one of the most underrated benefits of AI adoption in the revenue cycle.
Emma Sugerman added that this operational gain also has a direct impact on team satisfaction. RCM professionals who spend less time on mechanical tasks and more time on higher-impact activities tend to be more engaged with their work. In an industry with high turnover rates, that factor is far from secondary. The operational improvement driven by AI isn’t just a matter of financial efficiency. It touches on organizational culture, talent retention, and quality of service delivered — creating a positive cycle that goes well beyond the revenue cycle numbers. 🚀
When AI Underperforms: A Product Problem or an Environment Problem?
Arnav Simha addressed a question many organizations avoid discussing: when AI underperforms in the revenue cycle, is it a product problem or an environment problem? His answer was straightforward. It’s usually both.
He explained that AI tools should be treated the same way you’d train a junior employee. You test them, they make mistakes, and you provide feedback. Part of that feedback may involve improving how data is structured within the EHR or practice management system. Other times, the AI model itself needs refinement. By identifying failure cases and categorizing them, organizations can improve both the environment and the product until the system becomes a highly effective contributor.
This point is essential to the core concept of garbage in, garbage out, as Simha reinforced. The data feeding the AI needs to be coherent. AI won’t be very effective if it has to interpret information scattered across notes, fields, and documents in a chaotic way. Organizations need to normalize their data so AI systems can clearly understand the task they’re performing and the result they need to deliver. Ideally, AI vendors should help standardize this information — from the data in the systems to the results written back into them. The cleaner and more structured the data, the more effective the AI will be.
What to Ask Before Buying an AI Solution
All three experts also shared practical guidance for organizations evaluating AI solutions for their revenue cycles.
For Becky Carlson, the most important question is: what work is this AI solution actually going to do? And what metric will measure its success? Is the organization trying to improve eligibility verifications, denial management, or coding? Once that’s defined, you need a clear metric — like reducing eligibility-related denials, improving denial overturn rates, or ensuring documentation passes compliance review. Being intentional about the specific work and the success metric is critical.
Emma Sugerman brought a perspective focused on organizational impact. For her, the central question is: how will this empower my company and my team? She thinks about margin improvement, operational efficiency, and how to make the same resources go further. The tool doesn’t work on its own. It needs to amplify the team’s capabilities.
Arnav Simha suggested two questions. First: how does this help my organization scale? You need to understand where your current metrics stand and where they need to be to hit your goals. Second: what work is the tool actually solving? Every AI tool essentially performs a specific task — whether it’s making a phone call, reading a portal, or interpreting notes. You need to determine whether that task represents a significant pain point for your team.
The Future: Access, Simplicity, and Democratization
Looking ahead, all three experts shared visions that combine optimism with pragmatism.
Becky Carlson, after a decade in RCM watching teams spend enormous amounts of time just trying to identify problems, said she’s excited about the ability to analyze large volumes of data and identify patterns earlier. Instead of constantly reacting to problems, organizations can anticipate them and address them proactively.
Arnav Simha described the future Nirvana envisions as a scenario where healthcare transactions are as simple as buying a bottle of water at a store. You walk in knowing the price, pay with a reliable method, and receive the service. Today, bad or incomplete information makes that difficult. But AI can identify trends in claims data and uncover patterns — like recognizing when claims are consistently being redirected to a different payer. This allows organizations to make faster, more confident decisions and drastically reduce the time needed to resolve transactions.
Emma Sugerman brought the conversation back to access to care, which is Mavida Health’s core mission. The company provides deeply specialized care, and patients want to know that their treatment is in-network, financially accessible, and predictable. She also expressed excitement about the possibility of leveling the playing field for smaller providers. These tools shouldn’t only be accessible to very large organizations. If it’s possible to democratize access to these technologies, more providers will be able to accept insurance, which ultimately expands access to care for more people. 💡
What Still Isn’t Working the Way It Should
One of the most valuable moments of the panel was when the experts opened up space to talk about what still isn’t working. And here, honesty was the differentiator. Arnav Simha was direct in acknowledging that many Artificial Intelligence promises in the industry still run into data quality issues. AI systems are only as good as the data they consume, and a large share of healthcare organizations still operate with fragmented, inconsistent, or simply incomplete data structures. Training a model on bad data is a recipe for bad results, regardless of how sophisticated the technology underneath might be.
Another point raised was the misalignment between expectations and reality during the adoption process. Many organizations go into AI projects expecting immediate results and come out frustrated when the return takes longer than anticipated. This gap between market messaging and the practical implementation experience is real, and all three experts were unanimous in saying that transparency about the time needed to see results is essential for teams to stay committed to the process during the learning curve.
Finally, there was an important warning about the temptation to automate processes that are still broken. Automating an inefficient process doesn’t make it efficient. It just accelerates the problem. Before applying any Artificial Intelligence solution to the revenue cycle, it’s essential to understand where the structural flaws in the current process are. Technology amplifies what already exists — for better and for worse. And that, more than any specific feature, is the most important lesson the experts left for organizations that are still at the beginning of this journey.
