Going from agentic AI hype to revenue cycle reality
Revenue cycle management has set its sights on agentic AI to solve major pain points, but there are some implementation considerations before the tech can deliver on its promises.
Editor's note: This article is the third in a three-part series on agentic AI for revenue cycle management. The other articles cover agentic AI's role in revenue cycle management and top use cases.
The hype is building around agentic AI for revenue cycle management. Technology vendors and provider partners are quickly building a portfolio of use cases But, as it goes in healthcare, implementing new technologies takes time, patience and a lot of due diligence.
"Healthcare itself is a double-sided coin," explained Mark Sithi, senior vice president of product at R1 RCM. "We want to have groundbreaking technology, especially to better serve the patient, but at the same time, there's also healthy skepticism around the use of AI because it's a highly regulated industry and, by nature, we try to be minimally invasive to patients undergoing care."
So, how do healthcare organizations overcome implementation hurdles to make agentic AI a revenue cycle reality and realize the benefits the technology can bring? Vendors and providers must build trust in the technology, demystify misconceptions and address data challenges.
Trusting the technology
Healthcare is one of the most highly regulated industries in the world, as concerns about patient safety, clinical efficacy and data privacy are paramount.]. Building trust in AI in this landscape is a challenge, as AI has been known to hallucinate.
"Is AI going to hallucinate and suddenly bill Medicare for unnecessary procedures? I'm exaggerating, but those are the types of questions being asked," Sithi stated.
The risk of billing or coding a patient encounter incorrectly is too significant for healthcare organizations already operating on razor-thin margins. What's more, the government has made cracking down on improper payments a top healthcare priority.
"Just because there is an agentic AI tool doesn't mean it removes the False Claims Act," added Judson Ivy, founder and CEO of Ensemble Health Partners.
Data privacy is also top of mind for technology experts, particularly after the 2024 Change Healthcare data breach. Implementing any technology through a vendor brings a host of third-party risk concerns. Healthcare organizations need to ensure their vendor partners are safeguarding patient data, especially as AI agents tap into data across an organization and beyond to other healthcare providers.
"Our reputation is the most important thing to us, so we have to protect people's data and privacy," Ivy said. "We spend a lot of time de-identifying data. We spend a lot of time segmenting data. But still, it's a natural concern for people just because AI is new. However, data protection and data privacy are not new."
Working with the organization's IT team can help build trust in agentic AI implementation, as suggested by John Landy, chief technology officer of FinThrive.
"As you look at these solutions, get an IT security review in addition to ensuring it meets your business needs. That's the key thing we've seen," Landy explained. "We've seen people either try to solution something without making it a conjoined effort between the business and IT sides, but you'd be better off if you did a joint analysis on the solution because security is a big part of this and it's extremely difficult to invest the levels that you need to be able to provide the data without it."
Overcoming IT infrastructure challenges
Standing up the proper health IT infrastructure is also key to building trust with agentic AI, according to Dan Parsons, cofounder and chief experience officer at Thoughtful AI.
"We don't want you to just overlay technology on something that's fundamentally not optimized," Parsons said. "You want to reengineer the process from the get-go because there is a process that is more optimal for computers and AI to operate."
The old IT adage "garbage in, garbage out" also applies. If healthcare organizations don't have accurate and complete data feeding into their IT systems, those technologies will not provide high-quality output. Healthcare organizations must perform data clean-up to ensure agentic AI solutions draw on information that can generate the most accurate results or actions.
Having the right ontology or data platform allows a health system to see everything in a cohesive, connected dataset, which can be leveraged across all agents and all applications versus having everything in disparate places.
Mark Sithi, SVP of product, R1 RCM
There is a spectrum of data and infrastructure maturity, Parsons explained. Healthcare organizations -- whether internally or with the help of a third party -- should perform an analysis to understand what they need to support newer technologies.
Healthcare organizations should also be probing their vendor partners on their IT infrastructure, according to Landy.
"Vendors also have to make sure they're investing in the plumbing that allows you to really leverage AI," Landy explained. "So, for example, investing in the whole data platform, which some vendors are going to try to find a shortcut for."
The orchestration layer of agentic AI, which creates potential for more end-to-end revenue cycle management solutions, can really benefit from this whole data approach.
"Having the right ontology or data platform allows a health system to see everything in a cohesive, connected dataset, which can be leveraged across all agents and all applications versus having everything in disparate places," Sithi said. "That sounds obvious to a lot of people, but that's just not how healthcare works. Many health systems have applications and data that are not really tied together."
Will agentic AI replace humans?
It's true that AI is reshaping the workforce, and healthcare isn't immune. Healthcare organizations have expressed concerns about how AI could replace human workers, especially on the administrative side.
However, technology leaders vow this isn't true, even with more sophisticated AI agents. In revenue cycle, it is more about finding where agentic AI can bridge staffing gaps and complement the work teams are doing.
"We don't have any plans today for 100% agentic in any area that has compliance or regulatory risk. You can't use technology as the answer," Ivy stated. "We're very focused on how not to replace the human."
Humans should focus on the work that has the highest value-add for the organization, while AI agents can handle more of the daily, repetitive tasks that don't have as much of a regulatory and compliance impact.
"Then use human in the loop," Ivy emphasized.
The concept of having a human in the loop to provide feedback for AI agents is critical, technology experts agreed. Healthcare is human, and even the revenue cycle needs human oversight to ensure compliance and quality. Furthermore, some situations in healthcare just aren't meant for AI agents to handle completely.
"For our AI agents, that's a core part of their workflow: identifying exceptions, knowing what they can and cannot do, making the necessary documentation and then passing that work off to their human counterparts," Parsons stated.
As agentic AI matures, improving that human-in-the-loop experience will drive further adoption in revenue cycle management.
"Making that handshake between humans and their AI counterparts easier will be core to driving completeness and widespread adoption for these enterprise-use cases of agentic AI," Parsons explained.
Getting past the buzz
The revenue cycle agentic AI landscape is a bit of "a land grab," as Sithi put it. From established revenue cycle technology players to new entrants to healthcare, everyone is seemingly trying to get into this space. This buzz is going to drive adoption in 2026.
However, Landy warned that, at this point, many solutions may not be completely ready to deliver on the promises of agentic AI for revenue cycle management.
"There are a lot of vendors that have really good marketing around what they're doing, but when you go to analyze the solution, there's not a whole lot there," Landy said.
He encouraged healthcare organizations to conduct a vendor analysis to evaluate whether a vendor’s tool is really going to help the organization in a meaningful way.
Still, it seems like a matter of time before healthcare organizations use agentic AI in some capacity for revenue cycle management. Healthcare is moving from early adoption to "widespread curiosity," according to Parsons, which will drive adoption.
"Before folks might not even want to entertain it," he said. "Now every healthcare boardroom and advisory group in America is asking: What is our strategy with AI?"
Jacqueline LaPointe is a graduate of Brandeis University and King's College London. She has been writing about healthcare finance and revenue cycle management since 2016.
