Investors once told startups that large language models would fully automate clinical coding within a year. That prediction missed the mark, but AI is still finding a real foothold in healthcare's stubborn revenue cycle, according to Lee Kupferman, Co-CEO of R1's innovation lab. Speaking at HFMA's annual conference in National Harbor, Maryland, Kupferman explained where the technology delivers and why fragmentation - not the AI itself - remains the biggest hurdle.
Where AI delivers the most value
AI works best when it tackles straightforward, high-volume cases that don't require human judgment. A simple inpatient encounter for a known procedure with no complications is a prime example. In these situations, Kupferman said, "50 coders would all reach the exact same answer," making it an ideal task for AI to handle on its own.
That frees experienced coders to focus on complex encounters with longer documentation and varied payer rules - the gray areas where AI models still struggle. The goal, Kupferman believes, is routing the right work to machines and reserving human effort for cases that genuinely need expertise. Medical billers who want to sharpen their skills for this shifting workflow can turn to resources like the AI Learning Path for Medical Billers.
"You can get value out of [AI] tools in all of the revenue cycle, provided you have the right guardrails and you're honest about where it works well and where it's still got a way to go," Kupferman said.
The fragmentation that holds AI back
Part of what makes the revenue cycle so resistant to quick fixes is the deeply fragmented healthcare payment system. Hundreds of vendors sell narrow point solutions that don't communicate with each other, Kupferman pointed out. Coding teams often operate in near-total isolation from prior authorization staff, which means a denial that could have been caught upfront triggers weeks of rework downstream instead.
Kupferman sees this lack of connection as the single biggest barrier to AI delivering on its efficiency promises. The tools need to talk to each other for the gains to be realized, he said, yet most health systems still run with disconnected processes that undercut any single technology's impact.
A shift toward collaboration
The environment may be changing. Historically, both payers and providers have dismissed the idea of a more collaborative, AI-driven revenue cycle. Now, Kupferman noted, they're showing real openness to working together on fixing the payment process. The recognition that the current system doesn't serve anyone is shared across the board.
"Everybody is in violent agreement about what the problem is - they're just trying to figure out the best way to solve it," he said.
Why this matters for healthcare revenue cycle professionals
AI isn't replacing coders wholesale, but it is reshaping which cases they touch. The professionals who thrive will be those who understand which tasks can be safely automated and which require a closer, human look - and who push for better integration between the tools their organizations already use. The willingness of payers and providers to collaborate, even in early stages, suggests the long-static payment infrastructure may finally begin to catch up with the technology.
Your membership also unlocks:
