Artificial intelligence is already a staple for revenue cycle leaders, but emerging capabilities through agentic AI could bring automation to a whole new level: autonomous revenue cycle.
Autonomous revenue cycle brings together AI, machine learning and robotic process automation (RPA) to handle end-to-end processes with very limited human intervention. More vendors are touting autonomous capabilities, with some going as far as to signal autonomous end-to-end revenue cycle management.
Unlike traditional automation, which follows fixed rules for repetitive tasks, agentic AI can make complex decisions and orchestrate across AI agents to provide more cohesive, seamless automation.
However, healthcare is a dynamic, heavily regulated and complex industry, which begs the question: Can AI truly take over some tasks in healthcare, let alone most in the revenue cycle?
There is a lot of hype in the revenue cycle management market right now, meaning healthcare finance and revenue cycle leaders need to separate fact from fiction to prepare for this future state of revenue cycle management.
What does an autonomous revenue cycle mean?
Waystar, FinThrive, R1 RCM, Oracle Health and Epic: These are just some of the health IT companies touting agentic AI capabilities for revenue cycle management. Many vendors have already released these features as part of their products, while others are developing AI agents to support labor-intensive, rules-based tasks for short-staffed revenue cycles.
For example, there is a massive shift from computer-assisted coding to autonomous coding in which AI agents deliver "no-touch" or "zero-touch" coding. These agents can translate clinical notes directly into the codes needed for medical billing for certain cases, taking the human out of the process and, with it, human-related coding errors that lead to costly denials.
Medical coding has been a prime use case for autonomous revenue cycle capabilities, but health IT companies are rapidly expanding where AI agents can be used within the revenue cycle to support overstretched teams, eliminate bottlenecks and enhance efficiency. Using agentic AI, these companies are bringing autonomy to prior authorizations, insurance eligibility, denials management and A/R follow-up, among others.
Companies like Waystar envision an autonomous revenue cycle as a whole. Waystar recently announced it is building an autonomous revenue cycle using an end-to-end agentic network that acts within workflows, executes defined tasks and learns from outcome, all with minimal human intervention.
However, Chris Murray, managing director at Huron Consulting and leader of revenue cycle AI and automation at the firm, thinks healthcare leaders need to start more simply by defining autonomous as it works within the context of revenue cycle.
"This has always been an issue with AI," Murray said. "We always try to help our clients understand AI. There's RPA. There's ML. There's NLP. There are all these different flavors of AI, and agentic is a component of that. I think it's the same thing with a word like autonomy because, at least from my perspective, there will always be humans in the revenue cycle."
At its maturity, an autonomous revenue cycle may look more like AI and revenue cycle staff working hand–in–hand versus a fully automated workshop. After all, the revenue cycle isn't a closed system; staff must manage many changing factors, some of which they cannot control.
Challenges of a self-governing revenue cycle
The challenge with a self-governing revenue cycle is that providers are not the only stakeholders, according to Prashant Karamchandani, senior partner in financial transformation and co-leader of the revenue cycle transformation practice at Chartis.
You have payers that are changing rules and regulations. You have the data inputs from clinicians or other services you might be launching or doing. It's not a stagnant environment.
Prashant Karamchandani, senior partner, Chartis
"You have payers that are changing rules and regulations. You have the data inputs from clinicians or other services you might be launching or doing. It's not a stagnant environment," he explained. "To the extent those variables are going to exist or are going to have that level of complexity attached to them, it's harder to get to that fully autonomous level. You're always going to have to be relooking at things or redoing things along the way."
Compliance and data security in healthcare are also obstacles to full autonomy. Revenue cycles exist in a highly regulated industry that is vulnerable to data breaches because of the nature of the information healthcare organizations manage.
This means an autonomous revenue cycle may be automating 80% of tasks, explained Huron's Senior Director Clay Pemberton, a revenue cycle advisor at the firm. The remaining 20% requires human input due to compliance and security factors, as well as other operational considerations.
"You have to have a thoughtful approach to go job by job and say, should this actually be automated? Is it worth the risk? And then, which piece of technology should we use? Is this an RPA? Is this AI?" advised Pemberton.
The limits on revenue cycle automation make it difficult for end-to-end autonomy, but not impossible.
"We will see an end-to-end system, but I think when you actually look at technology itself, it's still going to be very specialized," Murray explained.
Currently, the highly specialized knowledge in different areas of the revenue cycle tends to exist in silos. For example, front-end revenue cycle manages patient access and eligibility verifications, while the backend handles denials and A/R follow-up. These bookends -- and all the areas in between -- don't necessarily talk to each other, especially without interoperable systems.
However, Murray sees agentic AI breaking down these silos to allow for visibility across the revenue cycle. So, AI agents may be very focused on a task at hand with bounded autonomy. For instance, an agent for eligibility discovery may focus specifically on a single payer or health plan. The scope of the agent is small, but an orchestration platform connects all the agents.
"They are working within one orchestration platform that is providing one source of truth and doing that programmatically so they can speak the same language at thousands of words a second, which I think is where we'll start to see end-to-end unlock and really drive significant value," he said.
Autonomous revenue cycle on the horizon
So, when can revenue cycle leaders expect autonomy? The answer may be a lot closer than they think.
AI optimists like Abhinav Shashank, CEO of Innovaccer, believe the autonomous revenue cycle is up to three years away.
A recent report from the health IT company defined levels of AI maturity, with the final level being autonomous healthcare administration in which AI manages high-volume administrative work with human oversight. Currently, the healthcare industry is between Levels 1 and 2, according to Shashank.
Level 1 involves AI tool experimentation, with pilots emerging in single teams and department-specific purchases, while Level 2 is when AI achieves functional adoption, with multiple use cases within functions like revenue cycle.
The report backs this stance with 70% of healthcare leaders reporting early to mid-stage AI maturity. However, 2026 will be a fork in the road, the report stated. Respondents signaled a shift from less experimentation to more embedded strategy for AI with a platform versus point solution approach.
Healthcare organizations that double down on platform AI for a unified structure will advance to Level 3 (enterprise integration) and Level 4 (systemic automation) much more quickly than those that rely on point solutions without a unified data layer.
"The timeline to even Level 4 is not decades away now, it's maybe two or three years," Shashank stated.
Karamchandani also predicts a shorter adoption path for agentic AI compared to AI implementation's past.
"There were companies in the past that said, 'We can do it today,' but that was 2018, and we weren't anywhere near ready for that. Now, our provider clients are hyper-interested in this and want to make investments and do it," he said.
Autonomous capabilities will emerge in the next three to five years, Karamchandani added, but true autonomy may take over a decade as organizations prepare for the shift.
It is a rubber-meets-the-road type of moment, according to Murray.
Operationalizing agentic AI for autonomous revenue cycle management is a lot more challenging than press releases make it out to be. Not only do AI agents need the capabilities to handle a dynamic environment, but healthcare organizations need a robust AI strategy to orchestrate agents across the revenue cycle.
Preparing for the future of revenue cycle management
Experts agreed that the autonomous revenue cycle is a long-term goal for vendors and providers. But just when this future state of revenue cycle management will be here depends on a thoughtful AI strategy.
You have to understand how to unlock the potential with automation and agentic tools.
Michael Duke, partner, Guidehouse
"You have to understand how to unlock the potential with automation and agentic tools," explained Michael Duke, partner and commercial healthcare lead for automation and innovation solutions at Guidehouse. "But in order to do that, you really have to understand data -- where it comes from and what it can do."
Data is at the heart of any AI tool, but it is also AI's biggest hurdle to ROI. A robust and connected data infrastructure is critical to the success of autonomous revenue cycle tools, especially if organizations want to reach an end-to-end solution.
"The institutions that will succeed in this entire process are going to have the best data and information infrastructures up and running," stated Shashank.
Healthcare organizations will need to map out data flows within workflows to identify bottlenecks and inefficiencies. Does a payer require PDFs instead of electronic exchanges? A problem as minor as that could make automating a workflow difficult, requiring staff to hold on to more menial tasks.
"Your constraint is not going to be the model or the hardware; your constraint is going to be the ability to manage your own information," Shashank said.
Detailed AI governance will also be a differentiator between successful AI users at scale and those relying on point solutions without a comprehensive strategy.
AI governance is not just an IT responsibility. Everyone needs to have a seat at the table, including revenue cycle representatives, to develop a comprehensive AI strategy that unlocks value at scale. This will help unify an organization's AI roadmap, aligning organizational goals for automation and data infrastructure.
However, Duke warned healthcare leaders that the roadmap may not be easy to carve.
"A lot of our clients try to just pave a goat path so they can move on it faster as opposed to straightening it out and shortening the entire distance," he said.
Healthcare organizations must reimagine workflows to enable AI, not just how agentic AI can generate efficiencies or results. Duke said leaders cannot assume the process they are considering for deploying AI will be the same process after implementation. This prevents missteps in health AI's past in which organizations simply automated existing, inefficient processes.
"If we're going to be successful, it's looking through a different lens, not trying to just replicate what we've been doing with some cool tools," Duke stated.
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.
