Sanbir Kapoor is the CEO of Unify Healthcare Services.
getty
It’s a situation many patients have encountered while navigating the healthcare system: They go to the doctor and get care, then the doctor submits a claim to insurance. The insurer denies the claim due to insufficient information. The result is a drawn-out process, filled with back-and-forth between the various parties, to sort out the documentation issue.
This is one example of how administrative inefficiencies in healthcare can disrupt revenue cycle management (RCM), a process that providers use to bill insurers and collect payments and that AI can streamline.
Administrative inefficiencies can have significant financial implications for health care organization. Consider this finding from the Centers for Medicare & Medicaid Services (CMS): “The Medicare Fee-for-Service (FFS) estimated improper payment rate was 6.55%, or $28.83 billion” in the 2025 fiscal year.
Three Key Use Cases For AI In RCM
As the CEO of a company that offers AI-powered RCM software, I’ve observed three key use cases for AI in RCM.
The first key use case relates to the example I mentioned earlier: document review. AI can help healthcare leaders review complex documents, extract relevant information and flag missing or inconsistent details. For example, say an important piece of information within a document, or even a document itself, is missing. AI can flag the issue for a healthcare team member to review. In turn, healthcare organizations can save time and reduce the delays and frustration that documentation issues can cause for patients and providers. They can also identify missing or incomplete documentation before it leads to payment issues. According to the CMS report, “2024 Medicare Fee-for-Service Supplemental Improper Payment Data,” insufficient documentation accounted for 59.8% of overall improper payments.
Eligibility verification is the second key use case. AI can check whether a patient’s insurance coverage is valid for a given procedure or medication, and flag potential ineligibility and other issues (such as a missing date of birth) before a claim is submitted.
The third key use case is denial management. AI can help healthcare teams anticipate potential problems and address the root causes of those problems before more claims are affected. AI can also help healthcare teams prioritize follow-ups or existing denials. For example, an AI solution might discover that a certain percentage of claims are denied due to missing certain information. With that knowledge, healthcare teams can then adjust their documentation processes to capture that information before submitting future claims.
The Future Of AI In RCM
AI can automate individual tasks, but I don’t see that as the only role it will play in the future of RCM. AI also stands to create revenue cycle operations that are more connected, predictive and proactive.
For example, organizations can implement AI-driven workflows across their entire revenue cycle. These systems can use predictive analytics to identify reimbursement patterns that could cause revenue leakage and alert the team, allowing stakeholders to address potential issues before they result in significant financial losses.
The Risks Of Using AI In RCM
While AI can benefit RCM, it also has risks. In my view, there are four key risks healthcare teams should be aware of when it comes to using AI in RCM.
For one, there’s the risk of inaccuracy—misinterpreted and incorrect information. When sifting through complex paperwork, for example, an AI tool might miss crucial context and suggest an incorrect billing code. That’s why there should be human intervention to train AI and review its suggestions.
Another significant risk is patient data privacy becoming compromised. Because sensitive information is at stake, healthcare organizations should do their due diligence to carefully assess vendors and understand how patient data is stored, used and protected. They should also have action plans in place for how any potential data privacy incidents would be handled.
Healthcare teams should also be wary of overrelying on automation. Accepting AI’s suggestions without question can have harmful ramifications for organizations and their patients. Again, having a human-in-the-loop approach to AI, retaining human judgment, is important. Leaders should also instill clear accountability for all team members who use AI.
Making decisions without the full context is another risk that AI poses. An AI solution may identify a trend, but leaders still need to understand the full context behind that trend before making any major business decisions. For example, an AI tool might find that 45% of a healthcare provider’s patients are over a certain age. However, that figure alone may not tell the whole story. Leaders may need to consider whether the data reflects the provider’s overall patient population or only a specific department or group of patients. It’s ultimately up to leaders to do their own research.
Implementation Best Practices
AI can create great value in revenue cycle management when healthcare organizations use it to solve clearly defined operational problems rather than adopting it for its own sake.
Before implementing AI, leaders should carefully consider the operational problems they want to solve and how AI can address them compared with other tools. AI is not always the best technology for every problem. For example, a healthcare organization might need to revamp its patient scheduling system, which traditional software automation could address more effectively. However, leaders should keep in mind that AI can’t compensate for unclear workflows, unreliable data or weak operational controls. That’s why leaders should fix any underlying problems first before trying to automate processes with AI.
From there, leaders should thoroughly evaluate vendors and select the one that’s the best fit for their organization. Leaders should ask vendors questions about implementation (such as how the technology would integrate into existing systems), account ownership and data privacy practices, to name a few things. Leaders should also research pricing and be aware of potential usage-based fees, which can accumulate.
Once they’ve chosen a vendor, leaders should run a focused pilot. If they pinpoint three operational problems, I recommend centering the pilot on just one. Leaders should measure the AI solution’s performance and evaluate its benefits and drawbacks. Leaders should only scale a pilot when the results justify it.
Leaders should also ensure that all decision-makers understand the healthcare business itself, not just the technology. For example, a billing specialist needs to know the reason why an AI tool flags a claim for denial rather than relying solely on the tool’s recommendation.
My approach to AI adoption prioritizes measured implementation over immediate scale. I recommend that healthcare leaders begin AI initiatives with a clearly defined operational challenge. Then, they should introduce the technology through a focused pilot, and evaluate both its efficiencies and limitations. Only when the results demonstrate consistent value should leaders expand implementation.
Ultimately, the objective should not be automation alone, but greater efficiency with accuracy, accountability and patient trust built into the process. Healthcare administration is complex. AI can simplify it—but only if leaders approach it thoughtfully.
Forbes Business Council is the foremost growth and networking organization for business owners and leaders. Do I qualify?











