Purposeful AI helps safeguard revenue cycle performance, healthcare leaders say

The healthcare industry’s rapid pivot toward AI promises a critical way out of the current margin squeeze, offering tools that can automate redundant tasks, predict denials and accelerate reimbursement. But simply layering new software on top of outdated, fragmented processes is a recipe for high costs and low yields. True transformation requires a deliberate alignment of AI, automation and analytics with the human teams that power them.

At a recent HFMA executive roundtable sponsored by Inovalon, healthcare leaders gathered to discuss their real-world experiences, frustrations and successes in deploying these technologies. From the necessity of robust governance to the hard realities of proving ROI, these leaders carved out a practical road map for moving past the vendor hype to build a resilient, high-performing revenue cycle.

How are you viewing AI’s role in revenue cycle, and what does your adoption strategy look like? Are you relying on internal teams to build these capabilities, or are you leaning into your electronic health record (EHR) or solutions from other third-party vendors?

DENNIS JONES:  We are an Epic system, which for us means that ongoing maintenance and updates constantly collide with our financial goals. You can present a project with a massive dollar yield, but if it requires significant IT resources, it drops down the priority list.

FAISAL HUSSAIN: We use a three-tiered, risk-based prioritization framework. First and foremost, we look at compliance and audit risk: Will this project improve documentation accuracy and reduce our exposure? Second is financial value: What is the true ROI when you weigh reimbursement gains against implementation costs? Third is operational feasibility: Will it meaningfully improve daily efficiency?

The biggest mistake operational leaders make is working in a silo. To win executive backing, you must look at the big picture and ensure your revenue cycle initiatives directly support the health system’s broader five-year strategic plan.

DESMOND JACKSON: We operate in a rural community, which means we struggle for capital and live in an “employee desert.” Because we lack a local talent pipeline, we leverage technology to scale. Implementing autonomous coding and bringing AI into our CDI [clinical documentation improvement] workflows has bridged those critical staffing gaps. As an Epic shop, our initial hurdle was the belief that our EHR could do everything. But many native features aren’t mature enough yet, and waiting on them sets you back. We’ve learned that third-party partnerships are essential for immediate solutions, and we are helping our IT teams understand when it is necessary to look outside.

What role could technology (e.g., AI, automation, analytics) play in improving revenue yield in your organization? Where are you already seeing results?

BJ TAYLOR: My main challenge with AI is vendor noise, which occurs when vendors pitch redundant bots that mimic automation we’ve already built. When evaluating AI, my filter is simple: What physical touches does it actually eliminate, and is it solving a real gap rather than replacing something that already works?

Right now, we are parallel testing Epic’s native AI to generate appeal letters. While it reads records and drafts letters, our nurses still review them. The current bottleneck is the learning curve; it takes time for the AI to grasp the precise, payer-specific terminology carriers want to see. Finally, automating routine tasks — the ones staff actually enjoy — means we must upskill our workforce into critical decision-makers, which is a major operational challenge.

LEAH KLINKE: To easily adopt Epic’s native AI features as they launch, we’ve spent two years aligning our system closer to Epic workflows. Outside of native tools, we are prioritizing autonomous coding but face a highly fragmented vendor market. To identify the right partner, we are bringing our professional, outpatient and inpatient coding groups together to target our biggest opportunities — specifically focusing on offshore-outsourced tasks and internal compliance gaps. Once we analyze these volumes, we will issue an RFP [request for proposal] for a specialized partner. While we have historically relied on homegrown bots, this will be our first time partnering with an external vendor to implement true AI.

JONES: While payer-related challenges are constant, untimely claims, appeals and record submissions are ultimately on us. We already use standard worklists to flag claims within 14 days of a deadline, but those still require manual cleanup. The real opportunity is for AI to take that next step: automatically run the checks, resolve the bottlenecks and push the claims out the door. That “find the problem, fix it and go” capability is where the true value lies.

JACKSON: We leverage AI to augment our labor, not reduce it. With CDI, we can’t pull clinical nurses off the floor, so AI assists by reading charts and queuing up physician queries. While we have experimented with Epic’s native autonomous coding, we’ve found that specialized third-party solutions are far more mature, especially on the professional side. It scales our capacity without adding headcount, running 24/7 like a virtual coder that never sleeps.

HUSSAIN: With so much noise among AI vendors, we always look for the human in the loop. Frankly, any health system without an AI governance council right now is shooting itself in the foot — you need that structure for clear accountability. Labor shortages are a massive problem, which is why AI’s greatest promise is as a force multiplier that augments existing staff rather than replacing them. The best way to prevent denials is proactively through what I call CDI 2.0. Since no health system has 100% chart coverage by CDI nurses, AI can prioritize which charts actually need human intervention, reducing downstream denials before claims are ever submitted.

TIM TRAVIS: We are always trying to reduce friction and fragmentation within the revenue cycle. Bolting on a separate AI solution at every nuanced step is incredibly expensive, and those systems rarely talk to each other well. Our approach is to use tools that work out of the box without modification. When you start modifying AI technology in the revenue cycle, you break it. Bolting it on as-is is ideal. Only when a good external solution simply doesn’t exist do we have our internal IT team custom-develop a piece of software.

Where are the guardrails for leveraging AI technology in revenue cycle? Where you are trustful of applying AI to revenue cycle workflows? Where do you draw a hard line against leveraging AI?

MONICA HICKS: To support our team, we created a leadership volunteer group and introduced a Level 2 position, giving everyone a clear, incentivized path to earn more by upskilling. We initially tried building our own bots to manage costs, but proprietary EHR systems are notoriously difficult to integrate with outside tech. Even a bolt-on workflow routing system we purchased felt outdated because it just prioritized high-dollar claims first. Some of those big claims actually resolve themselves, while smaller ones get denied or hit timely filing deadlines. We need intelligent systems that tell representatives why they are working a claim — flagging a looming deadline or missing data — rather than just chasing the biggest numbers.

KLINKE: Before transitioning to technology, you must build a foundation by measuring exactly what your staff is doing. We are currently redesigning our workflow tracking to pinpoint our heaviest manual volume so we can target those specific tasks for automation. To manage this safely, we also created a revenue cycle sub-group within our AI governance council. We force ourselves to ask if AI is truly the right solution or if we can use standard, rules-based automation at a fraction of the cost instead of just chasing the latest shiny object.

MIKE LORENZ: A lot of what vendors sell as AI is just standard, rules-based automation. We built a basic bot that automatically handles 70% of our record denials from our largest commercial payer, slashing our turnaround time from 155 days to just 35. While it took three years to reassure staff we were only automating tedious work, we hit an interesting hurdle: As soon as the bot began getting 400 claims automatically paid a month, new denials from that same payer jumped by 120. It shows that payer algorithms are highly reactive.

For our next phase, we are targeting true AI — specifically, conversational AI to handle status calls on an exception-based model, kicking the call to a human only if the bot gets stuck. Ultimately, I want real-time AI prompting that tells our human reps exactly how to counter a payer’s pushback during a live call. We are already preparing our team for this by having them use Microsoft CoPilot: If a payer rep challenges a claim, our staff can type the objection into Copilot to generate immediate, strategic talking points. It gets them thinking like critical decision-makers.

Conclusion

Health systems are operating in an environment where escalating operational costs and reactive payer behavior make automated precision a survival requirement. While an expanding array of solutions continues to emerge, early adopters recognize that achieving meaningful ROI requires moving past basic, rules-based automation toward true AI integration. Crucially, this transition relies on strong governance that aligns new AI technologies directly with institutional bandwidth and long-term strategic plans.

Ultimately, successful AI deployment is not about eliminating the revenue cycle workforce but realigning and elevating it. By routing high-volume, repetitive tasks — like autonomous coding, chart reviews and deadline tracking — to intelligent virtual tools, organizations can safeguard their revenue integrity. Moving interventions upstream ensures cleaner claims before submittal, transforming human staff from transactional data processors into critical decision-makers.

PANELISTS

Monica Hicks
is vice president of revenue cycle at College Medical Center in Long Beach, Calif.

Faisal Hussain,
MD, MS (HIIM), RHIA, is clinical revenue cycle physician leader at City of Hope, Duarte, Calif.

Desmond Jackson,
DBA, MBA,  PMP, is vice president of revenue cycle at Monument Health in Rapid City, S.D.

Dennis Jones,
MSOL, is associate vice president of revenue cycle at Jefferson Health in Philadelphia.

Leah Klinke
is executive vice president of revenue operations at Hackensack Meridian Health in Edison, N.J.

Mike Lorenz,
CRCR, CHFP, is director of revenue cycle at Integris Health in Oklahoma City.

BJ Taylor,
CRCR, is director of revenue cycle at Wellstar Health System in Marietta, Ga.

Tim Travis, FHFMA,
MBA, FACHE, is president and CEO at South Bend Medical Foundation in South Bend, Ind.

About Inovalon

Inovalon is a leading provider of data and solutions empowering data-driven healthcare. We bring together national-scale connectivity, real-time primary source data access, and advanced analytics to enable improved clinical outcomes and economics across the healthcare ecosystem. For more information, visit www.inovalon.com.

_{This published piece is provided solely for informational purposes. HFMA does not endorse the published material or warrant or guarantee its accuracy. The statements and opinions by participants are those of the participants and not those of HFMA. References to commercial manufacturers, vendors, products, or services that may appear do not constitute endorsements by HFMA.}