From Hype to Impact: Lessons from Cleveland Clinic on Generative AI in the Revenue Cycle

At the 2025 HFMA Annual Conference in Denver, healthcare leaders gathered to hear a session that was anything but business as usual. It was also one of the most well-attended events during all of HFMA.

Titled “No More Business as Usual: How Generative AI is Real and Already Paying Off in RCM,” the session brought together Cleveland Clinic’s Bob Gross (Executive Director, Financial Decision Support and Analysis) and Nick Judd (Senior Director, Revenue Cycle Management – HIM at Cleveland Clinic) with session moderator and AKASA co-founder, Ben Beadle-Ryby, to explore the real-world impact of generative AI in the revenue cycle and mid-cycle operations.

For health systems facing razor-thin margins and growing workforce challenges, the message was clear: generative AI (GenAI) isn’t just a tool — it’s a transformative force already delivering results.

“It’s about time we think about flipping the tables a bit,” opened Beadle-Ryby. “Generative AI and the emergence of large language models position health systems to have better control over what happens within the four walls of the health system — to drive more accurate quality and reimbursement through the mid-cycle.”

For a deeper look at how this collaboration came together and what it means for the future of AI in healthcare, check out the full announcement of the Cleveland Clinic–AKASA partnership.

Beyond the Hype: Tackling the Hardest Problems First

One of the most surprising lessons from Cleveland Clinic’s journey was their decision to start with the hardest problem in the revenue cycle: complex inpatient hospital coding.

“We sort of took an atypical approach and flipped the script,” said Judd. “We set out to focus on the most complex inpatient hospital coding first…not only out of necessity, given our workforce, but to be truly transformative.”

That workforce necessity is real: 55% of Cleveland Clinic’s coders are retirement-eligible. Leaning into GenAI is both a strategic hedge and a path to optimization.

Unlocking the Medical Record: Why GenAI Is a Breakthrough

“The watershed moment for me professionally was once we could see ChatGPT in action,” said Gross. “I knew that there had to be an opportunity for us to wrangle the messy, unwieldy semi- and unstructured medical record for purposes of automation.”

Historically, traditional automation couldn’t handle the complexity of patient records. But that’s changed thanks to large language models (LLMs).

According to AKASA analysis, most inpatient cases include 59 clinical documents and nearly 50,000 words — the equivalent of The Great Gatsby. Coders must distill that data into accurate billing codes from a set of over 140,000 options. Interpreting test results, reading through notes, looking at continuity of care. And they’re expected to get through at least two cases per hour.

It’s incredibly challenging for people and — up to now — impossible for technology.

As Gross noted, GenAI gives us the ability to tackle that mountain of complexity and “to make the semi- and unstructured medical record finally accessible for the purposes of automation and streamlining.”

Real Results: Accuracy, Quality, and Financial Lift

Perhaps the most compelling evidence came from Cleveland Clinic’s recent rollout of AKASA’s GenAI medical coding solution in one of its hospitals.

“We’re already seeing a 40% improvement in our present on admission (POA) capture,” said Judd. “That’s going to impact all of the raters, rankers, and risk adjustment outcomes related to that, not just revenue.”

Accuracy has been the north star throughout their investment in GenAI.

“Accuracy in documentation and coding was our number one priority,” Judd added. “The precision with which the solution delivered results was extremely compelling.”

The early results of the partnership revealed not only financial upside for Cleveland Clinic, but also improvements beyond what existing tools and workflows could deliver — all without compromising on compliance or clinical integrity.

Want to dig deeper into why accuracy matters — and how comprehensive coding can impact reimbursement and quality? Check out this post on ensuring coding completeness.

Empowering the Workforce: Coders as Validators, Not Replaced

While many fear AI will replace healthcare workers, Cleveland Clinic proves that the opposite can be true — if GenAI is implemented thoughtfully.

“It’s our intent not to replace people, but to level up how we’re using them,” said Gross. “To be validators instead of rote, straightforward coders.”

The response from Cleveland Clinic’s mid-cycle team reflects a mix of emotions and growing optimism as they engage with GenAI firsthand.

“There’s skepticism, some fear, but also a heck of a lot of excitement,” said Judd. “They’re seeing results. They’re being heard. And they’re helping shape the future.”

The interface and workflow were designed with coders in mind — sleek, efficient, and responsive to user feedback. In fact, the model incorporates feedback from coders in real time, enabling fast iteration and increasing trust, which is a key element for long-term adoption.

The AKASA Advantage: Deep Industry + Deep Tech

A recurring theme throughout the conversation was the need for partners who understand not only AI, but also healthcare.

“We recognized pretty early that we needed someone who understood generative AI exceptionally well — and revenue cycle exceptionally well,” Gross emphasized. “That’s the secret combination to making this work.”

That’s where AKASA comes in.

By building fine-tuned models specifically for revenue cycle and mid-cycle applications, drawing from Cleveland Clinic data and years of coding literature, AKASA enables scalable, clinical-grade GenAI solutions.

“This is not something we could have built on our own,” Gross admitted. “It would take us a decade.”

What’s Next: Scaling Mid-Cycle Intelligence

Cleveland Clinic is already seeing potential to extend GenAI beyond coding.“It’s not a far-off stretch to imagine automating portions of denial appeals or interpreting medical necessity for prior authorizations,” Gross predicted.

But success will require thoughtful expansion and shared learning.

“It’s important that we all begin to really lean into this space,” Judd concluded. “There is still hype, but there’s also reality. We’ll all benefit from shared success.”

The Bottom Line for Mid-Cycle Transformation

If the goal of revenue cycle management is to ensure accurate, timely reimbursement that reflects the true complexity of care, then GenAI is no longer optional — it’s imperative.

As Cleveland Clinic’s example shows, with the right strategy and the right partner, it’s not only possible but already happening.

“Let’s accurately reflect the care being provided — and make sure we get credit for that,” said Beadle-Ryby. “If we have high-powered AI to make sure we’re doing that, we should absolutely use it.”