From Great Gatsby to Great Gains: Cleveland Clinic’s GenAI Leap in the Mid-Cycle

If you worked as an inpatient coder, you’d spend your day reading The Great Gatsby — not once, but twice. Every hour. All day.

That’s because a single inpatient encounter generates, on average, 59 clinical documents and nearly 50,000 words — the equivalent of F. Scott Fitzgerald’s classic novel — and coders must interpret every detail, then select the correct diagnosis codes from more than 150,000 options. It’s a process that demands deep clinical knowledge, razor-sharp focus, and, increasingly, a workforce in short supply.

In a recent Becker’s Healthcare webinar on GenAI + RCM: How New Tech Is Solving Old Challenges, Bob Gross, executive director of financial decision support and analysis at Cleveland Clinic, and Benjamin Beadle-Ryby, SVP and co-founder at AKASA, explored how generative AI (GenAI) can “read” these Gatsby-length medical records in seconds, surface key insights, and deliver evidence-backed recommendations — turning a human bottleneck into a scalable, consistent process.

The AI doesn’t get tired. Monday or Friday, it’s consistent — and every recommendation comes with line-item evidence you can trust. That’s something we’ve never had before.

~ Bob Gross, Executive Director of Financial Decision Support and Analysis at Cleveland Clinic

Before diving into their real-world experiences, the speakers polled healthcare finance attendees on their own GenAI journeys and priorities:

• GenAI adoption in the revenue cycle: 21% not yet started, 30% exploring, 17% piloting, and 32% implementing
• Top use cases for GenAI: Denials (33%), prior authorization (27%), and coding/CDI (26%) topped the list

The split showed an industry still in the early innings — with half of respondents not yet past exploration — but already highlighting the same pain points where Cleveland Clinic has focused its early GenAI work.

Understanding AI vs. GenAI — and Why It Matters Now

The leap from traditional automation to GenAI is a bit like moving from skimming the Cliff Notes of The Great Gatsby to actually understanding the whole novel in context.

Artificial intelligence is the broad umbrella — machines that can reason, learn, and act.

GenAI, powered by large language models (LLMs), takes it further by interpreting and generating human language, which makes it ideal for working with the unstructured clinical data the healthcare industry runs on.

Want to know more about this technology? Here are 10 Things Healthcare Leaders Need To Know About LLMs and Generative AI.

Gross emphasized the need for trained models, noting, “You can’t just drop a general-purpose LLM into the revenue cycle and expect it to work. You need the right training, tuning, and curation for healthcare-specific workflows.”

So much of what impacts coding, reimbursement, and quality reporting is buried deep in the narrative. GenAI can pull it out, in context, and present it with an audit trail.

~ Benjamin Beadle-Ryby, SVP and Co-founder at AKASA

What this means for leaders: In revenue cycle work, the “novels” are medical records — long, complex, and filled with subtle cues that impact accuracy, revenue, and compliance. GenAI can parse them in seconds, interpret the context, surface the right details, and do so in a fully auditable way. That’s a leap in both speed and reliability over previous rules-based tools.

Why the Mid-Cycle Is Ripe for GenAI Innovation

When Cleveland Clinic evaluated where to begin, they looked for the place where Gatsby-sized records were creating the most operational strain. Inpatient coding and CDI were clear winners: high volume, rule-governed, heavily dependent on unstructured documentation, and chronically short-staffed.

Earlier this year, Cleveland Clinic partnered with AKASA to test whether fine-tuned LLMs could supplement mid-cycle staff. The objective: boost quality and revenue capture while reducing manual strain on the workforce.

As Gross explained when reflecting on Cleveland Clinic’s pilot results, “When we looked at where we could get the fastest, most measurable impact, the mid-cycle — and coding in particular — was at the top. The rules are clear, the data is rich, and the scale is huge.”

Today, coders wade through patient records, combing for clinical facts, interpreting them against coding guidelines, and making decisions that directly influence reimbursement and quality metrics. That level of cognitive load — multiplied across thousands of encounters — is exactly where GenAI’s speed, consistency, and contextual understanding shine.

What makes these models great is the curation, training, and tuning you do to make them purpose-built and suitable for a particular process.

~ Bob Gross, Executive Director of Financial Decision Support and Analysis at Cleveland Clinic

What this means for leaders: Start where AI’s strengths — scale, consistency, and contextual understanding — directly address your biggest bottlenecks. Processes that are both high-volume and rule-governed are the fastest path to visible ROI.

From Theory to Action: Applying GenAI to Real Workflows

Cleveland Clinic’s approach was to integrate GenAI into existing workflows, not replace them. The AI acts as a tireless second set of eyes, reviewing coded charts, flagging potential missed opportunities, and presenting recommendations with direct quotes from the patient record.

In the first 60 days of production with the AKASA mid-cycle Optimization Suite, Cleveland Clinic experienced:

• Higher ICD-10 capture and case mix index
• Improved severity of illness, risk of mortality, and HCC capture
• Roughly 50% acceptance rate for AI-flagged recommendations
• Full audit trails supporting every suggestion

The result? Humans still own the narrative, but GenAI helps ensure no critical plot point (or quality metric!) is overlooked.

The point is to supercharge the humans in the process. We want them focused on the clinical and cognitive aspects of their job, not the monotonous task of identifying codes over and over.

~ Bob Gross, Executive Director of Financial Decision Support and Analysis at Cleveland Clinic

Building Workforce Trust in AI

Trust was essential to making the AI a welcome partner. Gross stressed that trust isn’t automatic — it has to be built. Cleveland Clinic approached GenAI adoption as a cultural shift as much as a technical one.

“Education is the first step in demystifying and bringing down fear or anxiety,” Gross explained.

The team integrated GenAI education into leadership retreats and departmental meetings, showing exactly how it works, what it can and can’t do, and where the final decision still rests with the coder or CDS.

By revealing the AI’s evidence trail — like citing the exact “page and paragraph” from the patient’s story — staff could see its recommendations were grounded in the same facts they rely on, not in guesswork. That transparency transformed skepticism into curiosity, and eventually into adoption.

You have to bring people along in the journey. If they understand the why and see the proof, they’ll trust the technology.

~ Bob Gross, Executive Director of Financial Decision Support and Analysis at Cleveland Clinic

Gross also underscored the workforce challenge ahead: “Over 40% of our coding team is retirement eligible in the next few years. We had to think about how to keep operations running without burning out the staff we have.”

What this means for leaders: A successful GenAI rollout hinges as much on change management as on technical capability. Involve staff early, make the system’s reasoning transparent, and frame AI as a tool for empowerment.

Where To Start Your GenAI Journey

The team piloted GenAI in multiple revenue cycle areas, but the mid-cycle proved to be the clearest success. Why? Because The Great Gatsby doesn’t change based on who’s reading it — the rules are consistent.

Prior authorization and denials, by contrast, are more like a choose-your-own-adventure book: different endings depending on the payer, the patient, and the exact scenario.

If you want the biggest impact, start where you can affect 100% of encounters — that’s coding.

~ Benjamin Beadle-Ryby, SVP and Co-founder at AKASA

That consistency in coding made it the ideal proving ground, delivering measurable results and building trust that could then be carried over to more complex areas later.

Gross added that starting with a governed process like coding builds early wins that can be expanded to more variable areas later: “It’s easier to get buy-in and measure results when the rules are clear and the value is visible right away.”

Best Practices for Implementing GenAI in the Mid-Cycle

Gross and Beadle-Ryby emphasized that implementing GenAI in the revenue cycle isn’t about chasing a buzzword; it’s about thoughtful planning, disciplined execution, and aligning technology with operational realities. They outlined a roadmap for leaders ready to bring GenAI into their own workloads:

1. Don’t wait to explore

GenAI is already delivering measurable results in the mid-cycle. “The technology is mature enough now for you to feel confident trying and piloting,” Gross said. Waiting risks losing early-mover advantages in efficiency, revenue lift, and workforce enablement.

2. Decide buy vs. build — and be realistic

While some organizations are tempted to build their own LLM in-house, Gross was direct: “You’re not going to build and maintain this at scale on your own. This is a space where you want a partner to help you get it right.” The right partners bring not just technology, but hard-earned revenue cycle expertise.

3. Avoid rip-and-replace

Integration is key. Effective GenAI solutions layer onto existing workflows and EHRs without disrupting daily operations. This keeps adoption friction low and accelerates ROI.

4. Know what’s under the hood

Not all “AI” is GenAI — and not all GenAI is built for healthcare. Ask whether the model is trained specifically on mid-cycle workflows and healthcare documentation. Confirm the data sources, how accuracy is measured, and whether it can be fine-tuned for your organization’s documentation styles, payer mix, and coding rules

5. Start with a controlled rollout

Beadle-Ryby stressed the importance of piloting in a manageable scope first: “You don’t have to boil the ocean. You can validate the technology, prove value, and iron out change management before scaling.” This builds trust internally and creates early champions.

6. Educate beyond the revenue cycle

Stakeholder engagement shouldn’t stop with coding and CDI teams. Finance leaders, compliance officers, and even clinical leaders should understand how GenAI works, its benefits, and its guardrails. This broad buy-in strengthens adoption and aligns the initiative with organizational goals.

7. Demand transparency

Every recommendation should come with an auditable evidence trail, clear justification, and — ideally — a confidence level. As Beadle-Ryby put it: “When staff can see exactly where in the chart a suggestion came from, they stop thinking of it as a black box and start seeing it as a colleague.”

8. Measure beyond ROI

Financial lift is important, but leaders should also track quality metrics (i.e., SOI, ROM, POA, HCC capture), query rates, denial reductions, and staff productivity. At Cleveland Clinic, improved quality scores are just as strategically important as revenue capture.

9. Plan for the next step

Think of your GenAI adoption in phases:

• Review – AI analyzes coded charts and flags what’s missing
• Collaborate – AI works at the coder’s fingertips, augmenting human decisions
• Automate – Once trust and accuracy are proven, AI can handle more complex workflows autonomously

View GenAI as a superpower, not a replacement. The goal is to enable people to do more of the work that requires their clinical and cognitive judgment, and less of the monotonous, repetitive work.

~ Bob Gross, Executive Director of Financial Decision Support and Analysis at Cleveland Clinic

Turning Opportunity into Action

For most healthcare organizations, mid-cycle operations are filled with “Great Gatsby” charts — long, complex, and easy to misinterpret under pressure. Cleveland Clinic’s experience shows that when you give mid-cycle staff a high-speed, high-comprehension partner in the form of GenAI, you can improve both financial yield and clinical accuracy without overhauling your entire system.

Beadle-Ryby framed the broader vision, saying, “This isn’t just about doing the same work faster — it’s about fundamentally changing what people spend their time on.”

GenAI is no longer experimental — it’s operational. If your organization is still exploring or piloting, the message is simple: start with a place where the rules are clear, the stakes are high, and the cognitive load is crushing. That’s where GenAI will deliver its first — and most convincing — wins.

Watch the full webinar recording for deeper insights from Cleveland Clinic and AKASA. And if you’re ready to explore what GenAI could do in your mid-cycle, let’s chat.