The Problem Isn't That Hospitals Are Moving Too Slow. It's That They Think Slow Is Fine.

One year ago today, I started writing what would become RCM 2030: Strategy and Survival for Revenue Cycle Leaders. I had a thesis, a timeline, and a nagging feeling that I was going to spend a lot of time being told I was being dramatic.

I was not being dramatic.

This week, in five days, the federal government deployed an AI audit tool and aimed it at every Medicaid program in the country. Four of the largest technology companies on earth reached enterprise scale inside American hospitals simultaneously. The FDA loosened clinical AI oversight without public comment. An AI system got FDA clearance to detect sepsis up to 48 hours before a clinician would think to look for it. A ransomware ring targeting healthcare got taken down. And an AI oversight executive order got canceled hours before it was supposed to be signed.

I wrote the book on exactly this moment. Literally.

So I want to talk about the thing nobody is saying out loud, which is not "healthcare is changing fast." Everybody knows that. The thing nobody is saying is this: a lot of hospital and revenue cycle leaders are watching all of it happen and doing absolutely nothing, and they have convinced themselves that is a reasonable position.

It is not.

There are three flavors of AI laziness in healthcare right now, and you probably recognize at least one of them.

The first one is the pilot that never ends.

You know this one. The organization ran a proof of concept eighteen months ago. The results were promising. There was a presentation to the leadership team. Everybody nodded. And then it went into a folder somewhere and the vendor is still sending quarterly check-in emails that nobody reads. This is not caution. This is avoidance dressed up in process language.

The second one is the checkbox deployment.

The tool got purchased. The contract got signed. There was an announcement. And now it sits inside the organization doing something nobody can quite articulate, measured by metrics nobody looks at, governed by a committee that meets quarterly and produces a slide deck. The technology is real. The strategy is not.

The third one is the "our vendor handles it" abdication.

This is the one that keeps me up at night, honestly. The assumption that because your clearinghouse or your EHR vendor has baked some AI into their product, you have an AI strategy. You do not. You have a feature. There is a meaningful difference between those two things, and the gap between them is where your denial rate lives.

Here is what 2030 actually looks like if you stay in one of those three lanes.

Your payer will be adjudicating claims in real time via API. The interoperability mandates embedded in the One Big Beautiful Bill are building those rails right now. When they go live, probably around 2027 or 2028, the organizations that are not ready for real-time adjudication workflows will face structural delays in cash receipt that their competitors will not. That is not a competitive disadvantage. That is a cash flow crisis that compounds quarterly.

Your denial rate, if it is sitting in the industry average range of 5 to 15 percent today, will still be sitting there in 2030. Meanwhile, the hospitals that moved will be operating below 2 percent. The gap between those two numbers, measured against your net patient revenue, is not a rounding error. For a $500 million organization, that is tens of millions of dollars sitting in preventable denials while you wait for someone to hand you a roadmap.

The government's AI is already auditing your Medicaid compliance. AERO, which HHS launched this week using ChatGPT, is scanning five years of audit history across all 50 states looking for deficiencies that were flagged and ignored. Hundreds of grantees have not filed required audits in more than two years. The letters have gone to every governor and state treasurer in the country. If your hospital depends on Medicaid revenue and your state is scrambling to respond, you are downstream from that scramble whether you were paying attention or not.

The second wave of AI is not ambient documentation.

Ambient documentation won. That is done. KLAS Research said it plainly: ambient speech AI is nothing like any healthcare technology since the EHR. Physicians are demanding it. UToledo Health cut chart closure time by 29 percent in eight weeks. The first wave landed.

The second wave is what happens after the note gets written.

Scheduling. Coding accuracy. Discharge instructions. Patient outreach. Denial prevention before the claim ever goes out the door. Every single one of those things is a revenue cycle function. And the organizations that are asking "what do we do with the capacity AI gives us back?" are the ones building the second wave right now.

AdventHealth reported an 80 percent reduction in administrative work this week. Their Chief AI Officer said something worth writing down: "If you take a 10-minute task and make it two, and that happens a thousand times a week, that's real capacity. The question is how you reinvest that capacity."

That question is the whole game.

The organizations that have an answer to it are building a moat. The ones still debating whether to extend the pilot are going to look up in 18 months and wonder how everyone else got so far ahead so fast.

One thing worth doing this week.

Not a framework. Not a committee. One thing.

Find out which AI tools your organization has purchased in the last 18 months and ask someone, out loud, in a meeting, what the outcome data looks like. Not what is planned. Not what is in the pilot. What the actual measured result is in your patient population, in your workflows, with your data.

If nobody can answer that question, you now know exactly where to start.

The hospitals that thrive through this period will not be the ones with the most technology. They will be the ones whose leaders stopped waiting for permission to move and started asking the right questions before everyone else did.

That moment is now. This week proved it.