Every health system wants clinical AI. Far fewer want to talk about how to pay for it.
The economics are hardest where the need is greatest. Safety-net and public systems run on thin margins and an unforgiving payor mix. Clinical AI, the kind that touches diagnosis and treatment, costs a lot to build, is slow to validate, and can do real harm when it's wrong. Pitch it to a board on faith and you'll wait years for a yes.
The usual answer is to pitch AI as a cost-cutter. That's the wrong frame. The point of AI in healthcare isn't to cut cost. It's to expand capacity. Staff, beds, and clinic hours are finite; demand is not. The question that matters is how many more patients we can serve with what we already have.
Conviction doesn't fund a platform. Revenue does. So start in the revenue cycle, and let the back office pay for the bedside.
Start in the revenue cycle
Denials, charge capture, and payment variance are where AI proves itself fastest. Results show up in dollars within months. The stakes are lower than in anything that touches treatment, so you can iterate quickly. And the data already exists: claims, remittances, clinical documentation. Done well, these early wins fund the data platform every later use case needs. They also earn credibility with the clinicians and executives who decide whether clinical AI ever ships.
I've seen this up close in utilization management. Deciding whether an inpatient admission is medically necessary used to take a clinician hours of reading. An AI tool that reads the notes, labs, and encounter data as they arrive and checks the case against clinical criteria turns those hours into minutes. It pulls out the vitals, results, and presenting symptoms that matter, and suggests diagnosis codes with the evidence for each. Reviewers confirm instead of dig. The tool also catches supportable codes that human review missed. Physicians, skeptical at first, noticed and came to value it. The result is fewer denials, protected reimbursement, and faster patient flow. One rule doesn't bend: a human makes the call; the machine does the reading.
The data layer underneath
None of that works without the data underneath. Three lessons have held everywhere I've applied them.
Work backwards from a use case. "Get the data ready for AI" has no finish line. "Get these four feeds ready for utilization review" is a quarter's work.
Build a canonical model early. Model patients, encounters, claims, and providers as consistent objects with clear lineage rules. Every use case after the first gets cheaper.
Don't wait for modernization. Plenty of the feeds that matter still arrive as HL7 messages and CSVs over SFTP. That's fine. Waiting for a pristine platform is how AI programs die in year one.
Find a champion
People matter as much as data. The best predictor of success I've seen is a champion inside the domain: someone with real expertise and the patience to sit through one build cycle after another. Some clinicians work like startup founders; find them. Then run the early work like an internal consulting group. Keep it small and fast, and test the idea with the people who will use it before you formalize anything or buy infrastructure.
The most common mistake is the reverse: building the thing with no rollout plan, adoption target, or monitoring. You can't ship an AI model and forget it. It drifts as the world changes around it. And a model nobody uses has done nothing, however well it tested. Adoption is the metric that matters.
Equity is an accuracy requirement
Equity belongs on the first-wins list too. Language models perform worse in languages other than English, and models trained on unrepresentative data lose accuracy for the patients who can least afford it. If your patients speak sixty languages, a tool that only works well in English doesn't work.
What the money is for
The revenue cycle is the engine. Every recovered dollar and returned hour goes toward the real goal: more patients seen, sooner, with the staff and space you already have. Cost-cutting shrinks a health system to fit its budget. Capacity-building grows what the budget can do.
Start where the money leaks. Spend what you recover at the bedside.