At the American Hospital Association Leadership Summit in Colorado (July 12–14), Dr. David Kirk, CMO of Regard, joined Dr. Matt Werpy, Medical Director of Hospital Medicine at Monument Health, to walk through how a five-hospital system in western South Dakota moved from chasing documentation to fixing it at the source, and what that shift changed for physicians and patients.

Key takeaways

1. The documentation problem at Monument wasn't really about documentation. Physicians could only see a fraction of the chart, leading to missed diagnoses and lost revenue.
2. Physician-led decisions drove adoption, not a mandate from leadership. Monument's hospitalists chose the platform and voted to adopt it themselves.
3. Buy-in isn't won at the vote, it's won in daily use. Monument's note template wasn't popular at launch, and only won physicians over once they felt the time it saved.
4. Better diagnosis capture doesn't just raise revenue, it changes quality scores too. Severity of illness and mortality risk scores rose right alongside CC/MCC capture.
5. Once physicians trust a diagnostic tool on rounds, they start asking for it everywhere: overnight cross-coverage, outpatient encounters, and screening for conditions that usually only get caught after they become a penalty

Same problem, different hospital 

Dr. Kirk opened with a memory from earlier in his career: walking into a room of cardiologists trained at Duke and Cleveland Clinic to tell them their outcomes were below average. The care wasn't the issue, and Kirk's team had already exhausted the standard fixes, like revenue cycle pushing for more CCs and MCCs, Epic's own documentation packages, and retrospective queries. None of them moved the quality scores.

Dr. Werpy described a similar pattern taking shape at Monument Health, a five-hospital system serving a largely rural stretch of western South Dakota with about 4,500 physicians and caregivers. There was revenue leakage from downcoding, documentation lacking specificity, a rising volume of CDI queries, and DRG upgrades getting declined for insufficient support. Interestingly, there wasn’t a problem with the documentation, but with visibility. The root cause was visibility, not documentation. Physicians have time to review only about 3% of the data in a typical chart at the point of care. The rest goes unseen while they're focused on the patient in front of them. 

Chosen by the people who use it

Closing that gap meant finding a tool built to actually do that reviewing, not just document faster. Monument evaluated a handful of AI platforms before choosing Regard, and the decision came from the hospitalist group itself. It went to a vote at a hospital medicine meeting, and physicians overwhelmingly supported it. Even so, not everything landed immediately. Werpy was candid that Monument's standardized note template wasn't popular at first, since physicians take real pride in how they write. It won them over once they saw the value in practice, especially when handing patients off between hospitalists, where a consistent structure meant nothing got lost in translation.

That structure still left room for individual style. Within it, Regard let each physician document however they wanted, whether that meant bullet points, verbose notes, or something concise. Werpy said that flexibility is what got several of his most hesitant physicians to finally try it. This year, Monument went live with Max, Regard's newest AI agent, and physician adoption climbed from the high 60s to 90%.

The judgment stays with the physician

Werpy pushed back on the framing most vendors use. "I think artificial intelligence is probably not the best language to use," he said. "I'd say augmented." Kirk explained what that looks like in practice: Regard will flag a diagnosis like hyponatremia even when the physician hasn't documented it, based on labs and risk factors already in the chart. But the physician can always take it out. "Regard has no problem putting that in," he said. "Certainly I can erase it." The diagnosis is a suggestion, not a decision. Regard's job is to make sure it's on the table in the first place.

Part of what makes that workable is consistency. Regard's diagnostic algorithms run the same logic on every chart, so the same patient run through it twice returns the same result both times. That consistency, more than any single accuracy figure, is what let Monument's compliance and clinical teams sign off on the rollout without worrying about hallucinated diagnoses.

The numbers

Eight months in, Monument's two hospitals saw average Elixhauser scores per encounter improve 56%, HCC capture improve 52%, and CC/MCC capture improve 44%, with an overall capture rate increase of 38%. At Rapid City Hospital, severity of illness scores rose 17% and risk of mortality scores rose 21%. At the smaller Spearfish Hospital, which had no CDI program running before this, the swings were bigger still: a 296% increase in capture rate and 352% increase in HCCs per encounter. The results were big enough that Werpy, presenting his own data, called the numbers "gaudy" and admitted he still laughs every time he sees that slide. The platform paid for itself in three months.

What adoption unlocked

Once trust was established on inpatient rounds, use continued to spread on its own. Advanced practice providers turned out to be some of Monument's heaviest users, and physicians covering hospitals overnight leaned on the platform's natural-language chat interface to type questions about unfamiliar patients rather than dig through old notes. That same appetite pushed the platform outside the hospital walls: Monument's physicians asked for the same diagnostic support on outpatient encounters, and health systems more broadly are now using it to screen for malnutrition risk, missed DVT prophylaxis, and hospital-acquired conditions that otherwise only surface after they become a penalty.

None of this happened quickly. It took a vote, a note template that needed real use before it won anyone over, and a year of small moments where Regard flagged something correctly and still let the physician make the final call. That's what built confidence in the platform, not a pitch. And it's what turned a documentation project into something bigger: proof that Monument never had a documentation problem, a quality problem, or a revenue problem. It had a visibility problem.