A pattern kept showing up across a number of recent conversations with home health and hospice leaders, regardless of what the session on the agenda was actually called. Whether the topic was regulatory updates, value-based care positioning, five-year strategic planning, or PDGM and PEPPER reporting, the same underlying tension surfaced every time. Here's what that pattern actually looks like, and why the AI conversation inside it has changed.

The list keeps getting longer, not shorter

Agency leaders are being asked to do a lot of things at once right now, and none of them are optional. Improve quality. Protect reimbursement. Manage audit and denial risk. Adopt new technology. Operate more efficiently. Any one of these on its own is a reasonable annual priority. All five, simultaneously, with the same staff and the same hours in a day, is a different kind of pressure than hospice leadership has dealt with before.

That's the part that doesn't always come through in a regulatory update or a strategy session covered in isolation. Each topic gets treated as its own conversation, its own session, its own slide deck, but the leaders actually living it are carrying all of it at the same time, and the tradeoffs between these priorities are where the real stress sits.

What's actually keeping hospice leaders up at night

Underneath the formal agenda items, the same worries kept coming up in less formal terms. Audit and denial risk sits near the top of almost every list, and it's rarely framed as a fear of being caught doing something wrong. It's framed as uncertainty: not knowing whether documentation that felt sufficient in the moment will actually hold up months later if a reviewer looks closely.

That uncertainty connects directly to a theme that came up almost as often: the link between documentation and denials isn't treated as an abstract compliance topic anymore, it's treated as a direct financial risk. CMS's own improper payment data backs up why that worry is well founded, since unsupported medical necessity, not fraud, is consistently the leading driver behind hospice claim denials.

Data is being used defensively now, not just for improvement

PDGM and PEPPER reporting used to get discussed mostly as a performance benchmarking tool, a way to see how an agency compares to its peers. That's shifted. Leaders are increasingly using this same data defensively, pulling their own PEPPER outlier indicators before a surveyor or auditor does, specifically to find and fix a pattern before it becomes someone else's finding.

That's a meaningful shift in posture. Data reviewed for improvement gets looked at occasionally, when there's time. Data reviewed for defense gets looked at regularly, because the cost of missing something is no longer just a lower quality score, it's a real audit or denial risk sitting quietly in the numbers until someone finally looks.

The AI question has changed

This is where the conversation has moved the most. A year or two ago, most sessions on AI in hospice and home health were still working through "should we use this at all," largely driven by understandable caution around a new category of tool. That's not where the conversation sits anymore.

The real question now is narrower and more useful: where can AI actually remove work, improve consistency, and create measurable value, without introducing a new kind of risk in the process. That's a very different filter than "is AI good or bad," and it's a much better one. It means evaluating a specific tool against specific, concrete questions, security and compliance, accuracy measured at real scale rather than in a demo, and whether a human is actually in the loop reviewing what the AI produces, rather than a values debate about the technology in the abstract.

What this means for five-year planning

The leaders thinking furthest ahead aren't treating AI adoption as a side project to revisit later. They're treating it as infrastructure, the same category as their EMR or their coding process, something that needs to hold up under scrutiny for years, not just solve this quarter's staffing gap. That's a different kind of planning than evaluating a single tool for a single problem, and it's the mindset shift that seems to separate the agencies planning well for the next five years from the ones still treating every new technology decision as an isolated, one-off choice.

Talk through where this fits for your agency

This is exactly the kind of question our coding and OASIS QA review is built around: not whether AI belongs in the process, but where it actually removes work without adding risk. That combination of certified coders and clinical AI is how we hold to better accuracy than human-only or AI-only services, at roughly half the cost of a legacy vendor, with turnaround measured in hours instead of days.

If you're trying to figure out where that fits for your specific agency, we're happy to walk through what that looks like.