TL;DR: The biggest barrier to healthcare AI in 2026 is organizational, not technological — teams that define the same workflow three different ways, challenges that stay invisible until weeks after they start, and tools people don't trust enough to actually rely on.
Elevate, Medallion's annual virtual conference brings together healthcare leaders, innovators, and operators for meaningful dialogue on the most pressing challenges and opportunities shaping the industry. This year, one idea kept surfacing on its own, session after session.
A lot of what passes for "AI" in healthcare right now is just the same manual process with a nicer interface — and Elevate 2026 spent a full day making the case for how to tell the difference. This year, our speakers came from all corners of the industry yet landed on the same root challenge.
That challenge wasn't technology. It was the organization around it. When people that are far apart arrive at the same diagnosis on their own, it's worth taking seriously. Here's what it looked like across three of the day's sessions.
1. AI adoption isn't a technology problem, and the symptoms show it
Tiffany Long, System CVO Director at MultiCare Health, a panelist in the session, A Closer Look at Provider Credentialing: What's Working, What's Not and What's Next, put it plainly:
"I don't think it's a technology problem... we are looking at some real change that is beyond technology."
In a different session, Are Your Processes Protecting Revenue or Losing It?, Tami McMasters Gomez, Executive Director, Mid Revenue Cycle at UC Davis made a similar point about revenue cycle, saying "Revenue leakage is rarely one single failure point... it's usually the culmination of small breakdowns."
In the session, AI in the Admin Stack: What's Helping and What's Hype, Jennifer Mohler, Chief Administrative Officer at SMIL (Southwest Medical Imaging), made both points at once on denials: they're documentation and scheduling problems long before they're billing problems, and by the time one shows up, the real damage already happened upstream.
The blocker is alignment. When the teams touching a workflow don't share the same definitions, or can't see where it actually breaks, no tool reconciles that for them.
And the visibility gap is bigger than most leaders think: according to Medallion's 2025 State of Payer Enrollment and Credentialing report, nearly 47% of organizations polled said they don't know the dollar amount they're losing to slow enrollment workflows. You can't fix what you can't measure — and a lot of teams can't yet measure this.
2. AI's ROI comes from removing toil and surviving rigorous testing rather than the model itself
Stedman Hood, Co-Founder at Neon Health and a speaker on the session, Empathetic or Efficient? Can AI Deliver Both in the Patient Experience?, reduced the real work of deploying AI to one word:
"Testing, testing, testing, testing."
AJ Braga, Chief Information and Technology Officer at SMIL (Southwest Medical Imaging) and a speaker on the session, AI in the Admin Stack: What's Helping and What's Hype, sharpened the point from the operations side: a tool that "works on some but not others" can be worse than not automating at all, because someone still has to catch the misses, now without knowing which cases the tool got wrong.
Before you scale a workflow, confirm it actually removes work rather than shifting it to a different desk, and that it's held up against real edge cases at real volume. A pilot that succeeds on the easy 80% and quietly fails on the hard 20% hasn't earned a rollout yet.
3. Trust in AI has to be earned fast. The industry can't afford to wait.
In the session, Empathetic or Efficient? Can AI Deliver Both in the Patient Experience?, Jeri Garner, Principle Advisor at LEV Advisory Group offered the plainest test of whether a tool is ready for the people who'll actually use it: "Can my mother do this?"
Jennifer Mohler set the same bar from the buyer's side: "I would need to see it walk before it starts to run." The pattern held across every trust conversation: people route around a system they don't trust, and each workaround chips away at the return the tool was meant to deliver.
The complication is that the patience to build that trust carefully may be running short. Christopher Kerns, CEO of Union Healthcare Insight and a panelist on the session, Empathetic or Efficient? Can AI Deliver Both in the Patient Experience?, framed the pressure bluntly: health systems are "going to be investing in AI... because they're not going to have much of a choice." Trust takes time to earn, and demographic and cost pressure mean the industry may not have the luxury of taking it slow.
What to consider before you buy the next AI tool
Elevate 2026's throughline was that the fix starts with visibility into where provider workflows actually break. If you can take one thing to act on: start by getting the teams who touch a workflow aligned on how they each define it, before you invest in another tool to run it.
Many of those breakdowns can often trace back to that gap, and the time to address it may be shorter than it even feels.
That's an area Medallion can help with. Our platform helps provide a clear view of where things may be stalled in the process. If that's the gap you're trying to close, see how it works.
Watch the sessions on demand from our Resources page.
FAQs about healthcare AI adoption and Elevate 2026
How does AI reduce costs in healthcare?
Less by replacing clinical judgment than by removing the manual, repetitive work that sits around it: chasing documentation, re-keying data between systems, catching enrollment or credentialing delays before they become denials. Savings tend to show up when a tool removes a step a person used to do and holds up at real volume, not when it just makes the same manual step faster.
How much does slow provider credentialing and enrollment cost?
More than most organizations can name. In Medallion's 2025 State of Payer Enrollment and Credentialing report, nearly 47% of organizations said they don't know the dollar amount they're losing to slow enrollment workflows. The cost tends to show up indirectly: providers who can't see patients yet, delayed reimbursement, and write-offs that trace back to a paperwork gap weeks earlier.
How do you implement AI in healthcare operations?
Start with the workflow, not the software. Pick one process where breakdowns are costing you, get the teams who touch it aligned on how they each define it, then look for automation that is able to remove a real step and has been tested at your actual volume.
Who spoke at Elevate 2026?
A keynote from Blake Madden, Founder and Creator at Hospitalogy, and panel sessions with Tiffany Long, System CVO Director at MultiCare Health, Tami McMasters Gomez, Executive Director, Mid Revenue Cycle at UC Davis, Jennifer Mohler, Chief Administrative Officer at SMIL (Southwest Medical Imaging), Stedman Hood, Co-Founder at Neon Health, AJ Braga, Chief Information and Technology Officer at SMIL, Phillip Tornroth, VP of Engineering at Elation Health, Jennifer Urlaub, Senior Vice President of Revenue Cycle at Ivy Rehab Physical Therapy, Erik Pupo, Director, Commercial Health IT Advisory at Guidehouse, Mallorie Merboth, Director of Professional and Medical Staff Services at Boulder Community Health, Kim Bussie, AVP, Revenue Integrity and HIM at CHOP (Children's Hospital of Philadelphia), Christian Ellis, Credentialing Manager at Family Care Center, Christopher Kerns, CEO of Union Healthcare Insight, Shehzad Moiz, VP of Product Management at Simple Practice, Deb Muro, Chief Information Officer at El Camino Health, and Jeri Garner, Principle Advisor at LEV Advisory Group. We're grateful to every speaker and attendee who made Elevate 2026 what it was.
