🚀 Meet Us at HIMSS 2026 | March 9–12, 2026 | Las Vegas, NV 🎉
Most conversations about AI in healthcare focus on the technology itself, what it can do, how accurate it is, what it costs. What almost never gets discussed is the human work that happens before any of that matters: who inside a healthcare organization drives AI adoption, what they actually do, and why so many rollouts fail without the right people in the right roles.
That is what we wanted to understand, so we spoke to Amanda McFayden, Director of Clinical Operations at DENT Neurologic Institute, one of the largest outpatient neuroscience centers in the country. Amanda has been at DENT for nearly 20 years, starting at entry level and working through almost every part of the organization before reaching her current role. She has seen healthcare technology evolve across two decades, and over the last year she led the full adoption and rollout of Marvix AI across 60-plus providers at DENT.
In this conversation, we explored three questions: what role does the admin team and senior leadership play in adopting AI, what does that look like on the ground inside a large practice, and what can AI companies do to make the process work. Below are the key themes from that conversation.
AI stopped being optional for organizations like DENT not because of technology excitement but because the administrative burden on providers reached a point where something had to change. Documentation requirements kept growing, insurance authorizations kept getting harder, and staffing gaps from COVID never fully closed. Senior leadership at DENT recognized that absorbing this through EMR triggers and workarounds was not going to hold.
Their motivation was specific: reduce provider burnout driven by documentation overload. Better documentation also meant stronger insurance authorization outcomes downstream, but burnout was the primary problem they were solving for.
If you are evaluating AI for your practice:
Before talking to a single vendor, DENT's senior leadership and strategic development team sat down and defined what they actually needed. Amanda describes building a structured checklist with two non-negotiable must-haves: customization at the provider level and two-way EMR integration.
Customization mattered because outpatient neurology covers a wide range of subspecialties, and each provider has their own clinical voice. A tool that flattened that into a fixed template was never going to get buy-in. Two-way integration mattered because most neurology visits are follow-ups, and without the ability to pull previous visit summaries into the current note, the ambient AI scribe is doing only half the job.
DENT piloted other solutions before landing on Marvix. The tools that failed did so because providers said the notes did not sound like them. Once that trust breaks, training does not recover it.
If you are building your evaluation criteria:
Signing with a vendor is where the harder work begins. Amanda started with seven or eight provider champions drawn deliberately from across physician types and subspecialties, testing whether the tool could handle the real range of how DENT providers work. Once that pilot group was seeing results, their peers took notice. Providers watching colleagues take lunch and stop post-charting at night did more for adoption than any formal rollout campaign could.
She handled resistant providers by running short optional demos scheduled around clinic hours, tracked training completion individually, and followed up with anyone who had not finished their setup. That operational discipline from the client side is what separates a tool that gets adopted from one that stalls.
If you are rolling out AI in your practice:
Amanda's internal playbook started with three questions the team worked through before anything else: where are we starting, why are we implementing this, and what are we asking providers to change. They also looked hard at cost, not just the tool cost but integration cost and what the organization would need providers to do differently to make the investment worthwhile.
They ran surveys throughout the pilot to check whether the tool was actually reducing burnout, asking providers directly about fatigue and end-of-day workload. They built in formal sign-off steps so providers acknowledged what the ambient AI scribe was doing in their workflow. Long-term success for Dent means reduced provider turnover and improved satisfaction over time, and Amanda's view is that the cost of losing an experienced provider to burnout far exceeds the cost of a tool that could have prevented it.
For your internal playbook:
Neurology is a hard environment for ambient AI. The clinical language is complex, subspecialty terminology varies, and providers differ significantly in how they speak and how they prefer their notes to read. The tools DENT evaluated that failed did so because they could not handle that variation.
What worked about Marvix was a combination of clinical quality out of the box and the ability to customize from there. Corrections to names and spellings persist across all future notes, which matters in neurology where medication terminology is often unusual. Note sensitivity and output style can be adjusted per provider. And when DENT gave feedback, Marvix built it back into the product. That responsiveness is what Amanda identifies as the real differentiator between an AI company worth working with and one that is not.
What to look for in an AI vendor:
Integration is often treated as a technical checkbox when in reality it determines how much value an ambient AI scribe can actually deliver. DENT started without full integration during the pilot, and even then providers found the tool faster than their previous workflow. Once full integration went live, the Marvix summarizer began pulling previous visit summaries into the current note, and that changed the pre-charting and post-charting dynamic significantly for follow-up visits.
The Marvix E&M coding recommendations and ICD-10 suggestions that surface from the visit note also proved useful, particularly for newer providers without formal billing and coding training. Marvix surfaces the rationale alongside the suggestion, making the provider's review faster without replacing their judgment. Amanda describes the integration itself as one of the smoothest DENT has done across any technology layered over their EMR.
On EHR integration:
Amanda's core point is that AI adoption in a large practice is an operational challenge, not a technical one. It requires sustained hands-on involvement from the admin team and senior leadership that most organizations underestimate. Providers need relationships, not memos. They need training that works around their schedule. They need someone they can go to when they do not know how to articulate what they need from the tool.
This role does not need to sit at the director level. A clinic manager embedded in day-to-day operations can own it. What matters is that someone specifically does.
For clinical operators starting this process:
Amanda's view is that healthcare organizations in 2026 do not have the luxury of waiting. Payers are not getting easier, the staffing environment has not stabilized, and the administrative load on providers is not going to reduce itself. Her advice is to re-evaluate your processes and tools at least once a year and be honest about whether what you have is still the right fit.
The gap between a successful AI rollout and a failed one is almost never the technology. It is whether the organization has the right people, with the right relationships, doing the operational work that adoption actually requires. Amanda and her team at DENT are a strong example of what that looks like in practice.
To learn more about how Marvix AI supports specialty practices, book your 30-day free trial today.
.png)
