Should You Switch Your EMR for an AI Scribe?

The push to reconsider our electronic medical record (EMR) system didn’t start with artificial intelligence. It started with staffing.

Like many dermatology offices, we have historically relied on clinical staff who do far more than room patients. Our nurses and medical assistants keep the clinic moving, anticipating procedural needs and scribing. That model worked when staffing was stable and salary increases were predictable, but it is increasingly difficult to sustain. Turnover can vary, finding appropriate candidates is ever challenging, and the cost of experienced clinical staff continues to rise. Documentation has insidiously become one of the more expensive uses of clinical labor in the practice.

One relevant point of comparison: no one doctor, nurse or office manager we have ever spoken with actually likes their EMR system. Not a single person. And this encompasses all the products mentioned below, so perhaps that is the largest point of commonality.

Where We Are

Our practice has used Nextech in some iteration for the past 20-plus years, even dating back to its predecessors. In many ways, it still serves us well. Photo management is strong, before-and-after workflows are intuitive, we can host on premises for greater security than the cloud, and the platform is highly customizable. Those strengths are not easy to replicate.

And yet, it feels increasingly dated. AI capabilities, particularly anything resembling an ambient scribe, are essentially absent for dermatology. When Nextech launched its new artificial intelligence (AI) tool, Cora Scribe, in January 2026, the announcement made our situation clearer, not better: the company described Cora as “designed specifically for ophthalmology,” with expansion to other specialties planned only “over time” (with no timeline given). Our office manager attended the EDGE meeting earlier this year and came away with a similar read: ophthalmology is clearly Nextech’s priority. As I will show with ModMed below, customer service throughout the industry has and likely will always have its challenges.

A Look at the Alternatives

ModMed’s EMA is the most obvious comparison. It has leaned aggressively into AI, most notably with ModMed Scribe, now in its second major version, which is natively embedded in the EMA platform and trained on de-identified data from more than 750 million patient encounters. It was shockingly robust when we tested it recently. Even when we tried to trip it up with complex phrasing, information irrelevant to the encounter, or details from a different patient’s history, it performed well. On paper, it directly addresses the problem we are trying to solve.

Still, nearly every post in The Board-Certified Dermatologists group on Facebook about ModMed bemoans the costs and lack of service after the sale. The consistent feedback is that pricing escalates significantly at renewal and bespoke features are nearly impossible. Switching EMRs is already a major investment; doing so into an uncertain long-term cost structure gives us pause. Not all practices are comfortable yet with AI scribing due to the challenges in reading over notes and making sure that the notes themselves are accurate.

EZDerm is simpler and dermatology-focused, with strong body mapping, lesion tracking, and automated coding at a more accessible price point. When we tested it a few months ago, it had not yet matched ModMed’s depth in ambient AI scribing, but for practices that want a purely dermatology-built environment without enterprise pricing, it appears to be a serious contender.

A few newer entrants are worth watching. The subQdocs platform, built by a team that includes a board-certified dermatologist and Mohs surgeon, takes a different approach: rather than replacing the EMR, it layers AI-native documentation on top of existing workflows. It provides pre-visit summaries generated from patient history, ambient note generation during encounters, and dermatology-specific fluency across skin checks, biopsies, excisions, and lesion destruction. An “EMR Light” component for scheduling, prescriptions, and labs is reportedly in development. Edvak, meanwhile, positions itself as a fully AI-native EMR (meaning AI is fundamentally built in, not simply added on) with its Darwin AI engine structuring notes in real time while simultaneously generating billing codes. When we tested it earlier this year, however, there was still no iPad app, which ruled it out for us.

The Real Decision

The harder question lurking beneath all of this is not which EMR to choose but whether this is even the right moment to choose. Consider what AI has done to consulting. McKinsey and Boston Consulting Group spent decades building their value around the weeks of analyst work it took to produce a market assessment or competitive landscape. AI tools now generate a credible first draft of that same work in an afternoon, and the big firms are not simply buying better research tools; they are rethinking whether the underlying staffing pyramid still makes sense. The EMR category may be facing the same question. A general-purpose AI agent that handles clinical documentation, coding, and workflow management without a specialty-specific platform underneath it is not science fiction; it is probably closer than a 3-year renewal cycle suggests. The risk of locking into a platform today (at significant switching cost and contractual commitment) is that it bets against exactly that kind of change.

Even if the EMR category is not upended entirely, the competitive pressure from AI-native entrants may finally constrain what established vendors can charge at renewal. The pricing dynamic that makes ModMed’s escalating costs so frustrating to current users could look very different in 2 years if OpenAI, Anthropic, or some other company that does not yet exist gives practices a genuine exit option.

The staffing-plus-costs-of-documentation problem that started this conversation is real and it is not going away, but solving it by anchoring to a platform we may need to leave again in 3 years is its own kind of risk. AI may ultimately solve what we started trying to solve. The question is whether it will do so through our current EMR, a competitor, or something that has not launched yet. Like everything in our field, we must find a way to feel comfortable making important decisions with incomplete data. Warren Buffett’s longtime partner Charlie Munger once said: “It’s waiting that helps you as an investor, and a lot of people just can’t stand to wait.”¹

“

One relevant point of comparison: 
No one doctor, nurse or office manager we 
have ever spoken with actually ‘likes’ their 
EMR system. Not a single person.

”