John Beck has witnessed the full evolution of the EHR market, working first at Allscripts, then at NextGen before becoming senior vice president of strategy at Nextech, whose EHR focuses on specialty practices such as ophthalmology, orthopedics, dermatology and plastic surgery. With that long-term perspective on the health IT market, he spoke recently with Healthcare Innovation about why specialty practices might be early movers in the agentic AI space. 

Healthcare Innovation: I understand that you have a somewhat contrarian perspective on where agentic AI is likely to create value in healthcare first and who's best positioned to benefit? Could you explain that? 

Beck: Part of understanding this is understanding the difference between AI and agents and agentification. Simply put, AI is really good at finding an insight. Its job is to generate something that you hadn't thought of before. When you think about five claims coming up on a timely filing deadline, that’s what AI does. You don't have to tell it; it knew it was important and it surfaced it. When you say take those five claims and put them in a billing queue, that's an example of what an agent does. You combine an agent and the insight, and all of a sudden those five claims end up in a queue. When you think about agentification, it's entire end-to-end workflow. I'm going to call you back when the payments are received on these things. I expect that payment to be received from Aetna within 48 to 72 hours, based on what I see in your data. It’s the orchestration. 

When you think about the needs of a specialty practice, that's where the magic comes in because specialty practices are incredibly complex, with a lot of moving parts. They're not provider-constrained; they’re team-constrained, and anytime you have multiple team members and multiple departments and multiple things touching a patient journey, the opportunity for agentification to solve those things is exponential.

HCI: But isn't that true in primary care or in the hospital setting, too?

Beck: Sure, but the yield is different. I would suggest to you that the pressure on these specialty practices with the labor dynamics that they have is enormous. This is about team time. If you have a 45% shortage in technicians in the marketplace, that’s a really interesting thing to manage around. Agentification does not make you have more people, but it puts every person you have at the top of their license and levels out the peaks and valleys. That's a really complex thing to pull off, and that's what AI and agents are great at.z

Beck: I think the advantage we have is the understanding of the workflow and the data that we have. When you have 20% market share in ophthalmology and 30% share in plastic surgery, you have a lot of data and a lot of workflows at your disposal. You have a lot of advantages that are not obvious to groups that are either wider or not nearly as complex.  But if we rested on our laurels and simply relied on AI to provide more insights that needed action and did not build the agents, we’d be doing the market a disservice. I think there's a lot of room to work at the intersection of our complex workflows, our system legacy, and these new tools, which are fabulous.

HCI: Nextech recently launched Cora Scribe, an AI assistant. Can you talk about that tool and where that's having an impact with customers?

Beck: First of all, all EHR vendors, regardless of where they are, know enough to put a scribe at the disposal of the doctors. When you are serving a generalist market, it is actually a very easy thing to do. It is really dictation on steroids and putting it in the right part of the chart. Even in that simple construct, there are vendors that have done it well, and vendors that have done it terribly. A lot of the vendors capture the voice and come back 30 minutes later or the next day with a completed chart that still needs a lot of editing. Those were the early days of scribing. It was acceptable. 

When you start talking about specialty clinical workflows, we wrote something that's entirely different. It is entirely of our own making. It is real-time chart injection. It is from the provider and/or the technician's voice into the chart in real time in all of the right places, including some very granular data that's involved in when you do an ophthalmology exam, for example. The standard is much higher. 

The approach we took is much more interesting. It uses all of the shortcuts that the doctor uses. It uses everything that they already know about the doctor to create a full chart experience with a one-second delay. There is no after-hours editing. There is no going back and looking at somebody else's work. It is fundamentally different. Starting with clinicals over revenue cylcle was an interesting choice, but now it's paying dividends because it is entirely different from what everybody else has.

HCI: On the revenue cycle or patient engagement fronts, is Nextech building some agents for its specialty practice customers to turn on or customize for their use?

Beck: Yes, we are doing them ourselves. But anybody can write an agent. Anybody can now automate a bad practice. The question is which practices do we expose that sort of capabilities to? I'll give you an example: If a claim hasn't been paid in 60 days, rebill it. That is one of the worst practices ever. The claim wasn't paid because there's probably something wrong with it. Why would you rebill it and set the cycle again for another 60 days? You could write an agent to do that for you without a person having to do it, but it's not a good idea. These tools become very interesting in the hands of the right operators vs. the wrong operators, and I believe we're going to find we can probably solve problems much faster for our folks in really good swim lanes, while still giving them some choice in the last 10% of what they want to happen to that particular problem. 

HCI: Are the practices having difficulties solving those problems due to being overwhelmed with the amount of work that they have or the lack of staff members to do all that work?

Beck: Yes, it's a little bit of both, and they typically will bring in outside consultants to help them with this. I'm not sure if that's a help or a hurt. A lot of these groups are in acquisition mode. A lot of these groups are being acquired or disaggregated. There's still a lot of movement going on. But fundamentally, if somebody's house is on fire because they don't have enough med techs, taking the time to make their revenue cycle go faster isn't the right answer. The right answer is solving for the med tech issue. So I believe that the playing field right now — given labor dynamics in specialty practices — is complex enough in certain geographies where we would be better served helping these guys get 90% of the way and letting them get the last 10% themselves.

HCI: Have you seen a lot of consolidation in these specialties where there are centralized offices handling the administrative things for a lot of dermatology offices, for instance?

Beck: Yes. There are a couple of different cohorts to pay attention to here. Private equity is where you see the most aggregation. Private equity is all about don't disturb IT early, but in the end, make sure everybody does things the same way. That's how they get their leverage.

When you talk about smaller geographic consolidations, those are much more interesting. Those are not private equity-backed, but I find those ones to be a lot more creative when it comes to expressing themselves. There you see some really cool innovation. It’s in the 20- to 100-provider space pre-aggregation. That is just a really good place where folks are still basically street fighters and enjoying differentiating themselves.

HCI: But does that consolidation factor lead to churn in the EHR market? And is that good or bad for Nextech?

Beck: We love churn, because a lot of the legacy vendors are typically the targets of the churn. If you're the incumbent, and your platform is client/server and not receptive to cloudification, that’s not where you want to be right now. We think the churn in the market plays to all of our strengths. 

HCI: Are some legacy vendors falling away because they’re not modernizing their platforms?

Beck: They have to choose who they want to be. If you're a generalist and you're playing in specialty markets and also behavioral health and in FQHCs, that's not a winnable hand. You have to choose. You can't be great at everything at the same time. 

HCI: Is there anything else about agentic AI that you want to mention? 

Beck: Agents are going to become a commodity. They are going to be ubiquitous. The question is, do you recognize that AI is glue, not capabilities? You're not trying to add capabilities. You're trying to make the current capabilities have fewer breaks between them and make things run better. Also, the capabilities you have are typically used by different team members, so therefore you're addressing the labor dynamics piece at the same time. Not everybody understands AI and agentification as glue. Most people think I'm adding AI — it’s a feature and I'm going to sell that feature. That's not the way to think about this. It's an entirely different way of coming at it.