“As new requirements have emerged in the industry, what we've seen is that EHRs have continued to add layers, and the result of that has been a lot of patchwork systems. Clinicians and their staff end up doing a lot of coordination jobs just to make all these different fragmented systems work,” he explained. “The big realization for us was that you can't solve this problem by continuing to add more fragments. There are organizations adding AI piecemeal on top of their existing systems. We thought to really solve this problem correctly what we had to do was re-imagine what that EHR experience could be, not just for clinicians, but for the staff, and how we are orchestrating work that has to happen across patient, across clinical, and across revenue cycle. By starting with an AI foundational layer that's orchestrating tasks and activities, that’s how we're going to help unify these workflows, end to end.”

Michael Kozak, M.D., is a practicing physician and assistant chief medical officer at of HealthLinc, a network of FQHCs in Northern Indiana that has 14 clinic sites and sees approximately 60,000 unique patients per year.

He said that HealthLinc had previously been suffering with a lot of what Cohen described as fragmentation and has been partnering closely with Greenway over the past few years in the development of Novare.

He noted that clinicians are amazed by the types of information AI can help surface, but they do want that integrated into their work flow. “One of the nice things with Novare is rolling out small parts of it at a time, and every time one of those little parts has been added, the feedback from the clinicians has been very positive,” Kozak said. “Now we’re not doing that coordination between systems; the information is just served up to us. So far, what has been most profound is the AI ambient listening, because that's the most mature. But the summaries, as well as the clinical insights from some of the documentation — those are the next big steps.” 

Cohen said studies have shown a significant amount of clinicians’ time in the EHR is spent on messaging, ordering, and billing — tasks that Novare is focused on. “One of the interesting testimonials that we heard recently from a customer is that focusing on documentation improvement is having significant downstream impact and benefit on the revenue cycle team. Because the note is better and cleaner and the coding more accurate, that is helping to facilitate downstream coding accuracy and claims collection. So there's a huge systematic benefit to everything that's happening in the clinical workflow, and that's why a core focus of ours is how can we help to ensure most of the work that needs to happen is happening inside of the visit and not getting pushed downstream, which really just facilitates rework?”

Kozak added that the whole idea of this AI integration is that it's more intuitive, and HealthLinc is finding that to be true with Novare. “The training is so intuitive that even folks who are struggling to find the ‘on’ button on the computer are able to understand it,” he said. 

HealthLinc is already seeing some benefits in terms of burnout reduction. “If we're starting to annoy our providers less, and they feel more of that joy of medicine, they're able to take on another acute patient, they’re able to spend the time they need with someone who's suffering,” Kozak said. “The efficiencies are not quite as measurable as I think a CFO type would like, but they are very apparent to us in the anecdotes that we hear and the smiles that we see.”

Cohen said an early decision was to work with Amazon Web Services to build their own Greenway agentic AI factory. A number of agents that are deployed through that factory Greenway developed in house, but there are some capabilities that Greenway is partnering on. “That’s just because we want to move fast,” he said. “We’re not trying to reinvent the wheel. But we also know what we're doing with Novare is cutting edge stuff, and there are going to be a lot of things that are first in the industry that we are going to have to build — things like our AI patient contextualization and summary are things that we've built internally.”