Health-e Law Podcast Ep. 25
Human-in-the-Loop: Bringing AI to Skilled Nursing Facilities
Thank you for downloading this transcript.
Listen to the podcast episode released April 20, 2026.
Welcome to Health-e Law, Sheppard’s podcast exploring the fascinating health tech topics and trends of the day. In this episode, Ernie Ianace, founder and CEO of CareAlly, joins partner and host Michael Orlando to explain how AI orchestration can allow skilled nursing facilities to automate complex workflows, while keeping humans in the loop for clinical and regulatory decisions.
About Ernie Ianace
Ernie Ianace is the CEO of CareAlly, an AI orchestration platform purpose-built to address healthcare’s most urgent operational challenges: workforce shortages and care coordination breakdowns.
A founder and growth-focused executive with more than two decades of experience, Ernie has built and scaled technology companies across healthcare, senior living, cybersecurity, IoT, and AI. Throughout his career, he has led global commercial teams, launched new markets, and forged strategic partnerships that have collectively generated over $3 billion in exit value.
In addition to his role at CareAlly, Ernie serves as Chief Commercial Officer of InsightAlly.ai, the parent company behind a versatile AI orchestration engine serving healthcare, government, education, and enterprise productivity sectors. In this role, he drives strategic commercialization and market expansion, helping organizations streamline complex workflows and reduce operational friction at scale.
About Michael Orlando
Michael Orlando is a partner in Sheppard’s San Diego (Del Mar) office. He is team leader of the firm’s Technology Transactions team, a member of the Life Sciences, Healthcare and Artificial Intelligence teams, and co-leader of the firm’s Digital Health & Innovation team. Michael has more than 20 years of experience advising health technology companies, insurers, healthcare systems and providers, academic medical centers and research institutions, medical device manufacturers, and pharmaceutical and wellness companies on intellectual property and business transactions in key strategic areas, including EHR systems procurement and integration, telehealth, mobile health applications, clinical decision support technologies, artificial intelligence, data use, wearable devices, remote patient monitoring, and other medical devices, research and collaborations, patent licenses, software licenses, joint ventures, mergers and acquisitions, revenue cycle management, and other outsourcing transactions.
Michael founded a software-as-a-service company before entering private practice and completed an in-house secondment at a publicly traded biotechnology company, an experience that informs his practical and business-focused approach to client engagements.
Transcript
Mike Orlando:
From hospital boardrooms to startup war rooms, this is Health-e Law. Powered by Sheppard’s digital health and innovation team, we bring you quick and candid conversations with industry leaders, bringing sharp analysis and critical insights into what’s next.
Welcome to Health-e Law. I’m Mike Orlando, a partner at Sheppard and your podcast host. Today, I’m excited to welcome our guest, Ernie Ianace. Ernie is the founder and CEO of CareAlly and co-founder and CCO of InsightAlly, AI-powered platforms that help skilled nursing facilities, special needs plans and provider partners automate high-volume workflows like admissions packet review, case management, claims denial management and enrollment outreach. His approach is human-in-the-loop, AI assists, humans decide and every output is traceable to source data. Welcome, Ernie.
Ernie Ianace:
Hey, Michael, thanks for having me.
Mike Orlando:
Yeah. So, let’s dive in and start to get to know more about you and CareAlly. So, for listeners who haven’t worked inside skilled nursing facilities—what we call SNFs, for those who aren’t in the industry—or health plans, what is CareAlly and what does the day-to-day operational pain you are solving help in the SNFs, in the skilled nursing facility world? Can you explain a little bit more about what you do?
Ernie Ianace:
Absolutely. So, we built this persona-based AI orchestration platform. What does that mean? That really means we can help skilled nursing post-acute health systems take workflows that are manual, right now, and automate them with 100% accuracy and high precision. That’s kind of the net of it. Right now, if you use the big foundational models—the ChatGPTs of the world, the Claudes—they hallucinate about 30% of the time. You can’t have that in healthcare. So, we built a platform that can take data that are inside the health system or the SNF’s walls—whether that be policies and procedures, stuff out of their EMR, admissions data—and use that in workflows to automate things.
In a nutshell, it’s that simple, but it’s really a lot of complexity underneath that where we have lots and lots of checks and balances and lots of intelligence to automate those workflows. But then we can also, if it’s a clinical workflow or maybe a high-value revenue capture workflow, we can also put human-in-the-loop as gates at specific steps, which is really, really important. You don’t want AI right now trying to make any kind of clinical decision, it’ll get it wrong. But you can have it figure out, “Hey, these are some issues that we think you should pay attention to. Here, take a look, and click here if you approve or edit or whatever you want to do.” So, that’s kind of the gist of the platform.
Mike Orlando:
No, that’s great. So, let’s talk more about that workflow automation. So, you said that the starting point, especially in the SNFs, is this workflow automation before more advanced AI use and ... How do you pick the first workflow to automate for a client, and what makes a good starting point to prove the value quickly without threatening the existing staff, for example?
Ernie Ianace:
Absolutely. And let me take one step back and then I’m going to answer your question. The first one is health systems, SNFs, operators, they’re all fatigued with point solutions. Point solutions are a single app that does one thing. And the problem you have is, if you have 10, 12, 15 of those, you have to maintain all of them—very expensive. You also have to pay for all of them and they’re all an app which is not the same thing as, say, a workflow automation. So, they’re going to charge you more and your IT staff has to manage them all, you have to worry about security on all of them. So, we built a platform that you can add workflows to, but it doesn’t lock you in to having lots and lots more apps. So, that’s the first one.
What I tell customers, partners all the time is don’t boil the ocean. Let’s pick your biggest problem first that is just highly manual, right? What are you doing? I’ll give you an example. Got a customer that, when they went from an ISNP, an institutional special need plan, and rolled out a CSNP, they immediately went to the top of the Medicare Advantage roles in that state and their enrollment went crazy. Good for them. Bad for them was… is that the claims got denied at almost 50% because they were way more complex than their system had seen before, so they were throwing bodies at it: six, seven more people to try to do claims denial mitigation. That’s a perfect example; that’s super automatable. You basically take in claims, you read them, you look at them, you look at the contract from the payers and say, “This is why it was denied, fix it, resubmit.”
Simple, on the surface workflow, a lot of rules that have to go behind that, but it takes probably five people away from their claims denial mitigation department and allows them to redeploy. In our industry where there’s a massive labor shortage, if they can take nurses away from doing that kind of incredibly laborious manual task and redeploy them on paying attention to residents and patients, that’s a monster home run. So, don’t boil the ocean, pick the pain point that’s highly manual first, do that first and then start doing the next things.
Mike Orlando:
Yeah, that makes sense. AI is great for this resource-restrained task like that. So, yeah, that makes a lot of sense to me. Why don’t we talk about human-in-the-loop? So, I know you emphasize AI assist, human decide and you said there’s a lot of points where you could put in the human touchpoint to make sure that the decisions are made, especially in the clinical piece. With those outputs being fully audible and traceable, et cetera, I’m assuming, what does it look like in practice for clinical and regulatory decisions? And how does it change the comfort level for the operators and the clinicians and the compliance teams, et cetera?
Ernie Ianace:
Auditability is 100% a must. It’s great to say human-in-the-loop, but you got to be able to prove the human was even asked in the first place at the right time for CMS, for the payers, for compliance, for all that kind of stuff. So, that’s critically important. We built that almost first in … the audit trail has to be traceable step by step. Every single thing the machine or the system does has to have an audit trail and easy to find. So, that’s the first thing.
Some workflows, you don’t really need a human in them. They’re so basic automations, once you verified they were accurate for a month or two, you can probably take a human-in-the-loop and redeploy. But there’s certain things you never are going to let the AI make a decision, or at least not for the next few years where it’s a clinical decision support tool. Yeah, great, help them with the decision, but you can’t make the decision for them.
So, again, it depends on the automation, the workflow, what you’re trying to accomplish, where you have and if you have human-in-the-loop, but regardless, you must have auditability and traceability down to the step level. This happened here, then this happened next, then this happened next, and you need to be able to see a map of absolutely everything or generate a report, everything that made that decision happen and then the downstream effects.
Mike Orlando:
The other piece of that was what does it do in terms of the practical way that your clinicians work with that or the operators? How does it change their comfort level?
Ernie Ianace:
We’ve always talked about is the system needs to let the clinicians work at the top of their license. So it’s automating the busy work and the noise at the bottom that they’re spending more and more time every single day and not paying attention to patient care. So, the goal here is to get rid of the messy manual stuff and not try to force anyone to change the way they do things clinically. Just give them data at the right point. Because you can’t go into a busy skilled nursing operator and say, “Hey, you got to relearn new stuff.” That’s ridiculous. You’ve got to say, “Hey, we just gave you an hour back every single day and more data to make the decision when you’re in front of your patient.” That’s where you get the comfort and the trust built up. It’s critically important. If you ask someone to change fundamentally a workflow, they can’t, they don’t have time and to roll that out at scale is impossible.
Mike Orlando:
I think that’s a good lead into the next question which is, when you’re dealing with an implementation of this system and you’re bringing it into the SNF environment, how does that look different than other healthcare setting, like a health system or a health insurance plan?
Ernie Ianace:
It’s really a lot similar. They’re different use cases and different scale. So, skilled nursing is, unfortunately for them, a bit behind, technologically, than the bigger health systems who started automating 10 years sooner, but they all have the same problems. And the payers probably are the most automated already because they’ve been throwing billions and billions of dollars at solving that enhanced margin, and they have the resources. Skilled nursing operates at much smaller margins. So, one, you probably have to take smaller steps in skilled nursing than you can at health systems, than you can at payers. Kind of in that order, right? Because health systems have much larger budgets, they have giant IT departments. If you’re talking about the Mayo Clinic, billions of dollars. Where skilled nursing you’re dealing with thousands of dollar chunks.
So, it’s really… you got to think smaller, you got to give them immediate return on investment. They don’t have six, eight, nine months to get an ROI. You got to show ROI in month one. And if you’re not helping them save a massive amount of hours or generate more revenue for them, don’t even try, because they just don’t have the resources or the time. Where as you can think a little bit bigger, a little bit more monolithic and strategically at a health system. And at a payer, they’ve got money to burn, but it takes a lot longer to get them to trust something new.
So, it’s a little bit of difference, but it’s still the same thing. You’re automating manual things and you’re tying disparate systems together to get data from here, say, PointClickCare in a SNF, over to here to maybe their clinical EMR, maybe it’s CureMD or eClinicWorks or whatever, and mapping the difference between them. It almost looks the same in each bucket, it’s just more about scale.
Mike Orlando:
Yeah, that makes a lot of sense and, yeah, I understand the constraints there with the SNF and the margins, so that’s a real practical way to look at it. I also had a question about details of your solution you offer, if we could talk about that for a second. You described in some of the materials I read that you’re able to detect when a SNF patient is readmitted to a hospital and then alert the payer in real time something they might not otherwise know about. So, how does that real-time data sharing change that care management and the denial prevention and the outcomes in these value-based arrangements?
Ernie Ianace:
The good news is that machines never sleep, they never get tired, they never call in sick. So, if you give them the feed, the data feed, and those are available via TEFCA and CareQuality and the QHIN Networks, we can do things like detect that a patient that was just discharged in the 30-day window got admitted to their local hospital. And skilled nursing needs to know that, They’re still usually on the hook for that.
Or, if it’s an ISNP that their patient is in the SNF during the long-term period but they just went to the emergency room, the ISNP wants to know that. They all want to know that. It’s just really, really hard historically for them to know that until after the fact. They usually find out via a claim that they get whacked with that it happened. We can detect it in real time via these automated messages and exchange networks and let them know up into their workflow that said, “Hey, patient Jane Smith just got readmitted. You discharged her last week, but she’s now back in the ER.” And, by the way, it may not even be for the same condition which, if you’re in a value-based world, doesn’t even matter, you’re still responsible.
So, contextualizing that data, first detecting the signal, doing the right thing with the signal and contextualizing that signal like, “Hey, they are in a value-based arrangement or, no, they’re in just standard part B,” all that data matters. And if you need to get it to the right person inside a skilled substitute post-acute or health system, it’s all the same thing, it’s really visibility to actionable data they don’t currently have that we’re able to give them.
Mike Orlando:
All right, Ernie. So, listening to that and understanding the skilled nursing world, how many times do you hear from those operators that, “Hey, I know you have this great tool, but we have ChatGPT and we have Copilot and Claude, and they’re coming out with their own healthcare-focused models now.” What do you say to that?
Ernie Ianace:
I get that question at the front end of every first conversation and it’s a perfect question because it seems obvious, so why don’t we just use that. Honestly, sometimes it’s more expensive. But the bigger problem is those guys still hallucinate 30 to 35% of the time, because they’re designed to answer everything. They go out to the whole world and just get it… and they’re supposed to be helpful, so they usually say “yes” and just give you an answer whether it’s right or wrong, and sometimes it’s wrong on page 37 of their answer and you miss it.
And I’ll just give you one quick analogy. Walmart, there was just an article I just read in The Wall Street Journal yesterday, fired OpenAI. Five months ago, they announced this giant partnership. OpenAI was going to help them help their customers shop better. It was an utter failure because, again, it wasn’t tied to their specific rules, their specific data and saying here’s the guardrails and the governance. So Walmart had to go build their own system on top of the foundational models, those Claudes and ChatGPTs. You can still use them as long as you have arbitration, you have governance, you ask four or five of the models the same question because good chance, if they’re hallucinating 30 or 40% of the time, one or two of them are going to be wrong. But if you have six of them, four of them agree, that’s probably the right answer.
So, there’s ways to use them that allow you to take advantage of their power, but you have to build all kinds of controls to avoid all their hallucinations that ... They lie with a smile, and that kills you in healthcare. It’s completely not okay. It doesn’t work in law, it doesn’t work in healthcare.
Mike Orlando:
Yeah. Not to mention there’s a lot of regulatory issues around…
Ernie Ianace:
Privacy, PHI, all of it, right?
Mike Orlando:
Yeah, yeah. There’s so many things there but, yeah. I’m just curious because we also see those kind of questions coming up a lot and that makes a lot of sense. I want to ask a last thing. I know you have an upcoming webinar and that focuses more on moving from fragmented tools to a modular architecture. And so, for executives evaluating AI right now, what should be modularized first? How do you avoid this AI sprawl of these various disconnected, independent systems and what does the data governance choice look like? Why does it matter?
Ernie Ianace:
So, that’s a really good question. Again, I alluded to it earlier, you need a platform. These fragmented point solutions are killing healthcare. The big hospital systems, healthcare systems started moving away from point solutions six, seven years ago. There are still lots of stranded legacy systems that they’re stuck with, but you need a platform to start. And where we talk about modularization, it’s really the workflows that are modular. You can just keep adding new workflows and new modules and extending the ability of the platform. But what you get when you start with platform first and, by the way, you shouldn’t have to pay for a platform upfront, you should just pick someone that has a platform that has SOC 2, that has HIPAA, that has all the tight security and the regulatory guidelines nailed for you, and also has some of the interconnectivity to your legacy systems like your EMR and the TEFCA network and all that kind of stuff.
But, the way you deal with the modules is you just, like we said before, pick the first workflow that you want to automate. That’s module one. And it could be different for each skilled nursing facility. Somebody wants to maybe do claims denial mitigation, somebody wants to do decision support, someone wants to do admissions, ingestion of data, decisions and taking out care plans. Those are all modules or workflows, but it’s on a holistic single platform that you know has all of the security, all of the governance, all of the traceability and the interconnectivity first so that you’re not, again, having to go get nine or 10 or 12 different solutions from different companies and manage all that mess.
Mike Orlando:
Yeah, that makes a lot of sense. Yeah, and we’ll put more information about your webinar upcoming on that in our show notes so that anyone listening can go there and look at that. So, Ernie, our time’s up and it’s been really great talking to you. I really appreciate you coming on today. There’s a lot of great information in the skilled nursing space about AI and, anyway, look forward to talking further about this in the future.
Ernie Ianace:
Sounds good. Thanks, Michael.
Mike Orlando:
Thank you. That’s a wrap on this episode of Health-e Law, powered by Sheppard’s digital health and innovation team, where health innovation meets legal expertise. Until next time, stay healthy and stay informed.
Resources:
* * *
Thank you for listening! Don’t forget to SUBSCRIBE to the show to receive new episodes delivered straight to your podcast player every month.
If you enjoyed this episode, please help us get the word out about this podcast. Rate and Review this show on Apple Podcasts, Amazon Music, or Spotify. It helps other listeners find this show.
This podcast is for informational and educational purposes only. It is not to be construed as legal advice specific to your circumstances. If you need help with any legal matter, be sure to consult with an attorney regarding your specific needs.