AI in Healthcare: Building Smarter Solutions for Measurable Impact

[00:00:00] Auto Intro: Welcome to Health+Tech, the podcast where we explore the law that applies at the intersection of healthcare and technology. Hosted by McGuireWoods, episodes feature healthcare innovators and updates on the laws that apply to digital health businesses.

[00:00:18] Kristen McDermott: Welcome back to the Health Plus Tech podcast, the podcast where we explore the intersection of innovation and healthcare.

[00:00:25] Kristen McDermott: I’m your host, Kristen McDermott Woodrum, a healthcare partner in Atlanta at McGuireWoods, along with my co-host, Allison Mauer, healthcare partner in New York. Today, we are thrilled to be joined by Ghazenfer Mansoor. I will let him correct the pronunciation of his name. He is the CEO of Technology Rivers, a DC-based software development firm that has carved out a really impressive niche in AI-powered healthcare solutions.

[00:00:54] Kristen McDermott: Ghazenfer, welcome to the show.

[00:00:58] Ghazenfer Mansoor: Thanks, Kristen. Thanks for having me. You’re all right. Ghazenfer Mansoor is my name.

[00:01:05] Kristen McDermott: All good. Perfect. Perfect. I took four years of Latin, and languages and pronunciation are not my forte. So let’s get started. Let’s start at the very beginning. Technology Rivers was founded back in 2015, and from what I understand, the firm has grown into a specialized healthcare software development company that helps startups and growing enterprises build affordable, high-quality software.

[00:01:31] Kristen McDermott: Can you tell us about this journey? And specifically, what did you see in healthcare that made you think this was a space where a development firm could create real differentiated value, not just build software, but actually shape outcomes?

[00:01:47] Ghazenfer Mansoor: Thanks. Yeah. So we started in 2015 with the motto of building the product the first time the right way, helping startups in product development.

[00:01:58] Ghazenfer Mansoor: Our second customer was health tech. That’s how we gradually got into healthcare. It wasn’t really planned initially. We were just building products for all different types of startups. As we started getting into more and more health tech, we started working with one of the startups from Hopkins Innovation Center, and that’s when we learned HIPAA.

[00:02:21] Ghazenfer Mansoor: And then gradually we started doing more and more HIPAA-compliant products. So as of today, we have built over fifty healthcare applications. Half of them are HIPAA compliant. And as you focus more and more on health tech, you get more and more projects, and now this is purely our focus: AI, healthcare, and mobile.

[00:02:41] Ghazenfer Mansoor: So that’s the trio we focus on currently.

[00:02:45] Allyson Maur: That’s fascinating. As you started working more closely with these startups and service-based businesses and building these apps, you really had a front-row seat to see how companies were trying, and often struggling, to scale through technology. When you look back at the early years, what patterns or gaps did you start noticing that ultimately shaped how you built Tech Rivers?

[00:03:07] Ghazenfer Mansoor: So yeah, over the years we have seen many different patterns. Different startups would come in. I think the bigger challenge has been this feature race, where people would just start building things and then realize, okay, now we want to sell it. That’s the hard lesson that even I learned in my own startup before.

[00:03:27] Ghazenfer Mansoor: So that is one of the bigger pieces of advice: build something that your customer needs. Many times, we push them hard in terms of, okay, do we really need this? Is there an ROI for this? Obviously, every startup is different and their approach is different, but this is one of the most important things. And then we also work with healthcare service businesses where we help them improve their operations through AI and technology, and that’s where the scaling comes in.

[00:03:55] Ghazenfer Mansoor: You build those internal tools to bring efficiency to their processes. As you’re building, every process that you improve, every bottleneck you remove, it reduces mistakes, it improves efficiency, it brings more smoothness to your process. So gradually, your valuation increases.

[00:04:20] Ghazenfer Mansoor: It helps everywhere in terms of growing the business.

[00:04:26] Allyson Maur: Got it. That’s really insightful. As you continue to build a diverse portfolio of healthcare solutions, you’re building everything from telehealth to AI-driven analytics tools. Each of those areas has its own operational, clinical, and regulatory complexities. When you’re evaluating opportunities, how much of your decision is driven by the technical challenge versus the underlying business model?

[00:04:48] Allyson Maur: Things like reimbursement, regulatory pathway, or go-to-market feasibility. How do you decide which types of healthcare problems are worth solving and where tech can have the greatest impact?

[00:05:00] Ghazenfer Mansoor: It’s mostly the business problem, because technology is meant to serve people. Technology is meant to serve businesses.

[00:05:08] Ghazenfer Mansoor: So just building technology, if it doesn’t solve a business problem, it’s not going to have a big ROI. So, the key part is: how do you look at your business? What processes can you improve? Where can the bottlenecks be improved? What problems are you solving? Because as you’re solving those problems, that’s where it creates excitement, that’s where it creates profitability, and everything else.

[00:05:35] Ghazenfer Mansoor: As these businesses grow and the tech is solving the business problem, that’s where you get more customers. You increase your valuation. That’s a bigger part that we have seen when we work with those service businesses, because usually, especially for service businesses, the valuations are not that high.

[00:05:54] Ghazenfer Mansoor: But as you build that tech, it not only brings efficiency, but you’re also creating your intellectual property. And that intellectual property helps you increase your valuation as well. So not only do you increase your profitability, but most of the time the ROI is there right during the time you’re building, because as you move to system-powered growth rather than people-powered growth, the number of people needed to support those projects or that business isn’t growing as fast as the tech you’re building.

[00:06:29] Ghazenfer Mansoor: So obviously that not only improves profitability, but it also increases valuation, and now you’re gradually becoming a tech company rather than just a service business.

[00:06:40] Kristen McDermott: So, when you’re working with early-stage healthcare companies versus more established systems that have more resources and more personnel, how do their needs differ in terms of technology infrastructure and innovation priorities?

[00:06:58] Ghazenfer Mansoor: Okay, so startups usually need speed and flexibility to find product-market fit. They usually start with one problem and then figure it out. And along the way, as they’re building those products, they’re figuring out the product-market fit and gradually expanding. That’s how most startups are.

[00:07:18] Ghazenfer Mansoor: Enterprises are usually different. They are already solving certain problems, and their needs may be more on the integration side, more on compliance and optimization of their existing systems, where the tech is mostly solving an existing problem, but maybe for a broader set of customers.

[00:07:42] Ghazenfer Mansoor: Because enterprises are mostly building things for themselves, not just for one customer. That’s a big difference.

[00:07:49] Kristen McDermott: Yeah, that’s a keen observation. So, I want to dig a little more deeply into the types of solutions you’re building. I know there’s been an explosion of digital health tools over the past, say, six years.

[00:08:04] Kristen McDermott: What are one or two examples of solutions you’ve seen, either ones you’ve built or ones that are on the market, that are actually delivering measurable ROI or clinical impact?

[00:08:17] Ghazenfer Mansoor: So, the bigger ones nowadays are AI-powered process automation: all these clinical workflows, whether it’s scheduling, your answering machine, your document processing.

[00:08:29] Ghazenfer Mansoor: Document processing is huge, whether it’s an explanation of benefits or invoices or any documents in your business. Those automations are critical, and those are the ones we see a lot more demand for. Every business is different, so gradually we’re seeing more and more demand for those.

[00:08:49] Ghazenfer Mansoor: The second part we see a lot is AI-powered prediction, because AI is much simpler now than it used to be. So, you can do a lot more. Companies that have data can now do a lot more predictions based on that data. And then, based on that analysis and insight, you can build a lot more different systems.

[00:09:15] Ghazenfer Mansoor: So, we’re seeing a lot more demand for that, especially on the personal health side. For example, you load your personal health data, could be your blood work, could be any of your reports, and then you can get more insights on that data.

[00:09:34] Allyson Maur: Yeah, I think one of the things we’re seeing in this space is the tech is strong.

[00:09:39] Allyson Maur: It’s getting stronger, it’s getting faster, it’s getting better, it’s getting more accurate. But adoption seems to be the biggest hurdle, especially on the patient side. So, when you think about applications that are actually gaining traction and getting adopted, what design principles or features tend to drive real engagement and sustained use?

[00:10:02] Ghazenfer Mansoor: So, I think one of our areas of focus, and one of our recommendations, is always creating something really simple and easy to use. User experience is key. No matter how big a problem you’re solving, if it’s not easy to use for the user, you’re going to have a tough time getting adoption. Yes, sometimes users are forced to use it, but that’s not exciting.

[00:10:26] Ghazenfer Mansoor: You want it to be really simple. Then secondly, immediate value: it has to provide some value, it has to increase productivity for what you are doing. And I think another important one is personalization. Is it working for your specific case, or is it a broad solution for everybody?

[00:10:47] Ghazenfer Mansoor: I think as we work on these applications, if you have these things, simplicity, real value, and personalization, that makes a huge difference in terms of how users adapt to these applications. So lastly, I would say one thing: if it doesn’t fit into somebody’s day-to-day, it won’t get used.

[00:11:12] Ghazenfer Mansoor: So, you have to make it simple enough that people use it on a daily basis.

[00:11:17] Allyson Maur: Has to be easy enough to use, and you want your users to want to use it and to see the value in using it. On the provider side, we hear constantly about burnout and workflow challenges. Doctors don’t love adopting new tech.

[00:11:32] Allyson Maur: New tech can either help or make things worse. So how do you approach designing solutions that integrate seamlessly into clinical workflows rather than adding friction?

[00:11:43] Ghazenfer Mansoor: Yeah, so the important part is you build the technology around your business workflow, not the other way around. Many times, we see people build the technology and then teach people how to use it to solve a problem, and that adds a different kind of complexity.

[00:11:59] Ghazenfer Mansoor: Users are familiar with their business, so the terminology you use matters, as I was talking about with personalization. So you build something people can easily use: simplicity, and it has to be built around the workflow you’re used to, not around the technology.

[00:12:19] Ghazenfer Mansoor: The technology should feel invisible. So again, coming back to it, you don’t want to be building something and teaching people complex software. I think those are the older days.

[00:12:34] Allyson Maur: No square peg, round hole. And we do see some providers or institutions get excited about AI and try to adopt all the tools without understanding how they work together, or they don’t work together, and you get a bit of AI and tech fatigue, I think.

[00:12:52] Allyson Maur: Kristen, you want to talk a little more about AI?

[00:12:54] Kristen McDermott: Let’s focus a bit more on AI. It’s definitely one of the most exciting and complex areas in healthcare these days, and there’s a real difference between tools like RPA, which you mentioned earlier, and some of the large language model-driven or generative AI tools we’re seeing.

[00:13:16] Kristen McDermott: What do you think about the different use cases and risk profiles across that spectrum, and how does that shape your approach to compliance and deployment?

[00:13:26] Ghazenfer Mansoor: Yeah, that’s a good question. Compliance is critical in healthcare, so every application needs to be HIPAA compliant. I think as we design the system, we design the system, not just the models.

[00:13:42] Ghazenfer Mansoor: Each LLM has its own way of doing things, but as you’re building the system, it needs guardrails. In this case, that’s HIPAA compliance: which data needs to be given to the LLM. You can’t just give everything to the LLM and expect it won’t be used to train the model, because that brings risk.

[00:14:08] Ghazenfer Mansoor: There is some internal data that needs to be taken care of. There are different strategies that can be used, and that’s what we see in many applications where you have PHI data and you’re using AI. That comes with its own risk as well, so you have to be careful in terms of how you’re building, knowing the technology.

[00:14:27] Ghazenfer Mansoor: There’s a specific architecture, a specific way of doing it. This is on top of getting the LLMs to sign the BAAs. Once you have those things in place, and you’re building the tech the right way, you’re not only building the system, but you’re also compliant. So, building a system that’s compliant, obviously, then you have to validate that.

[00:14:51] Ghazenfer Mansoor: Does it work in your specific business, in your workflow? How does that work? And then the most important part that we see people miss is human oversight, because AI isn’t just about giving it some data and getting output. There’s a lot of hallucination, a lot of mistakes, so you still need human oversight, and those workflows are important.

[00:15:18] Ghazenfer Mansoor: So, as you’re building the system, we have to make sure we address all of that: compliance, validation, and human oversight.

[00:15:29] Allyson Maur: Yeah, Kristen is tired of hearing me say this, but garbage in, garbage out has never been truer than with AI, I think. We’re also hearing a lot about shadow AI, where providers or staff are using AI tools outside of their approved systems.

[00:15:42] Allyson Maur: Are your clients thinking about that risk? And if so, is it influencing how you design and position your solutions? Are you thinking about things like intended use, and how you promote the product versus how other providers or users may actually be using it in a way that wasn’t intended?

[00:16:02] Ghazenfer Mansoor: So that brings up a good question about governance, because obviously you have to be compliant. So, if people are using different systems, that means you have to follow certain regulations, and that depends on the organization. If people are using systems that aren’t approved, that’s a big risk.

[00:16:23] Ghazenfer Mansoor: So that needs to be handled through governance and regulation. When we build those systems, we’re only working with LLMs that sign the BAA and follow certain processes, because once these vendors sign BAAs, they are committing to zero data retention. The data is not being used to train the LLMs.

[00:16:47] Ghazenfer Mansoor: So, anything outside of that just increases risk. If we’re uploading any data to any AI, let’s say people are using ChatGPT or Claude for internal use and they’re uploading data, that’s not HIPAA compliant already. You’re creating a big risk.

[00:17:08] Kristen McDermott: I think that segues nicely into our next question.

[00:17:11] Kristen McDermott: One of the defining characteristics of healthcare technology is the regulatory environment. It’s complex, it’s evolving, and it carries real consequences. As you’re building solutions in this space, at what point in the development cycle are your clients really engaging on compliance? Are they thinking about HIPAA, data use, and security architecture from day one, or does that tend to come along later when they’re scaling or contracting with health systems?

[00:17:42] Ghazenfer Mansoor: We recommend that compliance be embedded in the system from day one. Anything you add later on adds more complexity. And yes, that does come up as a common question: can we add that later on? Many times, the system is already built, and now you want to add compliance. That means you’re going to change the foundation of your project.

[00:18:06] Ghazenfer Mansoor: That changes a lot. So, compliance has to be there from day one. In fact, people feel like if we have compliance upfront, it’s going to slow things down. No, it’s actually the opposite, because a compliant system will be much faster to build, since you’re taking care of that part from day one rather than patching it later on.

[00:18:28] Ghazenfer Mansoor: Because as you’re patching it, it adds a lot more complexity. So, compliance doesn’t slow innovation; it’s unclear compliance that does.

[00:18:40] Allyson Maur: So, with patient portals that implement end-to-end encrypted messaging between patients and healthcare providers, do you see patient engagement tools becoming more central to care delivery?

[00:18:51] Allyson Maur: And if so, how do you see expectations around security evolving? It sounds like you’re thinking security is going to evolve even more significantly.

[00:18:59] Ghazenfer Mansoor: Yeah, so when it comes to HIPAA, it’s not really just end-to-end encryption. Obviously, encryption is a big part of it, but now you also have these devices.

[00:19:10] Ghazenfer Mansoor: You have mobile devices where data is stored locally. You have browser caching as well. And then you have a server. So, you have to make sure that if you lose any of those devices, it follows certain regulations. And more importantly, authentication: who looked at the data?

[00:19:30] Ghazenfer Mansoor: At what time? For how long? What data was looked at? Audit logging. All of those things are an important part of compliance. So, it’s not really just encryption; knowing the HIPAA compliance regulations makes a difference in how you’re building these applications. And then, of course, there’s the additional part we already talked about: AI.

[00:19:50] Ghazenfer Mansoor: So now the data is also touching the AI, so how do you make sure any PHI data that goes into those systems is compliant? And lastly, this is another big part that many companies miss: integration with external systems. Yes, your own application, you feel like it’s HIPAA compliant.

[00:20:09] Ghazenfer Mansoor: But if it is integrated with twenty other different applications, using third-party software, whether it’s plug-ins or any of those, are they also storing or exchanging data? Are they following HIPAA compliance? Whether it’s an analytics tool, push notifications, or any other system that touches PHI data, it needs to be compliant, and that’s the part many people miss.

[00:20:40] Allyson Maur: I think that’s a great point that I don’t think a lot of folks who are developing apps and products are thinking about, that it’s not just vendor management, it’s integration partner management and diligence. You really need to know who you’re developing your app to integrate with, or developing your product to integrate with, and what they’re going to do with the data and what they’re doing with the people who are also accessing the data. Sort of the downstream effects of the subcontractor to the subcontractor.

[00:21:08] Allyson Maur: And how we’re actually vetting the security of all of those players within the integration chain.

[00:21:16] Kristen McDermott: Well said. I think those were great points and a nice recap, Allison. So, looking to the future, there’s clearly no shortage of innovation happening across the healthcare landscape. Over the next five years, what’s one shift in health tech that you think will truly get adopted at scale, and what will separate the solutions that make it from those that don’t?

[00:21:43] Ghazenfer Mansoor: I think the way technology is moving, there’ll be a lot more that changes in the next five years. A couple of things I’ll look at: predictive health, there’ll be a lot more predictive health solutions. And the second most important part is agentic AI, because as you’re building these different agents to solve day-to-day clinical workflow problems, agentic AI is going to make a big impact on how healthcare moves forward.

[00:22:16] Ghazenfer Mansoor: Because it’s not only about following certain rules and doing some agent work, but also making decisions on behalf of providers, patients, and users. So, these two are definitely the two big ones.

[00:22:35] Allyson Maur: Got it. All right, Ghazenfer, we really appreciate your time. Before we wrap up, I want to give you a couple minutes to talk about your new book, Beyond the Download: How to Build Mobile Apps That People Love, Use, and Share Every Day. I absolutely love the title, especially in healthcare, where adoption and engagement are the hardest parts.

[00:22:56] Allyson Maur: So, what inspired you to write this book, and why now, and where did you find the time to do it with everything else you’re developing?

[00:23:06] Ghazenfer Mansoor: Yeah, so I’ve been involved in mobile for a long time, even before iPhone and Android came out, since 2000, when it was just web-based, WML-based devices, much smaller.

[00:23:18] Ghazenfer Mansoor: Because of my involvement, I got more and more into mobile over a long time. Over the years, I built so many different applications. I’ve seen how these apps got adopted, some failed, some succeeded, all different types of stories and patterns. This book is about sharing those experiences: why some apps just got downloaded and never got traction, versus the ones that do get traction and get used on a daily basis.

[00:23:51] Ghazenfer Mansoor: So, this book is definitely about sharing my experiences building mobile apps and growing mobile apps. My goal is that more and more people who are building mobile apps build apps that aren’t just downloaded, but also get used. I want to help founders build products that people actually stick to.

[00:24:16] Kristen McDermott: For our listeners, many of whom are operators, investors, and healthcare leaders, what’s one practical takeaway from the book they could apply immediately to improve engagement with their own digital health tools?

[00:24:30] Ghazenfer Mansoor: There are many. The book contains about 32 different chapters talking about different strategies, but if I have to pick one or two, I would pick the user experience part.

[00:24:43] Ghazenfer Mansoor: You want to build an app that is remarkable, because the first experience is always the key experience. Usually, people decide in the first few minutes whether they want to keep using an app. So that first impression is the key impression. Obviously, other things like tracking do make a difference later on.

[00:25:07] Ghazenfer Mansoor: But if you don’t have a slick design and an amazing user experience, you’re not going to get adoption. You want to create those hooks so that people want to stay on it, they feel excited, and they want to keep coming back to the app.

[00:25:27] Allyson Maur: Ghazenfer Mansoor, CEO of Technology Rivers, we want to thank you so much for joining us on Health Plus Tech.

[00:25:34] Allyson Maur: This has been a fantastic conversation. We are absolutely thrilled that you joined us. Happy to hear about your book and your successes, and that you’re sharing your advice and experiences with others. I want to thank my co-host, Allison Mauer, and thanks for joining us, all of our listeners.