#512 From Mainframes to Cloud: Rob Duffy on Reinventing Healthcare Systems

In this episode of The CTO Show with Mehmet, I sit down with Rob Duffy, CTO of HealthEdge, to unpack how healthcare—one of the last industries to modernize—is finally undergoing a deep digital transformation. Rob shares how his journey through companies like Amazon, Kindle, Time Inc., and Expedia prepared him for leading one of the most complex transitions in technology: taking healthcare systems from legacy mainframes to cloud-native platforms.
We discuss the hype vs. reality of AI, what true ROI looks like, and why cultural change is the hardest part of transformation. Whether you’re a tech leader, investor, or founder, Rob’s perspective sheds light on how to drive meaningful change in regulated and lagging industries.
Key Takeaways
• Why healthcare has lagged in digital transformation and how SaaS is shifting the landscape
• The real challenges of replacing mainframes and legacy systems in regulated industries
• AI in practice: from demos and hype to real ROI and human-hour savings
• Building an AI-first culture and making adoption stick inside organizations
• APIs, interoperability, and the future of system integration
• Career advice for aspiring CTOs: why learning the business matters as much as tech
What You’ll Learn
• How to navigate transformation in industries with entrenched legacy systems
• The single most important metric for measuring AI impact
• Why cultural adoption is the true unlock for enterprise AI
• How APIs and interoperability shape the future of healthcare and beyond
About the Guest
Robert Duffy is an accomplished technology leader with an extensive background in product development and engineering. Rob previously served as the Chief Product and Technology Officer at Drizly, an Uber Company, where he played a pivotal role in scaling the company’s product and engineering teams post-acquisition. His leadership at Drizly was instrumental in driving innovation within Uber Eats' grocery delivery services, showcasing his ability to merge technology with user-centric solutions. Prior to his tenure at Drizly, Robert held significant positions at industry giants including Salesforce.com, Amazon, and Time Inc.
At Salesforce, Rob excelled as the Vice President of Software Engineering, leading the team responsible for the Lightning Web Stack, which handles billions of API calls per day. At Time Inc, he served as VP of Engineering for its video and content platform that powered more than 300 content apps for 91 media brands with 139 million unique monthly users. At Amzaon, he was Head of Strategic Initiatives for the Kindle Store Platform, the content experience powering hundreds of millions of e-commerce transactions on the digital stores (books, audio books, video, and mp3) across all of Amazon's devices (fire tablet, kindle, echo, fire TV, and mobile apps).
https://www.linkedin.com/in/platformduffy/
Episode Highlights
• [02:30] Why healthcare is the last major industry to transform
• [06:45] Mainframes, legacy systems, and the cost of change
• [12:00] AI hype vs. real-world ROI examples
• [20:00] Building an AI-first culture and driving adoption
• [27:30] Governance, security, and evaluating new AI tools
• [33:00] Pilot programs and scaling cultural change
• [38:00] AI’s role in cloud optimization and legacy refactoring
• [44:00] APIs as the next big unlock for enterprise systems
• [46:30] Rob’s advice for aspiring CTOs: learn the business first
[00:00:00]
Mehmet: Hello, and welcome back to a new episode of the CTO Show with MeMed today. I'm very pleased joining me from the US from New York, Rob Duffy, CTO of HealthEdge. Rob, thank you very much for being here with me today. Uh, I appreciate the time, I know how busy it can get. [00:01:00] But before we start, what I like to do is I like to keep it to my guests to introduce themselves.
So tell us more about, you know, your, your journey, your experience, and what you're currently up to, and then we can start the discussion from there.
Rob: Amazing. Thank you for having me. Really appreciate, uh, really appreciate having me on the show and, uh, nice to, uh, you know, be here and get a chance to talk to your audience.
Uh, so my name's Rob Duffy. I work for a company called HealthEdge and, uh, HealthEdge is a software as a service, uh, uh, uh, uh, insurance core administrative processing system provider. So if you're in the us obviously we have a, a paid healthcare. Uh, industry here, uh, many people consume healthcare and then the bill is paid partly by the insurance company and, and partly by them.
And the whole process of, of, uh, collecting the claim, adjudicating it, uh, you know, uh, making sure that it's coded and billed correctly. Figuring out how much the plan says the consumer plays and how much plan says the, uh, health insurance payer plays. [00:02:00] That whole, uh, process is, is very, very complex. It involves, uh, you know, hundreds of thousands of rules and, and we provide a software as a service platform that enables health insurance companies to do that, uh, or like automatically.
And normally we're replacing very old, antiquated systems like. Uh, name frames and we're putting this, uh, piece of technology into our customers. And if, if you can imagine doing that, uh, our customers systems and processes and people and technology, the entire ecosystem is sort of, uh, built around the old system.
When we come in with a new system, it really is a, a big transformation for our customers, not only from a technology perspective, but also from a people perspective and a process perspective. Uh, and doing so we really kind of digitally transform a lot of the health insurance companies that are on our platform.
Uh, and that leads to a much better consumer experience, uh, in health insurance, which has a very bad rap in the, in the US in general, from its poor consumer experience. So I'm excited that we're, we're providing better experiences to the [00:03:00] consumers of health insurance in the United States.
Mehmet: Great. Uh, we have a lot to talk about, Rob, because you mentioned mainframes, something which I didn't prepare, but I have to ask a few things.
Um, but before this, like, uh, what, what drove you to, to have this career and becoming A-A-A-C-T-O for a leading SaaS company? I'm always curious to know the stories behind, uh, what get you there?
Rob: So, my background, you know, I, I started my career. As a software developer, uh, and I've always, always joined companies that are kind of in a, in an industry inflection point.
So I, I, uh, early on in my career, I joined Amazon at the point where e-commerce was really replacing, you know, physical retail. From there, I moved to the Kindle organization at the point where, uh, e-books, uh, were replacing or overtaking, um, uh, uh, physical books. I left, I joined a media company, time Inc. Uh, at a point in the, the sort of media landscape where digital media was overtaking, uh, print media from, from a revenue [00:04:00] perspective.
Uh, I, I left and, uh, went to Expedia group, uh, when travel had shut down, uh, 'cause of COVID, no one was traveling. Uh, and then sort of brought travel back to the world, uh, through Expedia. Uh, and you know, at all these points in, in my career, I've really joined companies that are in this industry inflection point.
And healthcare is really one of the last, if not, if not the last industry to go through a digital transformation. And it's an exciting place to be. And, you know, I, I joined HealthEdge because. We're really leading that, uh, digital transformation in healthcare. We're one of the, you know, most successful, largest core administrative processing systems.
Uh, and with the, the, you know, the most, uh, cloud native of all of our competitors, and we're, we're bringing that expertise and that transformation to an industry that, um, has, has lagged behind in terms of its transformation for, for many, many years.
Mehmet: Absolutely. Rob, you mentioned something, uh, I told you about the mainframes and, you know, healthcare.
Now [00:05:00] I know the answer because I, I asked a lot of CTOs and a lot of CIOs in, in, in healthcare space, and I know, you know, what held them back. But from your point of view, why some verticals, especially healthcare, look like little bit behind the journey. So was it. Just, you know, because the systems were designed in a way, it's hard for us to take them to the next level.
Is it because of all the privacy, security, um, and, and other, you know, regulations that we have around the sensitive verticals. What caused this little bit slow, uh, adoption of digi new digital technologies for mainly healthcare, let's say.
Rob: I think if you think about it, you know, like finance is another good analog, right?
Mm-hmm. Like you have finance and healthcare are, are probably the, the, the last on the sort of transformation curve, which is kind of interesting. You know, you mentioned ai, we can talk [00:06:00] about this a little bit later because if you look at the history of cellular telephony. You know, th uh, third world countries skipped, uh, hardwired landline telephony and went straight to cellular telephony.
So like they, they are actually further ahead in some things like digital payments. And, you know, I think this is a similar, similar thing happening now, like in, in some of these older, uh, companies that haven't gone through a digital transformation, they're experiencing their sort of third world country cellular moment where they're skipping all of the steps that they need to do to transform and kind of looking ahead to an AI future.
But, um, we'll save that for, we'll save that for later. I think the question was, you know, why do I think that healthcare in particular is, has taken such a long time to, to go through its own digital transformation? And, you know, if you, if you think about the ecosystem that you have inside of a, um, you know, a healthcare company like, and it's specifically a health insurance company that, that are customers, the systems and the processes have been built up over, over a long, long time.
You know, they started, uh, on mainframe systems and each of those mainframe systems provided a, you know, a bunch of. Uh, [00:07:00] core processing and then other systems were kind of bolted on. And then they have to get information from all of these, you know, intermediaries and third parties. And there's billing and there's enrollment, and then there's like all different, different companies that provide different, different areas of capability within the company.
And what that means is it's just every things are everywhere. Like, and the data is everywhere. You know, in some cases data is in multiple different places and it's not always hard to tell which is the, the source of truth or which is correct. And sometimes they might differ. You just have this, this huge, um, you know, uh, ball of rubber bands.
That is, that is the, the central nervous system of, of a lot of companies. And that's very, very hard to replace. And it's, it's, it's very costly, right? Because you have teams that are put in place to support those things. You have processes that are put in place around the core capabilities of this entire ecosystem.
And sometimes you have cap, you know, you have processes that are put in place to work around the fact that that core system doesn't have capabilities. And, and you [00:08:00] know, in some cases you have technology that you've built to work around something that, uh, doesn't exist in a platform or does exist in a platform and is done in a strange way.
So the whole, the whole organization. Is built around this system, and if you start replacing parts of it and you start transforming parts of it, it is a huge initiative. You have to rethink the technology. You have to rethink the people, you have to rethink the processes. You have to, you know, restructure your teams.
You have to really change the organization. And I think no one, uh, you know, who gets to the level of like CIO or CTO. Looks at that project and is like, that's gonna be easy and low risk, and I have a high degree of, you know, confidence that I'm gonna succeed. Right. So you, you have to, uh, you have to find leaders that are willing to take a risk and you know, that believe in the future and understand and can sell it to their constituents.
And, and when you do that, then you can sort of transform. But a lot of times it's just, you [00:09:00] know, the, the, the, so the cost and the risk of some of these projects can be large and, uh, you know, they haven't in many cases. Uh, some of the alternatives, uh, outside of our, you know, cloud native software platform, um, are being very, very large and cumbersome and hard to put in place.
And like, you know, it's taken a lot of time and energy and effort to put them in, install them. Um, and the buying cycle then becomes 10, 15 years. You know, you buy something that's in there for 10, 15 years. So with, with our platform, we try and make it a lot easier to do that transformation and we acknowledge that installing this isn't just like, you know, you go, you click go, go go, you know, put in a license key and all of a sudden everyone's on a new core administrative processing system.
Right? Like it really is a, a transformation and we help our customers go through that transformation.
Mehmet: Cool. Now. I am, I am not sure if you will agree with B Rob, uh, the term transformation itself. So, you know, at least like we're taking something from state A to state B, although like it's [00:10:00] something which should be continuously, uh, enhanced now.
I remember I'm, you know, uh, I'm not sure about like which year exactly the term started to appear, but at least I can say 12, 13 years since we are hearing about digital transformation now, you men, we were mentioning about digital transformation. There are like some legacy systems still in place in majority of, especially maybe in in insurance, health, healthcare and so on.
But now we have this wave of, and you just mentioned, and it's good time to speak about it. Ai, right? Yeah. So now we have the AI transformation coming. Um, now I know when I was preparing, uh, so from the information the team gave me and, you know, some, some of the research I did, um, so we see people talking in antitheism about ai.
Right, but publicly, but behind the doors, some completely different conversations take [00:11:00] place. What do you think is this gap coming from? Is it because we didn't complete the digital transformation? Is it because we didn't start in a proper way? Is it still because we have these legacy systems in place?
What's your point? Why? Why? You mentioned this, why, why you are saying there are a lot of talks behind the doors about ai. Uh, I think
Rob: that, um, AI is, is, uh, a very, you know, new technology in the, in the business setting, right? Like, or that's not, that's not bad. Large language models are, you know, we sort of conflate AI and large language models, large language models, and the things that we can do with large language models are a fairly.
New technology within the enterprise. And with any new technology there comes, you know, a sort of fear or uncertainty, doubt. It becomes, becomes a lot of enthusiasm. You know, there's a lot of, uh, not knowing about it, like understanding it at a very high level, but not knowing the, the details of it. And then there's also just sort of, you know, fear and trepidation [00:12:00] that it might be another.
While Goose chase, right? And in tech we've had a few while goose chases like, you know, the metaverse and you know, like blockchain blockchain's gonna be everywhere. And, um, I think some, there's a lot of trepidation around just going all in on AI and then it turning out to be something that, um, you know, ends up slowing you down eventually.
Or it was just a waste of time or didn't deliver the results. And there's also, you know, a lot of, a lot of demos and a lot of, uh, you know, really good. Sort of quick hit one shot use case, uh, videos and talks and things at conferences where people look at 'em and they're like, that's great. And, and then there's not a lot of real hard evidence.
And, you know, people talking about we eliminated, you know, 40% of this work. I think one of the best use cases out there that does demonstrate the power is, you know, AWS. Um, talks about leveraging Amazon queue for doing Java upgrades. And, you know, they eliminated, uh, I can't even remember [00:13:00] what it was, but it was an, an astronomical amount of hours.
Something like, you know, 400,000, 300,000, 300,000 hours of work using Queue Developer to do this Java upgrade. Right? And it's like, you think about that and you're like, okay. Compared that a Java upgrade, which is probably the most boring and mundane thing you can possibly do to some of the demos you see, which are like, look, this is gonna completely revolutionize the the software development lifecycle, and I have 50 agents, you know, writing all this code and doing these code reviews and all this sort of stuff.
So there's this sort of like gap between the actual real hard. Results real hard. ROI, outcomes and the demos. And I, I think a little bit like, I, I think, uh, about this a little bit like robotics, right? Like when, when humanoid robots were, you know, uh, shown in science fiction, they were kind of like. Running around the house and doing things and, you know, uh, feeding your kids in the morning and giving them breakfast.
And then I think about the largest deployment of robotics in the consumer world, that it's like a small, uh, [00:14:00] thing that runs around your kitchen and, and cleans up crumbs, right? Like, it's like, it's not feeding your, feeding your kids in the morning or putting their shoes on or getting 'em ready for school.
Something that at night when you sleep runs around your house and cleans up some crumbs. And I, and I think that difference between. You know, the promise of robotics, uh, and the reality of robotics is the same. Uh, when we think about LLMs and the deployment of, of what they can do within the enterprise, and it really is like, there's the demos which are up here, you know, with, uh, thousands of agents running around, tidying up your, your, you know, your company and like, you know, marketing teams completely be replaced by, uh, you know, LLM agents.
And then the reality of it is. You know, actually the LMS are doing Java upgrades and, you know, writing unit tests and documentation and doing things that like really are the, those lower level tasks. And I think too many people focus up here on the, like the, the art of the possible. Um, and I, you know, I think as a, as a as technology leadership organization, we need to sort of think hard about how can we [00:15:00] take work from humans and migrate it to AI and agent systems.
But that work is probably in all likelihood gonna be. More boring, more mundane things that humans don't really want it do.
Mehmet: So I I get that. Rob, I, I agree with you. Um, you know, I, I used to be just, you know. Working on the technical side of the thing since I started the podcast, so I have to also see what other people talk about, what they, how they speak, how they present things.
And I agree with you Sometimes, you know, we see these cool videos and demonstrations of things, as you said, like agents working in the background about on, on something now. At the same time this is happening and you know, maybe you can confirm or maybe you have another point of view on this. So. In a company, on a uh, board level, maybe someone on the board watches some of these videos or demonstration or maybe one of his friends or her [00:16:00] friends, they see these things and they come to the board and say, Hey guys, this is this new AI thing that we have to bring.
And yeah, it is doing superb job like helping us. And then we take this and try to implement it in the company from the way up. Instead, which I think should happen is to look at the bottom, see what needs to to happen, and then try to build use cases where AI can fit. Are we seeing the more of the first one there, Rob, in your opinion?
Where just we are just trying to adopt a hype, or I'm not called it a hype. There's something called ai, which exists, exists for a long time. But I mean, the LLMs and the tools we are seeing, the co-pilots and the others. So, so do you think like now executives are trying to bring this as fast as possible because also they are afraid that they would go out of business because we, we are seeing also these, uh, analysts.
Talking about like how business are being disrupted currently because of AI and how jobs gonna be, uh, [00:17:00] different because of ai. So what's your point of view on this?
Rob: I mean, I think if you, you know, if you're taking technical direction from the board, there's probably a little bit of a broken relationship there.
But I, I just don't wanna say like, sometimes you do get suggestions on, on technology. Um, I think that, um, you know, I, I. I think all leaders, all CIOs, all CTOs, all you know, VPs, PS of, of engineering should be thinking about how they can deploy, um, LMS and artificial intelligence and agent workflows, uh, and leverage it through their, through their engineering teams and through their software development lifecycle.
I think there's a huge amount of um, uh, tasks, time reduction. That we can provide to our, our businesses as technology leaders by deploying lms. And you know, if you think about the software development lifecycle, everything from writing better PRDs to writing better user stories with better acceptance criteria to.[00:18:00]
You know, creating, uh, you know, scaffolding code to writing unit tests, writing integrations, all these tasks that take up a tremendous amount of time for our engineering teams can be accelerated using LLMs. And I think if you can demonstrate that and if you can instill a culture of thinking about. How can I migrate some of this work from the humans that we currently employ today to AI energetic systems?
Then you, then you are leading that conversation and you're not, like, no one is gonna come to you from, you know, a board perspective or, or you know, um, your CEO and say, Hey, go. We should try this technology. Because you can say, well actually we've got a fairly. Strategy here, and, you know, we're already trying these tools and here's some areas that we're getting leveraged.
And I, I don't think that, um, I don't think that there's gonna be a, a tool, a single tool here in this space that it's gonna be like, oh, I deployed this tool and, you know, I got a, a huge amount of savings across the company. I think the biggest change is gonna come from a cultural transformation where you get [00:19:00] everyone thinking about how they can.
Um, leverage AI and leverage agent systems, and you train people and, and, you know, you take your teams through, um, you know, boot camps and you, you get everyone thinking about AI first. And like, I think if you can do that, um, you know, that's where the real benefit and the real value's gonna come from. And that's probably gonna require more than one tool.
It's probably gonna require 20 or 30 different tools. If you think about, you know, the revolution of computing. You know, yes, you had a computer, but you know, you have, uh, maybe 40 or 50 different applications that you interact with every day to, to do your job effectively. And you know, I think the same is true for lms.
Like, yes, we're gonna use LMS and ai, but we're probably gonna use. 20 to 30 different tools and different agents and different things to get our job done. So I, I don't think there's one thing that's gonna, you know, come from, uh, come from on hire and, uh, you know, deliver tremendous amounts of transformation.
I think that the transformation's gonna come from a cultural change.
Mehmet: Great. And on the [00:20:00] cultural change, Rob, on this point specifically, um, so it, it's, it's kind of. Imperative for every leader to, to try to bring this cultural change there. Now, back to the point how healthcare is different. How easy is to do this cultural change in healthcare?
Rob: Uh, I think that, um, you know, if there is an, if in a industry like, uh, healthcare and an industry and a, and a, a company like a health insurance payer. Or health, the health plan. You know, cul the cultural change has to be leading to an outcome, right? Like, it can't just be like, Hey, we need to change culture.
We need to get a better ai, I think really needs to be like we're driving an outcome here. And you know, the, the industry as a whole is very pressured on, on margins. It's very pressured on, um, you know, reducing the, uh, [00:21:00] or increasing the efficiency and, you know, reducing internal costs. And I think. For me, you can, uh, have a conversation with a lot of health plans around, we are gonna reduce your operating costs.
And like at the end of the day, if you are, if you're having a conversation with a customer about reducing operating costs, you know, in some cases. Uh, it, it doesn't really matter whether you're doing that with AI or you're doing that with, you know, a, a software as a service, or you're doing that with a rules engine or you're, you know, it, it, the outcome is what matters.
And pushing on that reduction of operating costs and saying, Hey, we've gonna reduce operating costs, but we're gonna do it 10 times better than we were able to do it before, because we're gonna leverage ai. Then you have a lot of momentum behind that transformation because you can start talking about real hard outcomes and real business outcomes, and people start to get really excited about that.
And if you try and push a transformation without, as high to a real business outcome, you know, a real operating cost reduction or you [00:22:00] know, real increase in top nine sales, then it, it's very hard. It's very hard to drive that change. It's, you can't just have that living out there. Um, saying we have to transform, and someone says, why?
And it's like, well, everyone else is transforming. It has to be, you know, like it transformation has to be a a tool. AI has to be a tool, and the outcome that you're driving to has to be operating efficiency for your customers.
Mehmet: Absolutely. I think the trap, and maybe you would agree with me, Rob, and I know you said this, uh, uh, before, so, so the trap is we try to compare ourselves to someone else and think saying like, we achieved like this amount, percentage, uh, of getting like faster code or whatever.
Now you mentioned. In the business outcome about reducing the operational cost. Any other metrics that you think CTOs, even CIOs they need to focus on? This is really what they should be looking at, um, when doing. Um. Not only for the ai, actually, it, it applies to any [00:23:00] change that the, that they're planning to introduce.
So what are, like, in your opinion, the key metrics we should be focusing on as leaders? I honestly,
Rob: like, honestly, I think the only metric that you need to focus on is the reduction of human hours. Like, that's it, right? Humans are incredibly expensive. Uh, and, you know, AI energetic systems are, are, are. Three, four orders of magnitude cheaper to do tasks that, uh, humans, uh, are doing.
Um, they, they can't do some tasks, you know, they, they are terrible and they still need human review. But when you think about that example I gave you of a Java upgrade and how many hours were, were saved there, um, at Amazon, like, it's exactly, it, it, it's exactly that. You're just reducing the amount of, of human hours worked.
And honestly, I think that's really the only metric that we need to think about, um, in terms of the outcomes. Now, obviously there's control metrics that we have to think about and we have to look at things like, you know, the quality and the, you know, the. The cost of the agentic systems, you have to look at how much supervisory time you need to, to put on [00:24:00] top of that system.
You need to look at, you know, some of the, um, you know, non-functional requirements around it. But, and those are all sort of like input metrics, but the real output of, uh, any AI project should be a reduction in, in operating costs and, and a reduction in human time spent completing a task.
Mehmet: Right now. Let's talk about.
You know, a little bit, uh, and all the things related to data security, which is mainly risk privacy, and data integrity. Yeah. Um, what do you think are the frameworks or approaches that. Are proven effective. Right. To maintaining all the above to make sure, especially because we talk about healthcare insurance, so people's data are there.
Of course, we have the regulations. I know in the US you know, the, the, the HIPAA mainly is, is the main frame to, uh, or uh, framework to, to, to follow. [00:25:00] But. As a CT or CIO, or CO, maybe even. So how we should be, uh, preparing ourselves, especially in the age of ai, because AI means data, right? So when we talk ai, we're talking about data that we need to feed the labs with.
So what have you seen? Uh, working well.
Rob: So we have a, uh, you know, whenever we are looking to put leverage a, a new piece of technology, um, or, you know, a new AI system, we have a, a fairly robust evaluation, uh, that we put that, that technology through. But the company through, you know, we have a, we have a list of a list of questions.
We have a, we set up a, a steering, community steering group that kind of evaluates that tool. It's very important for us to, to really think hard about everything you're talking about. You know, is this a secure system? Can we, like, what happens to our data? Can we delete our data? Do we have data sovereignty?
Do we know where it goes? Like, you know, when, when we provide data to this company, are they using it then to train their internal models? Like how [00:26:00] can we opt in or out of that? So we have a big, a big questionnaire that covers all of our, um, concerns. Around leveraging some of these technologies and every time we get a request to use a new technology, we put it through this evaluation, and then we have a committee that sits in and looks at the results of that evaluation and decides.
You know, is this something that we want to, we want to invest in? Is this a tool that we feel comfortable with? And, you know, is this a company that we trust and we like, and we feel like they're, they're a good partner? And I think being very intentional about who you're partnering with and, and having that, um, you know, structure around your evaluation helps to alleviate some of the concerns that you're talking about.
And I, you know, I don't, um, as a, as a leader in this space, you really have to think hard about. You know, the pace of change and, and enabling people and developers to have access to the latest and greatest tools, but then doing it in a controlled way. And there's, you know, there's a spectrum of like bureaucracy versus the, you know, the wild west.
And we try and be somewhere in the middle [00:27:00] where we enable. Our teams to have access to the latest and greatest tools, but we do it in a structured and considered way, and we have some governance and, and, and controls around it. Uh, and I think we have, you know, I, I think we're at 50 50 right now. I think we have about 16 tools that we've approved and about about 16 or 17 that we've.
Uh, said no to you, and we're fairly quick. We can go from a request for a tool to an approval within, you know, three to four weeks. Uh, and we can, um, you know, make sure that we are not holding back or blocking. And we have some pilot programs that we run and we have a process that goes throughout. So being, being structured in how you deploy these tools and how you bring 'em into the organization certainly helps with everything that you're talking about.
But you do have to peel back those layers of the onion and look at the company that you're willing to partner with. Um, and the other thing I'll say is, you know. We have, we are, we are huge fans of AWS. We're huge fans of AWS Bedrock, uh, which is, you know, the AI developer platform. And, you know, that is a, is a very, very safe and secure space for us where, uh, [00:28:00] we're able to bring in models.
We, we know where our data goes. There's some controls and systems in place, and, you know, that's really providing a great. A tool for our developers to have access to the latest models and do it in a way that we know is safe and secure and that our, um, our team has some governance and control over it. Uh, and I think that like leveraging a platform like that, uh, you know, means that you don't have stuff everywhere and you're not like, you know, you have one, uh, development team going with this license, another environment team going with this license.
You know, you can sort of put it into a. Um, it's a single tool and then you govern that tool rather than having multiple different tools. So picking a platform at Bedrock is certainly an unlocked Verizon. I, I, I think, um, people should, you know, look seriously about, about, um, adopting them.
Mehmet: Cool. I believe what you described Rob, also applies not only for following the regulations and being compliant and all this, so this is applies also for organization who [00:29:00] want to be, uh, innovative, uh, in regulated involved also as well.
So it's not like just related to, uh, doing something on the data itself, like maybe any new initiative. So if you're telling me like. You know, it happens so fast. So that means, you know, you can innovate much, much faster. Now, me, as an emerging leader, let's say, let's, let's say I'm just get new to the job here, how I first get the support that I needed and how I take this and, you know, push it to my team.
What have you seen working Well also, uh, here
Rob: I, I, you know. In, in any pitch, when any, whenever anyone comes to me and says, I want to use X, right? Like, this is a great tool. I, I always ask for like, what is the outcome that you're looking to achieve? You know, like, what, what are you trying to do with this?
And does the, you know, does the cost. Uh, equal, uh, you know, a lot less than the [00:30:00] outcome that you're gonna, uh, provide. And, and, you know, not many people think about that. You know, they, they don't, they sort of like, I would say 50% of people can answer that question effectively when they first start talking about deploying a tool or like bringing something new into the organization.
And I think as leaders, you know, it, it, it is on us to always be thinking about why we're doing things and the business outcome that we're driving. Especially in technology. We should be thinking about what is the business outcome we're driving. And really the business outcome is either increasing the top line or improving margins by lowering operating costs.
Those are the, those are the two business outcomes, uh, that really matter, um, when you're talking about, you know, deploying something new or, or bringing, uh, something new to the organization. So I think if you can, if you can frame any. Request or for support or any, um, you know, new tool, a new way of doing things in that way, then, uh, you'll get a lot more support for, uh, from the business.
And it'll be easier for you to, to, to sell and, and get more momentum behind. And you know that you, there's ways that you can do that. You can start a pilot. You say, listen, we don't know. We [00:31:00] think we hypothesize that this is gonna save 20% of our developer. Um, you know, time doing test automation because we've seen some study here.
We'd like to bring it in on a small pilot. Uh, we're gonna do it five developers and then, you know, get some hard data on it. And if that works, we're gonna scale that out to other people. So I think if you can do that and you can parlay your project into real business value, you'll get more and more and more support.
And also you'll realize that some projects that you thought were worthwhile, aren't worthwhile. Like you'll actually like when you start thinking like that, you'll realize that, um, some of the things that you're trying to do, actually. Aren't really the, you know, the juice is not worth the squeeze and, uh, that you'll, you will, um, abandon some of those projects.
So I think, you know, that's number one. Get, get support by tying everything back to a business, uh, business outcome. And then number two, when you do have that support and when you're driving, uh, you know, change within the organization and you're trying to adopt things, um, especially something that requires a mindset shift and a human behavior change and tactic that we found, uh, works very well.
And we had this, we just, uh, deployed this, uh, you know, [00:32:00] leveraging. Uh, clawed, uh, across, uh, uh, 50, 53 people. Uh, and we put this pilot program together. We got everyone in a room, in a, in a teams room. Um, you know, we gave them the tools. We encouraged them to share the successes. We, we had a kickoff. We had prizes.
We gave out like, you know, silly prizes, like jars of spicy pits. We, we gave out, you know. Um, uh, uh, uh, tickets to an AI bootcamp for AWS. So we, we had some structure around it and we created this core group of 50, 50 plus people that were sharing stuff with each other. And leveraging that collective intelligence to really kind of encourage each other and transform each other.
And when, you know, people weren't engaging with it, we would kick them out. We'd have this like kind of squi game approach where if you weren't leveraging the tools, you know, then bring someone who is. So I, I think taking something, uh, you know, from that very early pilot, thinking about it from a business outcome perspective, getting support behind it, and then deploying it in a smaller group setting where, you know, you can put energy [00:33:00] into that room.
Energy into that group of people and get them transformed. Then after that you can scale it and roll it out across the organization. And you know, now we're looking at deploying 500 of these licenses across the company and we're already thinking intentionally about how do you take it, how do you take that small group energy and turn that into big group energy?
Because we we're actually, you know, worried that if we just give 500 licenses, the adoption of it is gonna be. You know, 50%, 40%, uh, and some people weren't, weren't really engaged with it. So I think you have to be very intentional and you have to put a lot of energy behind. Adoption and transformation and, and make sure that you are, as a leader, constantly driving people to these stores and constantly like, you know, reminding people, and I sound like a broken record.
You should hear me every single meeting, every single project review. I'm like, have you thought about using ar? It's have, like someone says, you know, this project is read, uh, you know, it's taking longer than we thought. I like my initial response now is always, and I think people are sick and tired of hearing about it now, but I'm like, have you, have you considered using AI tools?
You know, to [00:34:00] accelerate the remainder of work and maybe bring this project back from red to, to yellow or green. Uh, and you know, still people are like, oh yeah, maybe we should even did that. Even though I've said it a million times, you know, it's still not. Ingrained in people's minds to start thinking about AI and AI tooling first.
So when you, when you talk about adoption and you talk about driving change, and you talk about as an engineering leader specifically, you know, driving a technology trends, you know, uh, implementation, like LMS across your organization and getting people to use it for the first tool, you have to just put, like, you have to go all in.
Uh, and you have to put all of your energy behind constantly reminding people to, to go to that tool first. And, and it takes a lot of time and energy. It doesn't happen just by giving people licenses, you know?
Mehmet: Absolutely. So, so pushing people to. To think outside of the books, try to leverage, you know, the technology that we have to make sure that we can, uh, get things done as fast as possible.
So this is, comes from little [00:35:00] bit engineering perspective. Now, if, and, and, you know, I'm talking about here about developing writing code and all this, you mentioned a bit about AWS Bedrock, which is a fantastic platform indeed, from the infrastructure overall also as well, uh, you know, uh. Rob, um, cloud changed a lot of things, made things a lot easy.
But have you seen this combination of, you know, scaling in the cloud with ai also giving you and the team superpowers there? And maybe also how we can, you know, combine these two? Because I think what I, I saw someone shared really a good, a good article the other day. It's actually his thoughts about what happened in the cloud.
So when people move to the cloud, a lot of it was like just moving the workload as is without rearchitecturing. Yeah. You know, from on-premise to the cloud. And now AI comes and hey, like. Uh, [00:36:00] we are paying too much. So do you think we can apply this to rectify some of the mistake that we might have done and maybe it applies to the health healthcare a little bit.
I had the chance to work on some healthcare projects in the infrastructure part, and I have seen. People were scared to touch anything from, uh, you know, architecture, especially, you know, whether on the software layer or on the hardware layer, because yeah, we found it this way. It was working fine. And when they shifted to the cloud, they shifted as is.
Do you think, do you see a role to the AI also to optimize anything here? Maybe the architecture, the cost again?
Rob: Yeah, I mean, I think, I, I like, you know, the, the power of AI is gonna be in driving. You know, 20 to 30%, uh, more efficiency in not only the software engineering teams and the, and the cost of humans, but also in our infrastructure costs and, and enabling us to do, um, technology refactoring and [00:37:00] technology transformation much, much, much faster.
Um, and like, yes, 100, like 100%. And you know, I think that that old legacy systems, the hardest thing about old legacy systems is. You know, they are generally, they don't have good test coverage, right? Like they, they, you don't have, um, a, uh, certainty that when you make a change, it's not gonna break anything in the system.
And something that we are seeing with LLMs is that they can take a untested system. And right unit and integration and sometimes even UI tests for that system. And we can get from zero to a hundred percent test coverage very, very quickly in a, in a, in a period of time that would would've been unreasonable and unimaginable before.
And if you can wrap these legacy systems in automated tests, then you have more confidence that any changes you make to. I haven't broken anything, and then you can start using LLMs to start transforming some of that legacy technology and breaking it out and refactoring. We have, um, you know, we, we've seen great [00:38:00] success with some older, uh, stored procedure code where, you know, we had like, uh, Microsoft SQL stored procedures.
We've seen great success with LLMs taking those store procedures and turning them into T net modules so that that is improving, um, the, the transformation of that, um, older legacy code. And we're doing that because we've been able to wrap these older platforms in. Uh, automated tests and, and ensure that the new system performs exactly the same as the old system.
And we're actually discovering through that process as well, that some, some systems aren't actually doing the things that we thought that they were doing. Um, and you know, as when we're writing these session, writing these documentation, we're then going, fixing the, the systems and, you know, proving the performance.
And this is, you know, things that take. Uh, you know, a couple of sprints, two sprints, three sprints that might have taken six to eight months prior. Wow. You know, you, uh, you don't have the confidence that those changes are, haven't broken anything, and you have to tiptoe around the system. So, yeah, I think, you know, you think about that and you think [00:39:00] about as, as a CTR or CIO or you know, you know, people in your audience sort of like pitching, like how these AI adoption like no board and no CFO or no CFO is gonna say no.
If you can say, I'm gonna shave 20% or 30% of the costs of our infrastructure and I'm gonna say 20 or 30% of the cost of software development. You know, I'm not gonna, you're not gonna get that money back, but I'm gonna redeploy that, that, you know, that capacity somewhere else into new projects, whatever. Or maybe you, maybe you do, maybe you return that, that savings to the business.
Or maybe you can just say, Hey, I can guarantee our costs are gonna go flat while our top line grows 20, 30% over the next two to three years. Like that. That's really how people should be thinking and, and pitching. The power of LMS to their business stakeholders is in terms of a real world, you know, hard cost savings and, and efficiency.
Mehmet: Absolutely. Now, o one thing, you know, we, we talked a lot about AI and LLMs, which, you know, FF fact is LM stole the lights from every other technology out there, but [00:40:00] in Europe here, Rob, are there any other technologies that, you know would emerge with ai? I'm sure with AI, because I'm, I'm not saying anything gonna happen nowadays without AI or in the future that would have, you know, at least the same.
Uh, effect. Uh. On again, reducing cost, making things more agile and getting what at the end of the day, especially in healthcare, and actually it applies to every single vertical, getting better customer experience. So what do you think aside, LLMs can be a running, you know, like first runner maybe with, with, with the AI and LLMs?
I, I think the most
Rob: important thing is consistent access to. Uh, functions right within, within platforms, right? What we see is that we can create an agent, uh, that, you know, does something for a customer, uh, you know, maybe takes a regulatory update and turns it into, turns into [00:41:00] something. If we can get that LLM that or that agent to then act on other systems, then we can start stringing these things together, and that's either through access to an API or access to, you know, in some cases with more modern systems, some of these MCP servers.
Like that's the, that is the key for us stringing these agents together and creating actual, like much longer workflows that are gonna take more, uh, out of their hands of the, uh, you know, humans and, and create, uh, more efficiencies. But that requires that the systems and platforms that we use have consistent APIs, have well documented APIs that every function within the system can be done using an API.
And that's not always the case, right? Like we do have systems that in everyone's ecosystem. We have systems that don't have good APIs. We have systems that don't have every function available, uh, uh, to, uh, you programmatically you have to go through the ui. So I think that's gonna be the biggest unlock is, is we start to see.
The deployment of AI and gen systems, it's gonna force a [00:42:00] drive towards systems, legacy systems exposing their functionalities through API and exposing their data through API. And you know, that's going to, that's gonna change a lot of dynamics within our industry because many of those systems don't expose that functionality via an API or don't expose that data via an API because if they do.
Then it's easy to migrate away from that platform and, and replace it with something else. And they treat it as a sort of, uh, you know, defense mechanism. Uh, and, and improve their stickiness of that platform in their customer base by not exposing those things. And unfortunately those systems are gonna be, uh, you know, uh, you know, capability wise, they're gonna be bagging behind all of their competitors.
And they will, they will prevent the usage of that system inside an air, the gen system, and therefore almost make themselves redundant if they keep that strategy. And I, and I think what that's gonna end up doing is making all of these different systems. Uh, easier to replace, uh, all these different peripheral systems, much easier to [00:43:00] replace over time, and that's gonna change the dy the dynamics of the market a little bit.
And like that, I think, is gonna be a big unlock because then you're gonna see companies that have been hamstrung with all the technology, um, because it has been hard to replace, uh, you know, those, those companies will no longer be hamstrung by that technology, but, you know, in, in five, 10 years because that technology just won't exist.
Mehmet: Right, so, so the question would not be anymore, have you adopted ai? The question how much your AI systems are talking to each others. I think this would be, this is the question, which is I, I think, you know, from API's perspective, uh. A lot of companies, especially when, when, you know, the rise of SaaS, uh, started it, you know, exposing, you know, every single, uh, action you can do within any platform regardless.
Yeah. So you can do it. So this has opened the door, and I think without, you know, I can't imagine. Any system today without having the API exposed to [00:44:00] extract a kind of data from there. Of course, we need still the security, we need the rare guards. Everything needs to be in place. But yeah, to your point, because at the end of the day, if we don't get these.
Systems talk to each other, we might be stuck. And I think the next wave of, uh, innovation would be on that side. So, uh, absolutely. Um, Rob, as we are almost close to the end, uh, this is kind of a classical question I ask at the end if you want to advise. To be CTO. So someone who might be still a junior developer today, or maybe someone who's, they started their career, but they are aiming to become A-A-C-T-O.
What kind of advice do you give them?
Rob: Yeah, I would, I would, you know, learn the business, right? Like, I think as, as you, um, as you progress through your technology career, there's always. Someone above you who's a more, you know, um, more senior from a technology perspective until you get to CTO and then there's no one above you.
That's, [00:45:00] that is, um, you know, a more senior technologist and your peers and the people that you interact with are not technologists. And it's a very, very different dynamic, right? Like when you are a, a vp, you know, you can talk about tech and everyone around you gets it, your peers that you're trying to convince that you should do something.
Our technologists like, you know, you, you. Uh, if you built a system and you want 'em to adopt it, you can just talk about, you know, the speed of the response, the ease of the developer experience. You can talk about how this is gonna, you know, reduce the amount of tech that they have on their platform. It's gonna make their developers a little bit happier.
Like, you can talk tech to your peers, uh, and you can talk tech to your boss, and you get some, you know, oversight into the decisions that you're making. But as soon as you get to the level of A CTO, your peers are now. People in sales, uh, you know, the chief revenue officer, people in finance, the chief financial officer, the, this, you know, the CEO may or may not be technical.
You had to speak that language and you're now in a very, very different team dynamic where everyone understands the business and everyone understands the, the areas of [00:46:00] of their capability. And you have to learn those areas of capability. You have to learn what motivates the chief financial officer. You have to learn what motivates the chief revenue officer.
You have to learn what motivates the chief marketing officer and you know, to be part of that group and, and really understand each other. And understand where your swim lane is and what your role in, in that is, requires a deep understanding of everyone else's, everyone else's function within the business.
And I would say for people who aspire to be A-A-C-T-O, you know, your ability to demonstrate business acumen and understand. The, the facets of your business, and not just your business, but any other business that you, you might want to go into, um, is, is a critical, uh, a critical thing. And oftentimes, CTOs will fail because, you know, we see, uh, senior technologists promoting into that, that position, and they remain technologists.
They don't become executives, and they don't understand the business and how their, their work contributes to a much broader picture.
Mehmet: Absolutely. And this is why I tell, uh, people, especially the [00:47:00] younger generation. Try to understand business more. Don't do the mistake that I did. I took long time until I understood that if I want to really get something better, I need to understand the business side of it.
So when I talk, people understand what I'm saying. So a hundred percent on that Rob, and thank you for bringing this up. Uh, finally question where people can get in touch and know more about you and about HealthEdge. Uh, LinkedIn. I'm
Rob: always open on LinkedIn, so just connect with me. Send me a note if you want.
If you have any questions, let know.
Mehmet: Great. Again, thank you very much, Rob for, uh, you know, giving me the time today and having this, uh, great conversation. Uh, I appreciate it. So I gotta put the link it in profile, in the show notes. So for the folks who are listening on your favorite podcasting app, you'll find that in the show notes.
If you're watching on YouTube, you'll find the link, uh, in the YouTube description. And this is how I end my episodes every time. This is for the audience. If you just discovered us. Thank you for passing by. I hope you enjoyed it. If you did, [00:48:00] so please give me a favor, subscribe. And share it with your friends and colleagues and if you are one of the people who keep coming again and again, thank you very much for Loyality.
Thank you for the support. Thank you for keeping since the beginning of 2025. This podcast in the top 200 Apple podcast chart for the longest time since I started it two years and a half ago. And we are trending in multiple countries at the same time. Even during the. Slow months of summer. So maybe people are listening to us during their travel, which is great.
And thank you for, you know, all what you do for the show. As I said, voting for us, um, bringing us to the top charts, as I said, and thank you also to all my guests, including Europe, because this couldn't happen only by myself. This is the support of my audience and my guests, and as I say, always stay tuned for any episode very soon.
Thank you. Bye. Goodbye.