Everyone talks about human-in-the-loop. Right now, it is still AI in the loop.
In this episode, Alfred Olivares, Healthcare and Life Sciences Industry Leader at HTEC Group, talks about what it really takes to make AI work in medtech. He shares how his background as a manufacturing engineer shapes his view that companies should run a value stream analysis before starting an AI program. He explains why data governance, data ownership, and data maturity matter, and why AI models now need managed services with people who understand the business context. He also shares how HTEC took its own AFib detection device through to FDA submission readiness, and why he sees conferences as a place to build long-term relationships.
Tune in to hear why AI still needs people to validate the data and the output!
Resources:
[00:00:03] [SPEAKER_00] Hello everyone, welcome back to the Outcomes Rocket Podcast. We're here at the MedTech Conference recording episodes with some of the best thought leaders in MedTech. Today I have the privilege of hosting a friend, Al Olivares. He is just an outstanding leader at HTEC Group and I want to just welcome you to the podcast. Finally, we're getting this done.
[00:00:25] [SPEAKER_01] Al Olivares, Finally, and in a big room, I feel like I'm in like a big media room. It's awesome.
[00:00:29] [SPEAKER_00] You are. Here we are. Were you here for Device Talks Boston too?
[00:00:34] [SPEAKER_01] No, I missed that one. I was in Device Talks in Minneapolis. Minneapolis. And then we run into each other here, Hal, him, like all the places anyway. So it's about time we got a chance to talk a little bit. Shoot to that a little bit.
[00:00:48] [SPEAKER_00] Yeah, I'm excited about it. Al is one of my conference friends that I literally see at most as he mentioned. So, Al, help us understand a little more about you and then tell us about HTEC and the work that you guys are focused on.
[00:00:59] [SPEAKER_01] So me, my background, believe it or not, I'm a grunt manufacturing engineer that somehow went into technology and consulting. Started my career off with like the big firms, Deloitte, IBM, Cognizant. Worked in the mid-market firms with Kone Resnick. Did a little bit of software and now I'm with HTEC Group. So a degree in engineer that had the ability to understand business processes and have the conversation around how technology can enable it.
[00:01:25] [SPEAKER_01] And I think one of the great things about the world that we're in today with AI, you really need people like me that have that ability to translate business to technology and back and forth. Yeah. And be that liaison. So HTEC Group is a digital engineering firm. And what we mean by digital engineering is that we service multiple industries. I am responsible for the healthcare life sciences industry with the group.
[00:01:47] [SPEAKER_01] We look at developing devices for companies, for the big strategics, enabling them with AI and also enabling and developing software. So we've worked with software companies as well in the QMS space. We have a regulatory practice. We're one of the few vendors out there to compare to some of our competitors that actually developed a device. And we did it to not serve initially the healthcare industry, healthcare life sciences industry.
[00:02:12] [SPEAKER_01] We did it to work with that little apple in the sky, as we like to call it, right? Nice. What device did you guys make? It's called Humids. It's an AFib detection device. And we got it all the way through 510, ready for submission for the FDA and realized that, wait a minute, the minute we put this out there, we're no longer an engineering services firm. We're now a product company. So we kind of licensed it out.
[00:02:35] [SPEAKER_01] It's now being sold in China, but we're proud of that because it took us from the journey of the inception of the concept, working with the cardiac doctors, figuring out what the right algorithms are, developing the hardware, the device itself, how it would look, looking at the firmware, putting all AI inferencing technology and everything into it. So it was an interesting journey. I wasn't part of the journey. I wish I was. It would have been even more fun because I only joined H-Check about two years ago, but we've taken it all the way to market.
[00:03:05] [SPEAKER_00] I like that. It's kind of like that. I'm not only the president, I'm also a client.
[00:03:09] [SPEAKER_01] Right. I used to say that about my hair, too, but then I got rid of that when I decided to stay without hair.
[00:03:14] [SPEAKER_00] You guys remember that hair club for men? If you're old enough, you'll remember it.
[00:03:17] [SPEAKER_01] I don't know. If you're old enough, you'll remember it. For other people, you'll be like, well, why would you do that? Just go to Turkey.
[00:03:23] [SPEAKER_00] That's awesome. No, I love that. You guys did the process. You showed that you could do what you help others do. Yeah. And that's so important in this space, that nuance, that understanding. You mentioned your experience in business systems like EHRs and EMRs. Today with AI, it's so important to have clean data for it to actually do what it should and do good.
[00:03:50] [SPEAKER_00] Talk to us about that and juxtapose that with what you're seeing here at the event.
[00:03:55] [SPEAKER_01] So, look, we talk about data, but data has been a conversation that has been happening for God since cloud started in the 70s. Because cloud started in the 70s, let's all be very clear. Terminologies in the 2000s became more prevalent. Yeah. Same thing with AI, artificial intelligence. Technology has been there. But the key thing is always two things that I look at is data and the people. All right. So, we're going to talk about the data piece.
[00:04:20] [SPEAKER_01] One of the things when I'm talking to clients, what I'm seeing in the conversations here is like data governance, data ownership, right? The accuracy of data. What does it mean to build models and train models with bad data versus good data and spending the time to really understand where your data maturity is? It's really critical in the AI world. We've talked about what I've seen in the conference a lot is people are talking about AI. The conference doesn't have as many AI sessions as I expected. Yeah. That's one.
[00:04:48] [SPEAKER_01] They had a kickoff on Sunday, which I unfortunately missed because it was Google and Microsoft and I wanted to attend them, but I had some things. Yeah, that was on Sunday, right? That was on Sunday. Missed that too. And then they have about two AI specific topics during the day and the afternoon. But majority of it is around regulatory compliance, around market access, the reverse pitches. But it's almost like now AI is embedded into every piece that they're doing, right?
[00:05:14] [SPEAKER_01] So, if we bring that back to data, the reality of it is that if you don't have clean data and you don't understand where your data sources are coming from, whatever AI initiative you put out isn't going to come to full fruition. So, and if we take it a step back, and like I said earlier, I'm a manufacturing engineer by heart. So, shop floor production is where I always base everything on, right? Orthopedics, right? Helmedica. Helmedica, then Osteonics, and then Striker. Yeah.
[00:05:41] [SPEAKER_01] And I always tell people this is that when you do a lean initiative and you understand the value stream. So, a value stream obviously is the beginning and end of a product's life, right? I think in AI, and what I'm not seeing here is the conversation around where do you start your AI program more effectively? And do you start at one area or do you build a kind of vision? And I try to tell people, bring it back to value streams because of production.
[00:06:07] [SPEAKER_01] I was having this conversation yesterday that if I'm on a shop floor, an assembly line, and I've got machines in the bucket, and I go and improve one aspect of a machine and increase its efficiency, it has upward impact, it has downward impact. Okay? Yeah. So, and if you look at the value stream, you have data, feeds, material flow, and the people. So, people, process, technology. And if you don't do a value stream analysis first before you really go down the AI journey, I think you're missing out on opportunities.
[00:06:37] [SPEAKER_01] Because if you improve that one machine, the upward effect, right, at the front line is, well, maybe I need to now increase inventory levels or capacity of that area to keep that machine running if I want it to run 24-7. And AI is a 24-7 thing. Yeah. Right? And what does it mean downstream if I'm running that machine faster and producing more widgets? What does it mean downstream? Do I need to have more quality checks? So, do I now take an initiative of building an AI solution on down the stream to build quality, as an example?
[00:07:07] [SPEAKER_01] So, I think what I'm not seeing at the conference is those conversations. And then the governance aspect of it, which it's no longer typical managed services of software. It's really managed services of models. And the skill set required for that is people with experience. Not necessarily the developer that developed it, but people that have that value and experience of the business context. To ensure that whatever is being developed is not hallucinating.
[00:07:33] [SPEAKER_00] It sounds like that would be a really great workshop here. I agree. I agree. You guys should think about doing that.
[00:07:39] [SPEAKER_01] You should reach out to these guys. We're actually going to be hosting a roundtable at another conference where we're going to bring that topic up. Right? Really, what does it mean? What does it mean to go down an AI journey? What does it mean to really drive improvement? How good is your data? How quality focused is it? Because what you're going to feed it is what you're going to get. Right? And that's something that I think, if you look at history for the last 30, 40 years, even in the 70s, data is always one of those things that people kind of underestimate a little bit.
[00:08:08] [SPEAKER_01] And I think you can't do that anymore.
[00:08:09] [SPEAKER_00] I love it. Yeah. And it's one of those things where you think you have a great idea for the technology. But your idea and the application sits within a very complex system. And there's upstream and downstream impacts that potentially, when you look at that small focus area, it's a good idea. But it might not be when you take a look at the cost, the additive equipment, whatever it might be to the whole process.
[00:08:38] [SPEAKER_01] Correct. I use machines and manufacturing because that's a tangible. Yeah. It's tangible. People can digest that. When it comes to AI, the output is tangible, but developing the solution isn't as easy to visualize. Right? So, I think at the end of the day, AI is an amazing tool. I struggle personally when people say human in the loop. I had this conversation the other day.
[00:09:04] [SPEAKER_01] To me, it's AI in the loop right now because you still have humans that create, develop, think of the solutions. Do I think in the next five years, it'll really be human in the loop because AI will overtake a lot more things? Yes. But right now, I kind of like to stick with the AI in the loop. That's funny. I like that. Some people think I'm afraid they're saying that, but I'm like, listen to me.
[00:09:34] [SPEAKER_01] Just like in AI, how you can create context and feed it is how the models will work. Like, so to me, my mindset is AI is capable, but it still needs me. Yeah. And it's going to need me for a while in order to validate that the data is right, validate that the output is right, validate that it's learning because you're training it just like you're training a new employee, another person.
[00:09:55] [SPEAKER_00] Yeah. And, well, what I think about when you say AI in the loop, it's sort of like take that downstream five years. It'll be human in the loop.
[00:10:10] [SPEAKER_01] Listen to me. Let's be very clear. There's hundreds of already shown that an audit specifically, right, AI finds things that a human could not, right? So the people are going to argue with me and probably you'll get some comments saying, well, this guy's nuts. Why is he saying human in the loop? Right? AI in the loop. AI in the loop. AI in the loop. AI in the loop. AI in the loop. But no, it's not. It's still AI in the loop. I love it. I love it. Guess what? The person that, even though it found it, somebody else, a human had to go and validate that it was correct.
[00:10:40] [SPEAKER_01] We're not there where it's just AI.
[00:10:42] [SPEAKER_00] Everything AI says is, you know, the golden ticket. I love that, man. I mean, it's so good. It's so good. Because, again, like you did with value streams, right? You're sort of zooming out. And you have a talent of doing that, right? You have a talent of zooming out of a topic and actually seeing it for what it is and then working forward from that. So you and I should think about doing a piece on that. Yeah, definitely. I'm happy to get it. Yeah, I'm game. Okay, let's do that. All right.
[00:11:12] [SPEAKER_00] So let's move from that and tell me we've got a couple months left in the year and then we've got 2027 around the corner. What's top of mind for you and for H-Tech as it relates to the future of where the industry is going in MedTech?
[00:11:26] [SPEAKER_01] So from an H-Tech perspective, top of mind for us is how do we impact the industry in a positive way? How do we take our learnings? Because the company has been around since 2008 and we were doing AI work even then. And our engineering capability, because we have really bright engineers, we want to see how do we impact the industry? Like how can we do better things for our clients? How can we collaborate with our clients? How do we partner and co-create with our clients? And I've worked for firms where we always talk about that.
[00:11:56] [SPEAKER_01] And it's very difficult with large firms, but our firm is truly a partner. Like we will collaborate. We will invest time. We will build partnerships. Because for us, the value is in working with these organizations is that we're challenged with problems. They're problems and we're helping them. So we get that value.
[00:12:18] [SPEAKER_01] And that's something that, you know, the one thing I love about MedTech conferences and the healthcare conferences is that at the end of the day, we are truly in an industry that impacts lives. And it doesn't have to be a million lives. It can just be, it's just one life gets impacted. That's a great industry to be in.
[00:12:36] [SPEAKER_00] It's the long-term game. Yeah. I love that. Well, that's a great way to close. By the way, it just dawned on me, you know, so we met. I remember you invited me to an event you were doing at Health. Yep. And we were both sitting down getting our shoes shined.
[00:12:50] [SPEAKER_01] Yep. Yep. Yep. That's right. You remember that? That's right. We both walked in. You're not wearing your shoes. I'm not. Because I got my boots. I still wear my boots. People like laughing at me for wearing the boots. But yeah, you're right. We sat down. We just started talking. And we just had this conversation that I invited to an event that we were hosting. And you came. I love it. But that's the thing about these events, right? It's not just about consulting firms, organizations. We have to look at our bottom line, right? Yeah, of course. Generate revenue. But these conferences are about learning.
[00:13:17] [SPEAKER_01] And learning doesn't also mean you go to sit in a room and someone talks in a pan and you learn. Learn means you have a conversation with the folks. Learn from your peers. From your peers. Yep. Right? And that day I learned a little bit about you. You learned a little bit about me and you can't do that. And now look at this. I know. It's great.
[00:13:31] [SPEAKER_00] We're recording a podcast. And that's how it happens. Developing relationships. You don't want to be transactional. No. You want to play the long-term game. Your closing comment reminded me of that. It sort of like inspired this long-term thing and reminded me where we met. If people wanted to reach out to you, where can they do that?
[00:13:49] [SPEAKER_01] They can do alfred.olivaris at htechgroup.com. I'm on LinkedIn. You can always reach out to me. Look, not every engagement needs to yield revenue. What it needs to yield is a learning opportunity for both of us. So if anybody wants to reach out to me, they want to have a conversation. There's nothing more that I enjoy is having a dinner conversation, not about work, but about what's happening in the world and how we're impacted. So, and AI is one of those conversations you can have that's not about work. It's how it's impacting. So I'm happy to do that as well.
[00:14:17] [SPEAKER_00] I love that, Al. And I will say, as we did on this podcast on two topics, one being the value stream and how you apply AI. And the second one being about essentially AI in the loop. Al does have this talent of zooming out and helping you see the forest from the trees. So I would say, take them up on a conversation and you won't be surprised with the insights he offers. Thank you all for tuning in.
[00:14:44] [SPEAKER_00] In the show notes, you'll see all the ways to get in touch with him and the team at HTech. Signing off now. And Al, thanks for being with us. Thanks again, man. Appreciate you, man. Yep. Appreciate you. Thank you.