Inspire AI: Transforming RVA Through Technology and Automation
Our mission is to cultivate AI literacy in the Greater Richmond Region through awareness, community engagement, education, and advocacy. In this podcast, we spotlight companies and individuals in the region who are pioneering the development and use of AI.
Inspire AI: Transforming RVA Through Technology and Automation
Ep 3 - AI Strategies for Optimizing Business Operations w/ Matt Bartles
Ever wondered how AI can transform your business operations or how industry experts navigate the rapidly evolving AI landscape? Join us for an enlightening conversation with Matt Bartels, an esteemed AI and automation consultant. Matt shares his remarkable journey from senior management at Capital One to pioneering AI-driven solutions that reshape how businesses optimize legacy systems. Discover the crucial role of AI literacy and the efforts of organizations like AI Ready RVA in fostering a culture of awareness and trust, essential for businesses to thrive amidst rapid technological advances.
This episode explores the nuances of integrating AI into business operations, particularly in addressing the challenges posed by legacy systems. We speak with Matt about practical solutions for modernization, the tools that facilitate AI adoption, and the significance of performance metrics and user adoption.
• AI-driven analysis for modernizing legacy systems
• Importance of leveraging existing Microsoft technologies for AI solutions
• How to identify and enhance core strengths with AI
• The significance of adoption rates in evaluating AI performance
• Future trends in AI, including the rise of rational models
Welcome RVA to Inspire AI, where we spotlight companies and individuals in the region who are pioneering the development and use of artificial intelligence. I'm Jason McGinty from AI Ready RVA. At AI Ready RVA, our mission is to cultivate AI literacy in the greater Richmond region through awareness, community engagement, education and advocacy. Today's episode is made possible by Modern Ancients driving innovation with purpose. Modern Ancients uses AI and strategic insight to help businesses create lasting, positive change with their unique journey consulting practice. Find out more about how your business can grow at modernagentscom.
Speaker 1:In today's episode, I'm excited to introduce our guest, matt Bartles, a visionary in the field of AI and automation. Matt expertly balances the strengths of humans and machines in the workplace, recognizing the unique capabilities each brings to the table. With a comprehensive background in AI and automation, matt is currently an MBA student and conducting an independent study at the University of Virginia Darden School of Business, focusing on the foundational elements required for the successful implementation of these technologies in business operations. As an automation and AI consultant, matt has spearheaded strategies that drive business efficiency and led complex projects to success. Under Matt's leadership, his clients have seen remarkable success and his ongoing research continues to deepen his understanding of AI and automation, empowering him to help organizations enhance their efficiency and drive operational excellence. Please join me in welcoming Matt Martles to the show.
Speaker 2:Thank you. Thank you, I'm humbled.
Speaker 1:Well, it's great to have you here, matt. Feel free to introduce yourself in any terms that you wish, beyond what I just said in your intro, and roll into what inspired you to start working in the field of AI and what drives your passion for this technology.
Speaker 2:Yeah, so so well. Thank you for the introduction. You know I'm Matt Bartels. I'm an automation, ai and transformation consultant. I've been doing that for about three years. Before that I was a senior manager at Capital One where I spearheaded the automation program. It was an interesting journey over a decade, and before that I was a nuclear submarine officer in the Navy, and so it's been a great journey and I'm really happy to be here with you today.
Speaker 2:And so, in terms of what got me started working in AI, there's really two key factors. One was necessity and the other is impact. And so, after nearly a decade in automation, the industry had really reached a point of maturity. The automation tools, techniques, tactics had really plateaued and you know there's always something out there to automate, but you could really tell the industry had leveled off.
Speaker 2:And so, two years ago, nearly ChatGPT had hit the mainstream and it became clear that large language models were a game changer and a disruption for automation. And so we started to use these large language models for requirements gathering, documentation, process, re-engineering, and then, recently, the large language models can develop automation themselves, and so their low cost and ease of use made things that were previously too complex or impossible to automate. Now those things are achievable, and so embracing AI wasn't something that was optional as an automation consultant, so it became essential to learn AI to deliver value for our clients, and so the integration of automation into AI is going to be a really positive evolution for business operations but also our work life, and so naturally it's going to make things faster and smarter and more efficient.
Speaker 2:But, much like we saw in automation, it's going to empower talented individuals to contribute to more value, add things and get out of the mundane and repetitive tasks and really generate value for their businesses and also their clients. So very excited about that and really the ability to be part of that impact and amplify that impact and drive meaningful change that's really what keeps me passionate about AI.
Speaker 1:Awesome. Yeah, I feel pretty much the same with what pushes me forward. It's more than just being in cutting edge technology. It's about making impact for businesses in the world around us so that we can support growth in life, and that includes bringing people along with us. Great technologies have great disruption patterns, and here we have one of the biggest challenges ahead of us that the world has ever seen, I think, and it's coming fast, and that's what I think organizations like AI, ready, rva, is all about is people embrace that change, and so what you just said resonates with me a lot embrace that change, and so I really.
Speaker 2:What you just said resonates with me a lot. Yeah, I'm really excited about our AI Ready RVA because it really starts to set the foundation and tackle some of the biggest problems with awareness and adoption. Yes, sir, and hopefully trust.
Speaker 1:Trust Yep Cornerstone. Yeah, all right, let's move on. What types of AI technologies or solutions are you building and what problems are they designed to solve?
Speaker 2:Yeah, well, it's a great, great question. So I'm currently focused on AI driven analysis for legacy systems, and I realize that's a mouthful, so let me let me break it down. And so a lot of businesses and you've probably experienced experienced this in your career rely on these legacy platforms. These systems were invested in 10, 20 years ago or they were inherited through an acquisition. A lot of times, as companies grow through acquisition Geico, for example, demonstrates this they have hundreds of legacy platforms that they've acquired from the insurance companies that they have bought over the years, and so these legacy systems create significant inefficiency for businesses. They're difficult to adapt, to change, they require users to go outside the system and do things in email and Excel, and so it creates a lot of efficiencies. And, to even make things worse, there are a lot of times they're undocumented.
Speaker 1:Yeah.
Speaker 2:The original developers. They're long gone. You know these systems operate on tribal knowledge, yet migrating away from them is scary, it's risky, complex, and so 88% of businesses don't do it. They just stick with their legacy system. And so what? Ai-driven analysis for? Build a detailed understanding of the functionality, the business logic and the data models underneath these legacy platforms to create a blueprint for modernization. And so you know, essentially, use AI to uncover the requirements for the new system that enables businesses to like really start to move forward confidently with their upgrades and manage risk.
Speaker 1:Wow, yeah, I would love an example of that, if you have a quick one.
Speaker 2:Oh yeah, I can't give away the client, but it's in financial services and they have a 20-year system that is based off dot net, which isn't an ancient framework, but you know not, not, not the most easy to work with.
Speaker 2:Yes, and it runs off of stored procedures. So they have about 20,000 stored procedures based off of legacy products that they had issued, and it's this created this web of stored procedures that call each other. That's not decipherable by a human, and so nobody can really kind of tell the leadership. How does this system work? Right, like what is what is it? What is it doing? And that uncertainty prevents them from moving forward out of it. Right, because if they have a disruption, that's going to impact customers and they don't want to deal with that, but the vast majority of their IT budget goes to maintaining this platform. So they have to have something low cost and very accurate to be able to modernize, and so the only way to do this is using process mining on the UI layer and then AI tools to like code understanding and data architecture understanding, to kind of map this out and then define the features so that they can modernize.
Speaker 1:Wow, so you're helping them reverse engineer their legacy systems.
Speaker 2:Reverse engineer that's probably a better word for it. You have to put AI.
Speaker 1:I just think about if you need to learn about a system you've got to take it apart piece by piece, and that's what you're doing for these companies that have been meeting themselves in the corners for decades.
Speaker 2:And I understand how it happens right, because you know their core competency is providing financial services to their client at an efficient cost in a, you know, friendly customer servicing yeah, you gotta move quick in the corporate age right like yeah.
Speaker 1:Understood. All right, yeah, so you're building solutions to help companies you know unravel their systems. What tools and tactics do you lean into when helping companies solve these problems?
Speaker 2:platform, and that's mostly because it really provides holistic solutions for the majority of my clients' challenges, right. There's a couple of key reasons for this. The most obvious one is 99% of businesses have a relationship with Microsoft. There's very few exceptions, but it's about 99% and so implementing AI and automation in their environment is really as simple as a couple of clicks and issuing a license or two, right, and so they're already in the door. Also, this really addresses a lot of CIO concerns around cybersecurity, data integrity, and so it greatly quickens onboarding. Right, because you, the CIO is like, hey, if it's in my tenant, let's go ahead and do it right. Yeah, it's also very low cost. Microsoft licenses in the automation space are like a 10th or even a hundredth of some of the costs of the third-party platforms, so it's very cost efficient. And the big compelling reason is the Microsoft ecosystem has really evolved over the past four years and really it deserves to be commended.
Speaker 2:It's no longer just like Microsoft Office and Windows Platforms, like Microsoft Dynamics, the Power Platform and Copilot. These are comprehensive tools that are addressing major business concerns. So, you know, power Automate can do process mining and then automate tasks with low code a couple clicks. It's citizen development ready Copilot Studio. You can create LLMs that then go call the little automation that you built, and so now you have, you know, a very fast route to a agentic AI.
Speaker 2:And then Power Apps is a way to empower businesses to create low-code applications, and these things can feel like real niche applications. Or if you have, you know, a quick service that you need to do or mitigate some risk, you can rally around a quick low-code app that can facilitate that. And so, and then really, microsoft is up to its integration game. It's no longer a closed ecosystems and open ecosystems. They have thousands of integrations that you can tie into, like Jira or things like that, servicenow. So the integrations between systems has really matured. And then, finally, maybe my second finally is they have a really strong partnership with OpenAI, and we all know OpenAI as a leader in AI innovation, and so really my belief is that Microsoft is positioned to be and remain a dominant force in business operations for the foreseeable future.
Speaker 2:So this really kind of like it's a great ecosystem, it's low cost and it's got some longevity, and so that's really where I've kind of staked it. I try to remain platform agnostic, but having some subject matter expertise in Microsoft, I think, is a good thing.
Speaker 1:Yeah, very interesting. I started my IT career with Microsoft Server Technologies and made the shift when I started working at Capital One.
Speaker 2:Yeah.
Speaker 1:Probably about 10 years ago. Yeah, shifted into years ago. Yeah, shifted into Linux-based systems and from there never really looked back. And now you're kind of convincing me that I might be able to take another look at Microsoft technologies again in the future, and especially since they own 49% of OpenAI possibly more in the future. Yeah, and they're a real powerhouse.
Speaker 2:Well, you know, the Google platform, aws and a lot of the competition in the technology space had pushed Microsoft to be more forward thinking, yeah, and business friendly From an automation standpoint. A lot of users solved their process problems with the Microsoft platform, which is like hey, email me this spreadsheet and then I'll go in a system of record, and then I'll go do this A lot of times. These manual processes existed in the Microsoft ecosystem to begin with, and so that's really because it's native to that. It's really great for automation and the AI. The co-pilot space is really mature in the Microsoft platform.
Speaker 1:That's a real key insight. Yeah, that makes sense. What advice would you give to aspiring AI developers or companies looking to enter the AI space?
Speaker 2:Yeah, this is a great question and I get asked this a lot and a lot of people are interested in. This is my advice is start by understanding your core competencies, identify what you are strong at and how AI tools can enhance your ability to deliver on those strengths. And, simply put, use AI to get better at what you are already doing. Well, if you're a great teacher, for example, you can use AI to create custom learning planes for a struggling student, for example, and that's something you might not have done before because it was time intensive, but with a large language model, you can generate that quickly, check it, make sure it suits the use case and issue that learning plan. Another really innovative way I saw a teacher using AI is that they may use ChatGPT to make catchy songs, like a Taylor Swift song about the multiplication tables or whatever.
Speaker 1:And so, oh yeah, do a quick, nice little matchup yeah.
Speaker 2:Yeah, it creates, uh, engaging ways for children to learn complex topics, and it's easier and funner to learn, and so it's an example of hey that your core is teaching. Use AI to be a better teacher, so focus on your core strengths. Trying to compete at the frontier of AI is going to be incredibly challenging, right? A lot of these companies have years of head start on you. They have billions of dollars to invest and the capacity to absorb significant losses. They all acknowledge many of the billions that we are investing in AI that's out the door, and so competing at those frontier models, as tempting as it may be, I recommend that clients focus on their core strengths when designing their AI goals, versus, like hey, we're going to build the next great large language model. That's going to be out of touch for most of my clients and so, yeah, the success lies in finding your niche and leveraging the AI tools to accelerate your strengths.
Speaker 1:Just a quick follow-up question on that. When you're looking to learn something about AI, what do you do? Where do you go? What do you fall back on to learn about a new topic, because it's changing so fast? Just curious about your methods.
Speaker 2:Yeah, so I go a bunch of different places. It really depends on the use case. Because I spend a lot of time on the Microsoft platform, I'll go to Microsoft Learn. I look at a lot of influencers. The industry is moving so fast that a lot of times even Microsoft Learn is out of date of its own platform. So there's some great influencers who are constantly in the tools like, hey, this release just happened.
Speaker 2:And looking how other practitioners are using some of those tools has been light bulb moments for me. But you know, trying to keep up with the latest technology is going to be difficult. It's moving so quickly and you always have to keep an eye on what problems our business is having too, and you always have to keep an eye on what problems are businesses having too, and it's really straddling. Hey, you want to be up to date on the latest AI thing as much as you can, but you also kind of want to understand the business problem, and I think the successful people in the AI space, these new developers, will have the two. They'll understand the business problem really well and understand the tool and technology really well, and that will put them in the solutioning space, which is highly valued right now.
Speaker 1:Great, All right. So, moving beyond solving problems with AI, I'm curious what comes to mind when evaluating performance and models.
Speaker 2:Yeah, I thought about this one a lot, and most of my experiences in chat, gpt, in the various co-pilot, microsoft, llms, and I really think of one metric, and it's just adoption rate. Simply put, it doesn't matter how high performing the AI model is, if it's not adopted, it's not useful. And let me give you a quick example. Machine learning is really able to be superior than doctors at reading and diagnosing x-rays, right. So these machine learning models can look at thousands of x-rays. They can see things earlier than a doctor can, more accurately than a doctor can, and so these have been proven to outperform their human counterparts. But the adoption rate of these technologies is extremely low, and so this really highlights the barrier between adoption and performance and them being really kind of independent of each other.
Speaker 2:There's a lot of reasons for this, like both patients and doctors need to develop trust. They have concerns on regulation, liability if it gets it wrong, privacy. Will AI cater to my, you know unique needs and special sauce? So there's also some concern here, like from healthcare providers, like, hey, if AI can do this, what do you even need me for? Like, what am I here for? And so there's some fear there, and so I think the thing that this underscores is these performance metrics. They're always going to get better. The nature of this industry is these models are going to evolve, they're going to get great, they're going to get more accurate and stronger and faster and cheaper. The adoption rate is going to lag behind if trust isn't built. So you know, not a metric, but like hey, we need to focus on adoption rate because the performance of these models is already really good and the adoption is not.
Speaker 1:Fascinating, so imagine the course of things that you've seen in the past. Yeah, and answer this question what trends do you foresee in AI development going forward?
Speaker 2:Yeah, the trend that's really popular right now is discussion on really starting to see the limits of large language models in terms of how performant they are, and these large language models have just sucked up all the oxygen and AI in the past two years, so everything's a large language model.
Speaker 2:We've forgotten about random forest and the thousands of other machine learning models, and so I actually think the next wave of innovation is going to come from rational models, which is AI systems that think more like humans versus kind of like predicting things. And so you know large language models they're kind of always limited on their ability to look at old data and predict new data, whereas these rational models think of it like reasoning. Well, they reason. Let me give you an example like uh, imagine you memorize the encyclopedia, you read it a million times and you know water freezes at 32 degrees Fahrenheit, right, but a rational model understands like the molecular on the molecular level and temperature and pressure and all those factors and how that affects your freezing rate. So these rational models have much greater systems understanding versus really kind of just predicting the future based on what it's right.
Speaker 1:So these large language.
Speaker 2:Don't get me wrong. These large language models are incredible. They continue to impress me today. But these rational models is the next frontier for AI development and it's going to enable much bigger things, and so we're going to be moving on beyond just augmenting human tasks into these rational models being able to independently manage entire systems.
Speaker 2:Right, that's really exciting stuff, because right now, large language models haven't really been able to manage systems reliably. They're still like agentic is the buzzword, and that's like hey, one agent can do kind of one fairly simple task. But the rational models will be much more capable, for example, overseeing entire supply chains, right, or potentially like providing oversight for an entire data center, and so they have the ability to process vast amounts of data, react faster and predict outcomes based on the rational model versus the predictive model much more accurately than humans and MLMs. And so these models that have strong system understandings, I think is going to be the next development. I think it's a matter of time. I don't have a timeline on it, but that's one of OpenAI's charters is to create this, and so I think we'll see it eventually.
Speaker 1:I will follow those trends then, because it does sound fascinating and it's well.
Speaker 2:I mean, this system's understanding is something that I don't think us as humans really get right. And so if you think of like a giant organization like Capital One Bank or whatever, it's a giant human endeavor and when you zoom out to 35,000 feet and you look down, understanding all the moving pieces between you know, audit and credit and product and fulfillment and call centers and back office, all things. It's a giant machine that not one person has the capacity to understand, its whole thing.
Speaker 1:Yeah, plus they optimize on whatever is the lens.
Speaker 2:You've got these local optimizations and it's a bunch of people doing what they do best, but we're limited by our capacity and contained by specialization. In these big human endeavors, these systems that were created and I actually don't think we're as good as we think we are and these systems thinking rational models will be far superior.
Speaker 1:We always have room for improvement, so we don't get things right the first, second or third time. That room for improvement, oh yeah, so we don't get things right.
Speaker 2:The first, second or third time, that's for sure. No, no, I have many stories on that.
Speaker 1:Hey, matt, it's been great talking with you today about AI systems and the future of AI. I have one last question for you. Okay, I'd like for you to tell me if you could have any superpower yes, what would it be and why?
Speaker 2:Oh, thank, to tell me, if you could have any superpower, yes, what would it be and why? Oh, uh, thank you so much for this question. I'm very excited about this is uh, if I could have one superpower, it would be shape-shifting into like any animal I could be and, yes, I think this would be a great across many dimensions. One is my daughter would love this to create about a billion billion different games we could play, but the other is is I. I think it's like new and awesome experiences that you could have by trying things out, flying and swimming, and I also think I would get invited to really cool parties and get to meet new people. So I think the shape shifting into animals. Finally, I have to say, like I think it would, probably I could use it to increase awareness of, you know, endangered species and all those things, but also I'd get invited to really cool parties. I think that would be it.
Speaker 1:Yes, both A and B. A For everybody. That's cool. That's cool. I like that answer. Thank you for sharing what's yours, man, mine. Thank you for sharing what's yours, man, mine.
Speaker 1:You know, I think it's a little like yours, but I really like to bend space and time with my power. I want to be able to travel anywhere anytime. Oh, yeah, you know it's like I don't want to abuse it. I'm not going to abuse it. Yeah, well, you know it's like anybody could make a lot of money by doing something like that, but I want to help a lot of people. You know it's like a lot of change. And, yeah, you'd be able to like parse through a billion different scenarios and pick the right one. Right, kind of like in the Avengers. Oh, in the game, you know how Dr Strange says I've looked at 14 million scenarios and there's only one that that we're going to win with, and I'm not going to tell you it until you have to do it, and that's kind of that's. That's what I kind of lean into is is, I think, broadly about what issues, what, what's the what's the real issue and how can we solve it, and I think, about all of the various scenarios.
Speaker 1:It's much like that people problem you were referring to a few minutes ago about organizational structures. Yeah, yeah, yeah. What do you do with all of these various components and how do you get to the best answer? Not the one that gives you the most money, necessarily, but the one that saves the day in the end?
Speaker 2:Yeah, that's interesting man, but I like yours too.
Speaker 1:I might steal it for the best. I mean like you.
Speaker 2:I was mostly doing animal transformation to go to better parties and play with my kid. I mean you had optimized, you know, like world peace.
Speaker 1:so I've had to answer that question before so well.
Speaker 2:thank you so much for having me. I'm really excited about AI Ready RVA. I love the events. You know I please give me a call if you need anything. Yeah, this has been a great partnership.
Speaker 1:We have a ton of opportunities for people like you, matt, and we would love to reach out to you and share our event structures and leadership program with you so anytime you're willing to step up and help the community at large. I think we would very much leverage your skills and talent well, thank you so much.
Speaker 1:And thanks to our listeners for tuning in today. If you or your company would like to be featured in the Inspire AI Richmond episode, please drop us a message. Don't forget to like, share or follow our content and stay up to date on the latest events for AI Ready RBA. Thank you again and see you next time.
