Data, Supply Chain Issues and Industry Transformations – Digital CxO Podcast EP82

Hello and welcome to the digital CXO podcast. I'm Amanda Ani, and with me today is Mike Baard. How are you today?

I'm doing great. How are you? Doing well.

We have a big lineup as usual, and we're gonna start with data. Um, we have a couple of different articles to share. These are on digital CXO.

The first one is about data architecture. Um, and it says, data is an asset for businesses aiming for sustained growth in the digital age. Building a robust data architecture is essential to harnessing data's full potential and ensuring that it supports business operations and it's a key component to digital transformation.

We have another article, um, that backs that up too, and again, talks about the importance of data and then it brings up proxies which have emerged as versatile instruments for those that seek to adeptly navigate complexities of data technology challenges. So what are your thoughts on this? We're gonna geek out here for a minute.

And normally digital CXOs are probably not this deep into the weeds when it comes to data management, and yet it's the very thing that kills most of their projects. And, and the issue is that we don't have an appreciation for where the data's actually located at any given moment as it relates to a process. And so sometimes the data's in the cloud, sometimes the data is, uh, in know data center somewhere, and just as likely it's gonna be on the network edge somewhere because increasingly we're trying to process and analyze data at the point where it is created and consumed.

And if you look back at the history of it, it's always been this kind of batching mindset where I can upload it to some central location. Sometimes that could be done in 24 hours, and that accounts for why a lot of the times the applications we're using are a little outta sync with whatever is happening at the moment. But the issue that's really coming to the fore here is that, um, we need more event driven applications.

If you look at something like say Uber and the mobile app, it's very much event driven and people are engaged with that. 'cause even then, after you order a car, most people will watch the little car come around and, you know, they're kinda like, where's my car? And how far away is it?

They check in multiple times. We need digital applications that are like that. They are, um, event driven.

They're near real time. And, but all of that requires an understanding of where the data is at any given moment that we're trying to process and analyze. And that requires somebody who probably used to look like what we call that enterprise architect and digital CXOs would be well advised to go find themselves an enterprise architect who can explain to them what data needs to be where, and make sure it happens.

Because as far as I can tell, a lot of these projects are simply, um, running a ground because of fundamentals of it or just not understood. But Yes, and there's so many hats that CXOs must wear these days, but, um, they can't do it all. So they certainly need people that have that experience because of course we always hear about data being, you know, the key issue.

So making sure they have the right people on their team. Yeah, everybody will nod their head about data, but not in your head about. And understanding that data's an asset is one thing, but um, actually doing something about it and understanding how it flows required a level of skill and expertise that, um, a lot of digital c XOs simply don't have.

Right. And, and, and shouldn't necessarily be expected to have. You know, they need to make sure they have the right professionals in place.

Mm-Hmm, absolutely. So moving on, now we're gonna talk about the supply chain, redesigning it. So there, and this is on digital CXO, there's a growing number of challenges enterprises are struggling with as they try to manage their supply chains in an increasingly volatile world.

So it's good to understand the critical role that advanced technology implementation can play in easing this situation. So can you go into detail about that? I think we've been kinda backing our way into re-engineering the supply chain for a while now.

And every now and again, a crisis comes up, whether it's a ship hit and a bridge in Baltimore or, uh, some geopolitical tension involving a shipping lane. And we never quite get there. And part of the issue is, well, just the phrase supply chain in itself kind of condemns you to a certain amount of linear thinking about how your supply, uh, process should work.

And I guess what I'm just pointing out is, well, a, we now have enough data to maybe apply some AI algorithms to all this in a way that will, uh, give us more insights faster. So that's becomes more actionable intelligence. And we can reroute some of our supply chains, uh, more easily.

We can onboard new suppliers when there is, uh, some sort of crisis more easily and we can be more flexible. But we just gotta sit down and kinda map that out. But maybe the first thing we gotta do is come up with a new term because supply chain as in chain kind of invokes a a a way of thinking that narrows your your options.

Yes. And especially in this day and age when we tend to see a lot of bottlenecks in the supply chain, you need all this information and efficiency Great add. Um, but it's not clear to me that the folks who are running the supply chains know what the order of the possible is.

I think that they have an over dependency on some legacy application somewhere, and they're waiting for some vendor to magically come fix all that for them. Um, that's one way of thinking about it. The other way to think about it is take all that data and stick it in something that looks like, uh, you know, what they call a rag system in the realm of AI and expose it to an LLM and find some way to make sure you're doing that securely.

And you'll be surprised what comes up without having to wait for some vendor to tell you something. Yeah. And we're hearing a lot more about that as an option.

And, and if you don't, uh, see it, there's this underlying theme here of data still data's at the root of everything. If You, if you have data, you have power, you just have to figure out how to tap into it. Mm-Hmm.

Okay, so moving on. This is an interesting article about reinventing the banking experience. So recently Catchpoint analyzed over 50 banking websites to identify the hallmarks of top performing sites.

The study focused on key metrics, DNS time, time to first by document completion, webpage response, largest Contentful paint and cumulative layout shift and overall site availability. So these three sites are the ones that top the list. Franklin Templeton, bank of New York, Mellon Core, and Thrivent Financial.

And then it goes on to explain other things that are looked at in the banking world. So what are your thoughts on this? I think banking is similar to just about every other eCommerce application.

And what we're not really understanding though, is well, I, there are transactions involved, so that's adds a little bit of nuance to the whole thing. But, um, similarly, e-commerce sites have transactions. The issue is that we have so many dependencies on our website now that between where me the end customer actually clicks on something and where the application actually resides to record something, there's all kinds of, uh, dependencies that sit between that.

Everything from, um, DNS servers and all kinds of networking infrastructure that most digital CXOs may not want to concern themselves with. But all of that stuff conspires to add latency and latency is the enemy of customer experience. The, if the longer it takes for the round trip to occur, the more that the end customer has a sense of that, that thing is klugy doesn't work, doesn't have their right level of experience because while everybody's expecting a near real time experience these days, and once again it comes back full circle where the data architecture really does matter here.

And understanding and gaining some observability and visibility into that is crucial. Yeah. And it seems like from other articles that we have posted and just from hearing from other people and even my own experience, I feel like the banking industry is a little bit more behind when it comes to digital transformation and customer and user experience than other industries.

Yeah. Well we can talk about other industries in a minute, but I mean, I don't know, maybe they're all equally behind. I'm not sure.

Yes, this is true. In fact, we're going to be speaking about the next industry now, and that is, uh, the medical industry. So this article is on tech strong ai, and it says generative AI tools are swiftly transforming the healthcare industry from medical offices to operating rooms.

A white paper from the US National Institute of Health recognizes the numeral advantages, limitations, ethical considerations, future prospects and practical applications of AI in healthcare. But with news about erroneous results from citing fake legal cases to performing poorly answering drug related questions, it has also been generating controversy and concern in the medical realm where accuracy means life or death. A more nuanced concept is needed such as collaborative ai.

So can you go into detail about that? Well, I think we can all agree that hallucinations are a bad thing, especially when it comes to medical scenarios where we need something that is, shall we say, closer to a hundred percent accurate. And here's the subtle thing about generative ai.

When it comes to digital transformation, for the most part, generative AI is uh, what they call probable meaning that the thing is making a best guess to appeal your prompt based on the data. Here we have data, again that has been used to train a lot of the time, even though the, the LLM is trained to be overly uh, helpful is the only way to phrase it. And so in their zeal to be overly helpful, they will hallucinate and give you something that they think is the right answer or the answer you're looking for.

And that turns out to be something of a disaster. And the issue becomes, um, most of the workflow in the medical industry is what I would describe as deterministic. And that means that it has to be pretty much a hundred percent accurate and done the same way time in and time again.

And the LLM might not do the same thing the same way twice. And so that requires, uh, a next generation of LLMs, shall we say, that will be used in this space. You'll see what we call domain specific l lms or they are a variant of what people are calling small, large or small language models versus large language models.

And they're trained for specifically one use case or two use cases, and they're relatively easy to, um, embed in an application. And I think that that's kind of where this whole thing needs to head from here is that, um, large language models are helpful and interesting, but um, in a computer science kinda way, if I need them to help me write a better email, okay, great. Um, you know, that, that's nice, but is that gonna move the ROI ball?

Probably not. I think what we really wanna see is these more domain specific LMS that are trained for particular vertical industry, and that's where digital CXO should be focused on is going back to their squads and saying, how can we customize something that looks like an LLM today that's small, that is gonna work in our industry for our particular use case? And there may no ever be something that's a hundred percent, but you know what, 99 98 is a lot better than your average human does anyway.

So there's progress to be made here. I just think we need to kind of step back from the LLM hype and start thinking about the next generation of these language models. Yes, absolutely.

And remembering, uh, when it comes to that collaborative aspect that the human element's still very important to make sure that the technology is, is giving the correct output. Yeah. Although I scratch my head sometimes about all this because, well, we've all been in the medical office and the probability that the humans are gonna make a mistake is frigging eye.

So I kind of sit back sometimes and I wonder how much worse could the machine really be A system of checks and balances between human and and technology? All right. And then lastly, we're gonna talk about AI and the music industry.

This is also on Techstrong ai, a new partnership between Secret Technologies and AI powered Search Engine and Live One, an award-winning music, entertainment technology platform holds the promise of sparking a creative revolution in the music industry by giving artists the ability to tap into vast catalogs of beats and sounds. This partnership will deliver to music creators later this year. The industry's first AI powered search platform for Beats and Sounds, and it promises to do it for music creators, what the MP three player did in the general population, giving them access to limitless catalogs of beats and sounds.

So this is certainly a digital transformation in the music industry. Yeah, I imagine it's kind of like Spotify for sampling and you know, there's a lot of folks out there that kind of, you know, create their own music and put all this stuff together, whether it's hip hop or whatever else it may be. Um, I think this is, you know, a good use of AI in the music industry.

And man, I could be wrong. I'm not in in the music space, but for those of you who've been following it, the controversy around the usage of AI in the music space is really high. Um, artists are particularly worried about, you know, their entire songs being, uh, shall we say, either lifted in part or in whole and being used to drive some other song.

And, uh, again, going back to our earlier conversation about the LLMs, they are not as discriminating as we would like them to be. And, um, so we're starting to see a lot of pushback. There's lawsuits, and at the core of the argument I think is, uh, whether or not, uh, it's okay for an LLM to scan some sort of website to get some sense of what that thing is about under the heading of fair use versus copyright.

Um, I don't know whether this lawsuit issue's gonna wind up, but it would seem to me that, uh, especially when it comes to music, um, you know, rights have been fairly well established for as long as I can remember. So I don't think that just because something found something, uh, on a website somewhere that that constitutes, you know, quote unquote fair use. Um, whereas I don't know, maybe a news story it does that I, I guess I'm not quite clear that every piece of content is gonna have the same kinda rights issue to associated with it, but looks like all this is gonna play out in court soon, but in the meantime, hey, you know, if you wanna make house music or whatever it is, it looks like it's getting a lot easier.

Yeah. So it seems to be there are many benefits from ai, but as you mentioned, there's all these problems too, even with the voice, um, technology being able to mimic the voices of singers and DeepFakes, who knows if it's a real music video of that person or not. So these are issues.

All right, so you're saying that there'll be an image of me singing Purple Rain somewhere? It could, it could happen. It could happen.

I mean, we've seen this. So that brings us to the end of our list of articles shared today. Do you have any final thoughts for our audience, Mike?

Um, I think it was always about the data and we lost sight of that, or we continue to lose sight of it. And I think everybody might wanna take a giant step back and say, all right, let's organize our data in a way that, uh, may maximizes as much value as we can get out of it on the assumption that, you know, there'll be some AI capability or some other reason to drive it. Um, and yet we all seem to think that we can just kind of magically say, we're gonna write some code, gonna build some app and we're gonna create some awesome experience.

And then when we get there, we find the same thing over and over again. We just don't have the data to drive it. Most of the startups that I talk to, part of their biggest issue is they have some awesome application, they just have no access to data to drive the thing.

Um, the paradoxes, the larger legacy platforms have access to all the data in the world and just not a very compelling user experience. So, um, somewhere between these two extremes is the middle I we should start working towards. And I think maybe digital CXOs can provide the, the leadership to do that.

But like I said earlier, you gotta understand what the core issue is from an IT perspective if you're gonna have a meaningful conversation about all this. Yeah, it takes a large team of knowledgeable professionals in different areas to make this happen. Alright.

Alright, well thank you for tuning in again this week to our podcast. And if you didn't catch last week's, it's there. Go back, catch up and again, let us know what you're interested in reading about and what was the most interesting to you today on the podcast.

Until next week, have a great day. We'll see you next time.