Bridging IT and Smart Infrastructure: How AI-Driven Workflows are Reshaping IT Leadership

Welcome to the Hitchhiker’s Guide to IT, brought to you by Device forty two. On this show, we explore the ins and outs of modern IT management and the infinite expanse of its universe.

So buckle up and get ready to explore the ever changing landscape of modern IT management.

Hello, and welcome to the Hitchhiker’s Guide to IT, where we explore the latest innovations shaping IT operations and infrastructure. I’m your host, Michelle Long Moody. Today, we are diving into how AI is transforming IT management from automating workflows to optimizing cloud environments and scaling infrastructure for AI driven demands. With AI rapidly changing the way IT teams operate, organizations must rethink how they balance automation, security, and efficiency. Joining me today is Jeff Hagan, CTO of Contextable. Io, a technology leader with deep expertise in scaling platforms, automation, and IT infrastructure. Jeff, thank you so much for being with me today.

Thanks so much for having me.

Looking forward to getting into the conversation. Before we do that, however, can I ask you to give us a brief bio if you can, please?

Oh my goodness. You know, I’ve been doing this for a very long time. I could tell you that my first job as a software engineer was in nineteen eighty.

So that’s gonna that’s gonna date me really quickly. Right?

But, so, so I’ll skip over a lot of that, right, and just give you the highlights.

I’ve been the CTO or the CIO, and that’s a that’s a bit unusual in and of itself that I’ve been in both roles a number of times. I’ve been the CTO or the CIO for seven different companies, I think.

And at some point in that journey, I also transitioned from being a a a technology executive for hire to being a founder and, starting my own companies.

That’s pretty incredible, Jeff. And clearly, we have the right man for the job here at the task at hand, which is talking about, you know, how we’re seeing evolving data centers and how AI plays a role in that. And as you said, you have seen quite a bit and especially during this day and age, and, you know, we’re gonna talk much more about this, but I think we’ve seen such a transformation just in the last ten to twenty years more than, you know, society has ever seen when it comes to technologies. So I can’t wait to hear your insight on our questions today. So we’re gonna start off here. AI driven automation is reshaping IT asset management, IT operations, and IT service management. How are IT teams adapting, and what would you say are the biggest opportunities AI presents in these areas?

You know, IT, ITSM is all about sometimes predicting that there’s going to be a failure.

Right? And then, and then evaluate and and then letting somebody know that there’s gonna be a failure. The opportunities with, AI in that space are to do a better job of failure prediction, right, of understanding failure modes, predicting failures, or when a failure occurs, doing a better job of kind of incident and alert analysis to decide whether this is actually a failure that a human being needs to know about, or not, and maybe which human being needs to know about this failure, or what are the steps to resolve it. So So so I think, you know, lots of opportunity, but I I think, you know, and we’re I’m sure we’ll talk about this more, but I think the opportunity is, needs to be balanced with, you know, taking on automation projects that have a clear return on investment and, frankly, aren’t too big.

You know, you mentioned AI is everywhere. I mean, you cannot escape it. So with AI taking over more operational tasks, how should IT leaders rethink their strategies? And then you talk about that balance between humans versus machine. What role then will human oversight play as AI driven automation continues to evolve?

I think human oversight is actually the biggest single challenge and risk that exists with AI.

We are not, like, you know, we’re we’re not we’re not in danger of AI taking over the world and destroying us all. Right? At least at least not in the short term.

But but what we are in danger of is forgetting how things work.

As we automate things with AI and, ultimately, as we have AI making decisions about what next to automate and automating that, I actually do worry about, a world in which developers, engineers, but even business leaders forget the nuts and bolts of how things actually work.

And there is definitely a big learning curve there and a lot of challenges as you said, and managing AI workloads across hybrid and multi cloud environments presents a challenge as performance, security, cost, which is a huge one. So what are the best practices for IT leaders to ensure scalability while, of course, maintaining efficiency?

I think, you know, we’ve all experienced what I’m about to say, with large language models, especially when they hallucinate.

Hallucinations are a symptom of poor training or poor quality data going into the model during the training phase.

AI in the IT world is frankly no different. You’ve gotta start with good quality data, because if you feed too much bad data into your AI, you should expect bad results.

Right? So, you know, ensuring high quality data, whatever you’re doing, whether that’s, you know, whether that’s, using a retrieval augmented generation RAG to, to create an IT customer support bot that is gonna answer questions from a customer about a problem that they’re having. Well, that’s still IT. In fact, you know, support is a big huge part of of IT.

That support bot needs to answer those questions.

The support bot is, you know, of the quality of that bot is a function of the quality of the data that went into creating that thing. Right? And whether you’re fine tuning a model or whether you’re using retrieval augmented generation, the quality of the data that you put in is gonna have a big determine determination determining factor on the quality of the result and, therefore, the quality experience that your customers are going to have. Right? So think about the quality of the data.

Maintain human oversight.

There is no scenario in today’s world where you should be deploying AI without some kind of human oversight for the reasons see number one. Right? Because of the quality because of the data quality issues, and you may you may not fully understand what data quality looks like until you see a bad result.

But it’s never gonna and this goes back to number two, you know, human oversight. It’s never gonna be AI’s responsibility.

AI is a tool. And I think the second you start looking at the the AI that we have today, and I’ll I have to, I guess, I have to to qualify everything I say. I’m talking about the AI that exists today. You know, what it’s gonna look like two years, five years from now, like, we don’t we don’t have AGI today. Right? We don’t have an AI that could actually replace a human.

And anybody who says we do, see number one, doesn’t understand how things work. Right? Doesn’t really understand how these models work.

It’s math. It’s not intelligence. And so, you know, when we think about, you know, best practices and and things that you should be doing, you know, taking a human out of the loop is think about that as adding risk because that’s exactly what it’s doing. Automate, review, automate some more, review.

You know, automate a little bit more, review, but still have humans in the loop reviewing at some point. Right? And eventually, guess who those humans might get replaced by? A different AI to do the review.

Right? And that’s another best practice, which is there is no such thing as an as an AI enablement project that we do where we’re using one model and assuming that it does things correctly.

Yeah. It’s amazing. And not to to go back to when you said that was dating yourself of working in this industry since the eighties, but I think of how you can cash checks on your phone. Like, everything that we do that, you know, ten, twenty, thirty, forty years ago, it just you couldn’t even think. So I wanna ask you this because you have experienced, Jeff, scaling platforms like SmartThings.

What lessons can IT leaders take from large scale infrastructure growth when preparing for AI driven workloads?

AI is that that that nothing’s actually substantially changed because of AI in terms of IT infrastructure growth. Right? What we’ve already seen like, see what you think of this. What we’ve already seen over the last, you know, twenty years is an explosion in data and our ability to process that data.

And by that, I’m referring to the, you know, in in IT environments to the mountain of log log data that gets generated out of every application, every environment. There’s just insane amounts of data going from everything from a server in your ERP system to the laptop computer that you’re using at work. They’re all sending that in today’s world huge amounts of data.

Right? And so and so we’ve already I feel like we’ve already learned the lessons as an industry on how to process and deal with large amounts of data like that. Right?

What what we have to get ready for is more AI processing of that same data, and, and in some cases, the added cost of doing that. Right? Because if you’re already dealing with massive amounts of data, and now you say, oh, but now I want AI to process it, that sounds expensive to me.

Right?

There’s a reason perhaps that you don’t want AI to process that data, and you wanna you wanna stick to traditional models traditional mechanisms for, you know, whether it could still be machine learning, right, but but not large language models because that may not be what they’re really good at. So I think maybe that maybe maybe what I’m getting to in my own mind, Michelle, is is that the answer to that question is you have the right tool for the right job.

Right? Large language models, that kind of AI isn’t good for everything.

Right? And trying to force fit it onto a problem like massive explosion in data, it may be the wrong answer.

I want So I wanna dive a little deeper there because we need to look at what not to do. Right? We we know the benefits here. We know the right things to do, but sometimes more importantly, you mentioned the difference between CIO and CTO and and what is driving through, what is driving around, and and when to know what to do.

Right? So what are some of the biggest mistakes that IT teams make when trying to optimize their environments for AI powered workflows? And then how can they drive around? How can they avoid them rather than driving right through?

So I think, one of the one of them I’ve already said, right, is biting off too much. Right?

And get it and and the reality of that is not, that you you you bit off too much, you tried, and it failed.

That is that that happens. That absolutely happens. But the important, aspect of that that attempt to AI enable a business process and biting off too much and then failing is that you at that point, you’ve now lost the trust of the leadership in the business, and there’s not gonna be another one. Or if there is, it’s gonna be a lot.

Right? So so whether it’s your CIO, your COO, your CFO, or your CEO, right, they are both they are all being pressured by their boards of directors to implement AI, reduce costs, right, increase margins. They’re all feeling that squeeze, and yet it doesn’t mean that they’re willing to just throw, you know, unlimited amounts of money at trying to AI enable their business. They’re not.

Right? And so the reason you wanna take bite off little AI projects and prove that you can do it and be successful not at a prototype, not at a proof of concept, but be successful at taking that AI enablement project and getting it all the way into production.

That’s the key. Right? And that’s and so, you know, you wanna start small, fight off a little project, prove that you can do it, do it do the proof of concept, get it all the way into production and operations so that so that you’re you’ve now delivered on an ROI for that project, and now you’ll get to do another one. But if you go too big too fast and fail, and we’re not in a we’re not in a point yet. I I don’t think I’ve ever seen any statistics, any reports on AI enablement project failures.

But our experience is that there are plenty of them because of what I’m saying. Right? Because because people over overreach. They think AI can do more than it can.

Right, in multiple dimensions. They think AI can do more than it can, or they bite off more they bite off too much scope, and, ultimately, that results in failure. So little bite off little bits. Right? I think if I if I’m giving, CIOs, advice on how to a AI enable their business, that is that like, I would put that as number one, and and and everything else I would have to say would start at number five because that should be, like, number one, two, three, and four.

I think, the other thing that I think I had already mentioned, the second thing is, what I mentioned before about taking a multimodal approach to the problem. Meaning, don’t assume that you can build it with one particular model, whether that’s a large language model, a machine learning model, you need to take a multimodal approach to the problem and test your AI enablement solution, your automation, against multiple different models to see which one is actually gonna give you the best results. And then over time, retest because new models are coming out every single month, and the model you use to solve a problem one type of problem last month may not be the best model for solving the problem this month.

Finally, then the third the third kind of component is not taking people out of the equation. Right?

So automate, but figure out where to still have a human review in the process in order to make sure that everything is working the way that it’s supposed to be working. It doesn’t mean that you might not be able to get rid of that human eventually, but one of the one of the things that we see people doing is implementing an AI automation, taking the human out of the loop, and then no one’s there to see the problems, there are always going to be edge cases. There are you know, matter what you do, there’s always gonna be an edge case that doesn’t work quite the way you think it should, and that’s what the human is there to catch so that you can modify the enablement, modify the orchestration, however you built it, and get and handle that case for the next time. So maybe eventually you get rid of the human, but you can’t do that right away.

As we see AI becoming more and more central to IT operations, how can organizations ensure their infrastructure is both secure and compliant while still being able to leverage AI’s full potential?

So as I mentioned earlier, security and compliance are an area where AI is never gonna be responsible.

The good news is that what I’ve already said about not taking a human out of the loop aligns well with a lot of compliance requirements.

So your security needs and your compliance requirements are going to drive, you know, how how much you can automate and where you still need to have humans in the loop.

The good news is security and compliance are ultimately going to cost you less over time because of AI enabled tools on the security and compliance side. So remember, AI is working its way into both the systems the business systems that you’re trying to AI enable and make more efficient.

AI at the same time is working its way into security and compliance tools for helping you to to understand compliance requirements, to document what you’re doing against those compliance requirements, right, and to ultimately achieve compliance through some kind of review process, AI is going everywhere. Right? So so the good news is that automation on the business side, right, may be subject to a security or a compliance requirement that you have to meet.

You’ll be able to to achieve that compliance more easily because the what you did on the business side will be more easily documented by AI to document what was done and provide that to the security and compliance people to need to understand how things work.

Right? It’s an interesting, situation where, you know, I started our discussion today talking about the importance of not forgetting how things work.

One of the tools that you can use to not forget how things work is AI.

Right? Because because AI is not trying to hide anything from you. It will happily tell you how things work if it if it knows.

And so, you know, in our company as an example, because we use AI to build AI enablement solutions, We then also use the AI to document how those solutions work.

Right? Because the AI does a better job than a human could do. Right?

And so we use the AI to produce the documentation to to describe how this AI enablement, this automation solution actually works. So, you know, point being in this in this world of I need to automate this over here, but I still need to stay compliant.

A lot of a lot of what we have to do to to achieve compliance is about documentation, is about, you know, documenting the how things work of this new solution that you’ve created. AI is a great tool for doing that.

And I love how you gave that example of kind of coming full circle there that the human brain is an amazing thing, but sometimes it needs a little help and could use a little help to do an even better thing and and do a lot of, I mean, absolutely.

Look. I’m a software I’m a software engineer, but I’ve written a lot of code in my life. And all software engineers have have the same experience that I’ll describe for you, which is I I wrote a bunch of code. I tested it.

It it went into production. It’s been working fine. Six months later, something breaks, and I’m thinking, oh, I don’t actually remember how that works. I’ve act like, I wrote it, but I don’t but I but I’ve gotta go back and look at the code and refresh my memory as to how the code that I wrote actually works.

AI is actually a great tool for that, and I’ve used it for exactly that. Right? To to review a piece of code and remind me, like, how this works. So, you know, we all need to understand how things work.

We need to not forget how things work. And, you know, the the the talk that I gave to the group of high school students, my message to them was, you’re you’re about to be in a competitive world where AI is involved in everything you do in your professional life.

Who do you think’s got a better leg up?

Like, the guy who understands how the AI works and therefore can use it more effectively or the guy who doesn’t understand how it works and they’re just kinda poking around at the edges and trying to get good answers out of it. But because they don’t understand how it works, they’re not as effective as using that tool. Right? The same thing is the same thing is true in every industry and every discipline.

Yeah. And when it comes to that knowledge, I wanna ask you about that. How how does contextual intelligence improve IT automation? And then how do platforms like contextual dot IO enhance visibility as well as efficiency when it comes to modern IT environments?

So, we think, of course, it’s interesting that that contextual, dot IO, as our product also, you know, means something, because contextual intelligence is what you’re trying to to achieve with AI. And I I I don’t wanna make it sound like that’s too mysterious in but but the real answer is it’s different for every use case. Right? What does that mean for you to have a a contextual understanding of a particular problem, of a particular business process, automation, whatever?

That’s actually what large language models are doing when they use retrieval augmented generation. They’re trying to give the language model context.

And by the way, that’s why it’s called a context window. Right? The context window being that that window of information that you can pass into the model for it to consider when responding to you.

So with that, looking ahead, what trends do you see shaping AI driven IT infrastructure in the next, let’s say, three to five years, which is an eternity, I guess, when when it comes to technology. But what should IT leaders be doing now to prepare for these changes that we know are going to come?

Build expertise in AI, not in how the model works, but in how to use it. Right?

Don’t assume that the AI is gonna do that for you. Right? You need to understand how things work. I think, but also don’t don’t underestimate. It’s interesting, Michelle, that this is the Hitchhiker’s Guide to IT because I think it was Douglas Adams that actually said that technologists tend to overestimate how much things will change in the short term and underestimate how dramatically things are gonna change in the long term.

As a wrapping up here, Jeff, if people wanna learn more about contextual dot I o, where do they go?

Well, they go to contextual dot I o.

I guess I kinda set that up for a pretty easy You walked well, you walked right into it.

A big thank you to Jeff Hagen, CTO of contextual dot I o, for sharing your insights on how AI is really reshaping IT operations and infrastructure and some great insight there, Jeff, from your wealth of experience as you told us about the beginning of the podcast. Only to be very interesting to see where things go from here. But thank you for your time today. Really do appreciate you being here.

Absolutely. Thank you, Michelle.

And I wanna thank all of you for tuning in and listening to the Hitchhiker’s Guide to IT brought to you by Device forty two.

If you enjoyed this conversation, be sure to subscribe for more discussions on the latest IT trends. And for more information on how Device forty two can help you gain visibility and control over your IT environment, you can visit their website. Thanks for listening, and be sure to tune in again. We hope to connect with you on another podcast soon. I’m your host, Michelle Dawe Mooney, and we’ll see you on the next one.