Helix: Monetization of AI

We never give up to fight is a must. We going in, in, we bringing the fire. We about to combust. We ready and willing to do what we gotta do. You cannot stop. We put in the time we Hello folks, and welcome. My name is James Hilliard and this conversation is part of our connection Helix Applied AI series.

What we're doing is we're taking buzzwords out there and turning those into practical playbooks. Hopefully give you some good information here today. Not only to learn, but also maybe come up with some new questions about our main topic, which is the monetization of AI. We're gonna look at some tangible ways that you might be able to deliver value.

Maybe it's from new revenue, maybe it's cost savings. We're gonna talk a lot about that. Uh, maybe it's that accelerated time to value. So a lot to go over. Really happy to have Jamal k with us here today. He's the Chief Growth and Innovation Officer here at Connection. Good to see him. A friend. Good To see you, James. Always.

Yeah. We, This is thick, thick paper here. There's a lot that we could go over. Bottom line as we go through the conversation here today, again, we'll hit a lot of high points, hopefully giving you all the idea that I wanna follow up. I want to talk in more depth with, with someone. So that's what we'll do is as we talk about things, one of the things that I wanted to start with is a, a almost a yes or no question.

Mm-hmm. And it has to do with the, the monetization. There are individuals out there that are looking at AI, they're working on some projects, and I've seen numbers that 80% of folks that are working on something never even gets to production. So it begs the question, is anyone out there really making money on AI that don't have millions and millions of, uh, shares of stock and major companies that are the foundation of, of this era that we're in?

Yeah. No, that's actually a pretty tough question to answer. It's not a yes or no, uh, in some ways, but I will say for the sake of being exact into the rules you set. Yes. Yes. Okay. Yeah. Again, simplified. We, we will, we will go deeper, but is is the Yes. Say that there's new revenue and that's how they're making money.

They're making money because they're saving cost, or which are easy to put on your profit loss sheets, right? You can quantify those. Or is it well, we'll be a little more efficient. We might get better talent that might lead to something. So that's where the monetization is. And The answer is yes, both on both ends.

Right? So you and I think that's the classic situation, which is soft dollars versus hard dollars. Soft dollars being more cost savings driven, or an enhancement in productivity driven approach. And the hard dollars being where you can have, you know, new revenue lines or, um, have the ability to cut costs from, um, you know, resources that you might be leveraging, uh, currently.

So I, I think both ends of the spectrum work, both the soft side and the hard side, Um, monetization, let's talk mindset. Mm-hmm. Any organization, small, a medium, large enterprise out there is our mindset, Hey, we need to make money with AI in 60 days. Should it be we're gonna be doing it in the next few quarters.

Should it be a long play that we'll make some money in two or three years? That'll be fine. I, I think that's a really fair way of looking at this. And I think the alignment with how we would look at it from an artificial intelligence perspective is the short, medium, and long term, right? Mm-hmm. So from a short term perspective, and it's also given the maturity of where artificial intelligence is, you can make an argument that in the short term, your classic soft or hard dollar monetization approach would be productivity enhancement.

So it's almost like the return on employee proposition. How do I take my existing workforce? How do I make them more efficient? Give them a set of co-pilots or some other tool sets, make my developers more efficient. So you're looking at the ROE model, which is not necessarily, it's not return on equity, it's return on employee.

Mm-hmm. That's the, my approach. And then you're looking at the medium term, which is return on infrastructure. Now you're starting to say, okay, my workflows within the business, how I run my organization, uh, I need to sort of map those processes out. And maybe I wanna bring in a few agents that can automate certain things.

So you're structurally, you know, changing the, the, the mechanics of your business and you're bringing in a higher level of automation and so on and so forth. That's your medium term play, uh, from a monetization soft and hard dollars perspective. And by the way, we'll have a conversation on how the service providers and so on and so forth, uh, and, and product manufacturers impact that as well.

And then on the upper end of the spectrum is the return on future, the ROF. And, and that is a mindset and approach, almost a strategic alignment where as a company, you've made the decision and something often board driven down that we're gonna materially change how we run as a company. We're just going to come up with new models, come up with new revenue lines, uh, just fundamentally shift the core construct of an organization.

Right? And now you're looking at the longer term, the auto F approach. When folks acted and thought in that manner in the past, there was a much longer runway to get there. The changes and the advancements that we see in AI are so quick. How long is that? Can a company say, okay, we're fundamentally gonna change things.

We're gonna look towards the future. We'll make money here off AI in three years. If they have that mindset. Are they outta business in three years because that's, that's too long? Do they have to be looking at 18 months? Do we have to be a little more focused on more near term future? Uh, Again, difficult question to really pin down, but I think in your premise and your core thesis is absolutely accurate.

I remember six, seven years ago when we were going through a process of, of, uh, enhancing some of our tech stack. Um, I had this conversation with our software engineers and developers, and I would sort of make the argument that the world shifts every year. And so when you're looking at the classic software lifecycle approach, products were being built out seven, eight years ago through the waterfall and mean early stage Scrum and, and agile software development as well.

But the whole notion there was the argument that we would make is that the world's compressing the timelines on how fast you deliver products are compressing. So if you've got a product timeline that's a two year timeline, um, by the time you get to the second year, your entire premise on building that product might have shifted.

So it becomes ineffective. And so we used to start shifting our timelines down to, look, the world can shift on us every six months. And that's kind of the, the calculus that we apply now. Um, which is, yes, it's important to have the longer term horizon 'cause you've gotta have the north star in terms of where your business ought to be.

But in terms of your cycle, anything beyond six to 12 months is, is a stretch. I think so much change can happen that I think you've gotta recalibrate. And, and so from our perspective and, and the guidance that we give our customers is, you know, you've gotta do this constant recalibration of a much smaller cadence.

Someone just heard that and someone out there is like, that's, that's, that's short. And I'm not used to that. How have you seen leaders and companies make that adjustment to still have a North Star still be able to look at that, but actively work better? Are are, are they doing a good job of it when you sit down and talk to them?

Do they have that? I don't know if we can, but they learn that no. Okay. We can, we can make the adjustment here. We, we've been here, I Use the analogy, it's almost a schizophrenic approach to how you run your businesses in some ways. Yeah. And there's this massive oxymoronic approach where, um, you know, you, you've gotta bring that long-term view, the longer arc on what is materially changing your business.

Mm-hmm. And usually as human beings, we are better geared towards shorter cycles of change where that cadence and frequency is something that's more naturally aligned to us. Yeah. But it's the longer arc that we often miss. And I think from a leadership perspective, it's very important to be really well honed in on what those longer arc shifts are.

'cause that's fundamentally gonna change your business. Um, but at the same time, be able to react appropriately to the shorter arcs as well. Do do companies and senior execs do it? I think, again, depends on the company and the culture. Um, the, this approach often lends to this notion that, you know, you can take some risk, it's okay to sometimes stumble if you stumble forward and, and learn and improve, I Stumble Quickly and stumble quickly and, and run through that cycle.

The challenge in any sort of ecosystem, which is especially in the public ecosystem, where you're a public company and you're operating at a quarterly cadence, that often becomes, um, you know, difficult because you've gotta manage your P&L. You've gotta manage some of those elements, your expectations, And those will be scrutinized and Those will be scrutinized. So that becomes a challenge.

So the operational execution becomes a challenge. And I often find a lot of execs go into the proposition, well, understanding the needs mm-hmm. Of having that approach. But then when it comes to the operational execution, they get, they get challenged. We'll talk a little more people, and then we'll get into some more of these, these questions.

And again, we want to lay out a bit of a, a, a timeline and a playbook and, and prioritization. I think that's one of the biggest things that the teams that I've talked about, AI, that's where they're struggling right now. So I want to dive into that. Um, chief AI officer, is that something that I know, I know it's growing, is it gonna be table stakes in your view, in, in the near future?

Or can our CIOs, can our CFOs, can, you know, the teams that are already in place drive this and have the North Star, but also manage the day-to-day to get to this monetization? Or do we start thinking more and more that someone needs to lead this charge focus there? The short answer is yes, with a slight caveat.

So, o there's no reason why the CIO, the existing CIO structure combined with the CTO structure and its underlying organizational support, should not be able to drive an effective AI strategy. There's no, no reason why that should not be the case. But I think even from a CIO's perspective, and I've had conversations with a bunch of CIOs having a partner with you in this journey, which is usually a pretty arduous and tough journey.

And this also includes the data aspect. And you're looking at the CDO equation of your Chief data officer and what that represents. I I think it just helps amplify and speed up the process because there's still a whole broad set of internal challenges, inertia and battles that need to be fought. So having an effective CAIO in an organization, um, that that particular position wakes in every morning and thinks about how to drive a higher level of automation and AI mm-hmm.

Uh, involvement across those three spectrums that I talked about, the auto e, the ROIs, and the auto fs, I think it's, it's a really good resource to have. It drives, it's driven through purpose. It's driven through structure. Um, because the CIO's challenge is a pretty broad encompassing challenge that straddles everything from the minutia of simple IT support functions and challenges to the more sort of organized structure challenges on enterprise scale applications, infrastructure.

So I, I, I think yes, they can do it, it will be a harder nut to crack, but having A-C-A-I-O alongside would just make the process better and Simpler. And if you know someone that has that job title now, let us know, because we know that there aren't a lot of those folks right now. So that brings us back to the, now we've talked a little bit about, about the future, uh, have those long-term plans.

You have to have that North star. What about, and maybe this is a, a good little short example time from you teams that you have worked with that have had a very easy win. Something where they were able to maybe within even a quarter, you know, 90 days go and implement something that led to an improvement on that, that P&L.

Either it was some new profits that they were able to go because of a new product that came out. But, uh, of the organizations and companies you've worked with, what's a little example there to just show that it can be done, it's not all this way out in the future. Things are happening today and and you're involved with those?

Yeah, and again, that depends on the level of maturity that exists within an enterprise. From an AI perspective, some organizations have had almost a decade plus sort of experience with this thing we call artificial intelligence. They've been using it in predictive analytics. Yeah. Uh, they've been using it in, um, computer vision, um, for quality assurance.

Yeah. Manufacturing has manufacturing lot of this absolute so that, that ecosystem exists. So if you look at the mature end of the spectrum, it could be something along the lines of a computer vision based, uh, you know, capability within the manufacturing process. And the reason why I use that example is it's a mature AI technology.

It's been around for a while. Um, the general field of artificial intelligence works really well within the computer vision ecosystem. We have the ability to detect and understand images in a, in, in, better, faster, quicker than humans. And so leveraging that fundamental construct within AI to effectively drive value is something that can be done in a reasonably mature organization that's been working with this, uh, this ecosystem.

And on the lower end of the spectrum, you know, it would be cliche if to just use the terminology copilots, because I think that's, that value still has to be fleshed out. I would sort of codify it more along the lines of conversational access to corporate information to drive, um, more meaningful productivity.

Um, corporate ecosystems have this explosion of data sets and different constituents within corporate ecosystems, whether it's your sales teams or your marketing teams and, and your operations teams, your finance teams, your IT organizations and so on and so forth, need to have access to information all the time that's resident within an organization.

How can they, within a context based mechanism, be able to interact with that data? That certainly enhances productivity, it compresses timelines to execute on certain operations. And that has certainly the impact of what one would quantify as as soft dollar value. And, and, but, so let's go back to us wanting to be able to quantify that number though.

Especially a publicly traded company, we can say, Hey, we got the most efficient people in the world. They're really, really productive. But how do we see that on that, uh, you know, quarterly filing? How do we see it in the K one? How do we see it, uh, you know, out there, Right? So short of, short of filing it in your 10 Ks or things of that sort, um, the approach that one would take, and it's, I don't wanna say it's fuzzy math, but it's still math.

And that's a function of looking at your, um, you know, your delta or your timeline to deliver on a on an operational function. Uh, you look at the, um, uh, you know, the compressed timeline for certain tasks, um, and, and the value that comes out of it, you know, the, whatever the quantifiable dollar value is.

You've got one single resource that now has the ability to do things much faster, right. At higher volume. There's a value associated with that. And then you subtract from that any of the change management costs or any of the AI costs from a tooling perspective or hyperscale cloud cost perspective. And that gives you a ballpark figure on, on sort of the, the classic, uh, you know, real term value that you're extracting out of it.

But this math will evolve. It'll become more mature as use cases become more prevalent. When folks engaged with the Helix team, are they coming in with, Hey, Jamal and team, we're trying to, uh, increase productivity by 25% and we wanna see a 30% reduction in cost. And you start hearing numbers like that and you say, okay, we can help you work towards that.

Or you say, we have to look at those numbers different specifically to your organization. You know, how did you come up with 30% cost savings, and how do you think that AI is gonna do that? Is that something that you have to educate people on? Say, we might have to really just rethink what our numbers are because, because when people invest in new tools, new technologies, often we wanna see that major number that we can write down, right?

Right. And the answer is yes. I, I'm usually very skeptical with approaching projects because that, and projects in, in line with this notion that we're gonna bring in a 35% cost reduction. That's, I'm usually very skeptical in, in, in approaching projects that way. Um, and that often gives rise to an overhyped expectation of what artificial intelligence can do.

Right. Can deliver. Yeah. And then you've got customer sat issues at the tail end of it. Um, from our perspective, you know, we, you know, we, we wanna sort of approach it. We don't wanna boil the ocean with our customers, so we usually give them that guidance that, you know, let's identify two or three good lighthouse projects that have small blast radiuses around them, that if things go south there, it's not, you know, uh, the end of the world.

And, and then let's sort of build incrementally, because there's one element of artificial intelligence and it's adoption within organization that's very important. You can have these theoretical exercises on the value, there's a cultural component as well. Mm-hmm. And the cultural component's, quite often, the most difficult component to, to help navigate organizations around the, the, the classic reticence or, or pushback from employees around concern around automation. And Why do I wanna adopt this and learn this if it's gonna take my job away?

Absolutely. So, so you've gotta really help navigate organizations and employees around that process. And so, which is why coming in with those high fluting numbers, uh, of 35% reduction costs usually often will have a negative connotation with it. Um, we want to enable organizations to be more efficient, more productive, more automated, that they can repurpose their resources for more higher function roles and responsibilities.

I remember there's a buddy of mine who used to sit on, um, the board of, uh, at Amazon and McDonald's and other places. And he, he's, he uses this really simple function. It makes so much sense. He said, the baseline of where human intelligence is has just shifted up. Where what was expected in terms of all of us having this baseline sense of knowledge and capability was at a certain level now with artificial intelligence that is getting recalibrated.

And so from our perspective, that's fundamentally what we focus on, is how do we take that baseline intelligence and efficiency and efficacy of an organization and move that two rungs up. And that's through not just classic AI. That's through, you know, cultural shifts is through process redesign, it's through agent frameworks.

It's through more meaningful machine learning AI, it's through data curation. There's a whole long set of journey, and I think, or enterprises that have the ability to understand this as a long journey of evolution, uh, almost analogous in some ways to the web. If you think about it, there must have been some companies in the mid nineties that were saying, what is this thing called www?

Yeah. I, I sell books. Yeah. I don't need to get on the web. Well, didn't work out too well for them, but that journey of transitioning your business into adopting something like the web in those days, and now adopting something like AI is a long journey. But the difference is it's a much shorter long journey.

Hundred percent. That's, that's the big, the timelines are compressed. Yeah. Timelines are far more Compressed. Is it fair to say that most organizations will not create AI tools to reach a level of monetization? They will go out and rely on other companies that are doing that. And so they need to be choosing wisely, um, who and what tools they're bringing in that are gonna help them towards this? These Are all rhetorical questions.

You are absolutely right. The answer is yes. Yes. And yes. Um, most companies, given the frame of where they're gonna initially apply AI are likely to use existing tool sets. Yeah. Um, And off the shelf, almost Off the shelf tool sets. And then what you're also seeing, and this is becoming more prevalent, existing applications, existing systems themselves are bringing in, you know, enhancements through some module that's an AI enabled module.

Uh, and you see that in the Adobe stack. You see that in the Microsoft stack. You see that in every stack, even ServiceNow, Salesforce, uh, with Salesforce, uh, their genius platform and so on and so forth. Every platform has the, that enhancement through some level of, uh, artificial intelligence. At some point, when that business either going through that cycle or perhaps even being mature, we'll say, you know, there is a need for me to build my own.

And by the way, this is a classic buy versus build. Mm-hmm. Mm-hmm. And so now when you're starting to leverage AI around proprietary data sets, you've got, you know, proprietary workflows. You've got certain specialized tooling and orchestration internally. Now you are, and you're doing some level of, you know, rag, uh, rie of augmented generation, uh, functions.

And so you're bringing in this, this entire new set of process, orchestration, automation, and control. Now you might need to develop something on your own. And so that capacity, and again, it's gonna be a challenge for smaller organizations to have those skill sets, but I think to really get that higher level of exclusive exceptionalism from artificial intelligence will require some level of proprietary buildup. And I'm thinking as you're, you're talking right there, smaller orgs would want to probably focus on the cost savings model of monetization first, which might allow them to have then enough money to go out and hire the folks they need

to now make it a net new product or offering that they're putting out there. That sounds like, because if you just try and go out there and get the AI person now working at your org to build it for yourself, that could be a losing proposition. Those people are in high demand right now. Right. Right.

Um, and you also, there might be, I'm not gonna say charlatan's out there, but they're gonna, some people saying, I know this. Yeah. I can do this for you. And all of a sudden you go down into this rat hole that does not work for you. You've lost a lot of money. Of course your customers have lost face in you and all that. That's A challenge. Yeah. And that's, that's a challenge that's uniquely aligned with organizations with more limited resources.

They've gotta make those bets. And what compounds these challenges against that, again, that compressed timeline and what that represents in terms of competitive landscape shift? Yeah. That they're likely to see, um, you know, the effectively deployed artificial intelligence in corporate environments does give an unfair advantage, just materially unfair advantage to those companies that have adopted it.

And, and that gap, and you and I have discussed this, it's almost like a flywheel. The, you know, the, that initial investment and you're pushing against the flywheel and, you know, mentally picture that analogy. A flywheel is very heavy initially to push, but with every push that initial momentum sort of catches on and it becomes inert the push, push push, and the inertia keeps going.

And so that, that is the challenge I think that smaller organizations gonna have. To your point, the good thing is you're seeing a lot of innovation come out from tool providers and hyperscalers and other organizations. So there still is a significant amount of tooling that can be deployed to get that initial benefit.

But I think at some point, everyone's gotta think about, you know, how do you bring your exclusive ip? Way back in the day I was watching the.com bubble grow in the San Francisco Bay area where I'm from. Right, right. Tons of competition. So much. So it was pretty cutthroat competition there. My observation, I love yours is, are things a little bit different now?

Is there a little more of the AI ecosystem coming together that, that companies are working together and bringing this knowing that so many things, and there's so mo many moving parts here that folks that are looking for that support, they're looking right now at the buy before they're building that it, it, it's, it's not as cutthroat.

I, obviously there's competition out there, but I just see more of an ecosystem and, and teams are working together, and they, they understand more of that. Now, maybe those are lessons learned from the.com era. Um, I don't know if they're lessons learned from the.com era. I think what is abundantly clear for any organization to drive meaningful value?

And I think you, the last three years have been a little bit of a hype cycle, right? Absolutely. Especially with chat GPT, it's sort of in some ways got democratized. It got simplified. And not every resource out there in this, in the world is now somewhere AI enabled in some respects or the other. Um, so I, I, I think what is becoming pretty clear is to drive real business meaningful value is a team sport within the AI proposition.

Because if you think about it, the AI proposition, orchestrates data infrastructure, perhaps hyperscale security policy, orchestration, guardrails, then a stack of applications, some level of ip. So I mean, you, as soon as you start looking at that ecosystem, you begin to realize the complexity. There is no one single organization that you can go and build a relationship with it, and they'll solve everything for you.

Everything. Right. So it is certainly a team sport just by the mere fact that without teaming, you're just not gonna be successful. Some of the tools seem to promise so much. Do organizations need everything that's being promised right now? Are there too many shiny objects in the AI sphere now that folks need to be a little more discerning on?

What do they really need AI for in the organization? Do they need to hone in on we need to make just more productivity that's gonna be our, our North star and our thought? What, what are your thoughts there in terms of Yeah, I, I think that's fair. Yeah. I, I think there are lots of using your terminology, shiny objects out there.

Um, and, but you can sort of, six years ago, I, I remember, or five years ago, um, artificial intelligence, Was it five or six? It was, it was, I think five, five or six years ago, five. Artificial AI had still become, was becoming a big thing. And I remember every product that we would look at as someone who's consuming products Yeah.

And tools. Um, everything had become AI all of a sudden overnight. And it wasn't Right. But it wasn't AI, it wasn't AI chat bots. That's not an AI. Yeah. They were chat box following a prescribed workflow and so on and so forth. There's not artificial intelligence now. I think you're seeing more meaningful AI, uh, be part of products out there.

Uh, whether you need all of them. I, you know, I think that's again, a calculus that organizations make. Yeah. Yeah. Not gonna go super deep tech. That's not what this conversation is about. Again, it's one to, to share some ideas, get you all thinking about things, and ultimately have, you know, ongoing conversations that are specific to what your needs are.

What are you trying to do, what are you trying to monetize, et cetera. But let's, let's go to a little technical, uh, for any of these evolutions, whether it was the pc, the.com, the, the web, now AI, uh, all runs on data. Mm-hmm. And we have to make sure that the data is correct out there. What are some things before an organization, maybe there's someone out there and they're like, we got some stuff we gotta clean up before we can really even think about this monetization.

We want to get there, we want to be there in the next several months. But what are some tangible things that teams are doing? Maybe teams come to you and the Helix team and they're saying, Hey, we want, we want to do this. And you push back a little bit and say, okay, but is this an order? Is it this order? Is this here?

What are some of those things that teams are still trying to get in order so that they can go down with these types of projects and truly make that change on the money sheet? Right. So, so data alignment is often predicated on your use case alignment. So it's reasonably simple exercise that if you go identify, let's say I, for example, a computer vision project and that computer vision project's looking at an end product of a manufacturing process.

And that requires a level of images. And now that requires a lineage of images over the course of the last three, five years. Mm-hmm. And they have gotta be in reasonable quality. And then how do you sort of extract or apply some meta tagging to that data set and then sort of identify the right products, how they're versus failures within that products.

And you now have created enough of a data repository that you can then apply into a, you know, into a, um, a model that you can then train or fine tune or Pune. And now you can at the end of that process, through an inference effort, be able to get the results you're looking for. So I think that's an important consideration to keep in mind is that when you identify use cases, your data strategy will be absolutely aligned with that.

The other thing I will express is, you know, 20 years ago when I was working in artificial intelligence, um, you really had to curate and cleanse and, and construct and orchestrate your data in the best way possible. 'cause it was a very unforgiving exercise. Um, now depending, again, and, and I am trying to dumb this conversation down quite a bit, because the AI ecosystem is very broad from, you know, generative AI to, you know, predictive analytics to, uh, gen and everything else.

Yeah. Other sort of areas of artificial intelligence. So context based conversational access to information will allow you to sometimes have bad data in it, and there's no, you know, and so the AI systems themselves have become a little bit more forgiving in terms of how you manage your data, and then you bring in the appropriate rags and, um, and those can also have, you know, data that's not totally scrubbed.

Um, so I, I think that's the element. So the, the argument would be, yes, there is a data process that has to be adopted, it is predicated on the use case, and there are plethora of use cases that you can apply, and then they will impact your data strategy. But then at the same time, AI tooling has become, um, more forgiving if you don't have, you know, perfectly clean data, uh, that you might have historically, We brought up a little bit ago the, uh, hype cycle.

Mm-hmm. Um, I wanna bring up a different cycle, which goes to when we saw the, the big growth of cloud computing. Mm-hmm. Um, all of a sudden a lot of people started running cloud and started realizing, oh wait, this is a lot more expensive than we thought. We didn't know. Is there a danger there within the consumption of some of the AI tools out there as well, that people, we don't want them.

Right? We, we used to have people do this all the time. They would do the old shadow, it, they'd go on, they'd just drop a credit card mm-hmm. And spool up some servers and, Hey, now we've got this new service. But all of a sudden getting that data back outta the cloud, there was got, are there concerns that need to be thought about there on this AI journey? Mm-hmm.

Yeah. There are concerns and they manifest in those exact examples of, uh, um, you know, shadow AI or overblown cloud infrastructure costs and so on and so forth. But I, I've never been of that school of thought that's, you know, either one or the other. A repatriation or on-prem is the way to go versus cloud.

'cause cloud becomes really difficult to manage and, and there's a cloud bloat or cloud is more efficient versus on-prem. Why would you do that? Because it provides you with, um, you know, the economics of volume, uh, processing. Uh, for me, the function's always been your skillset. It's your skillset that in large measure drives how efficient model you run.

Yeah, sure. And so I've, I've had some really meaningful conversations with CTOs and CIOs, uh, who are just super bright individuals and have a very super bright set of resources that have completely optimized the cloud ecosystem down to the nth degree, where they can burst up and burst down based upon need.

They have serverless callouts, they, they just smooth, smooth, they just have it down. It's so smooth. And I've really tried to rack my brain to maybe make an argument that why wouldn't you consider a certain portion of your workload? And they're like, why would I, I've, I've got everything running down to a T.

Right? But that level of DevSecOps and maturity, maturity is not an easy skill set to have. So, but it's a great north star to sort of chase. Um, having said that, there are still certain considerations. There'll always be certain set of workloads that need to be OnPrem that may have, you know, more security. It Depends on what verticals are, right.

Verticals, the DOD ecosystem, there could be air gaps, so on and so forth. Um, so I'm not saying that there are no needs, but I think depending on which side you choose, whether it's the cloud ecosystem or the on-prem ecosystem, it's a function of your resources. You can get the best resources to make each of those environments, um, as hyper optimized as possible.

And at the same time that hyper optimization can straddle a hybrid ecosystem, which is usually where people will fall. Yeah. Which is there will be a certain set of, um, utilization of the hyperscale ecosystem, and then certain workloads will be on-prem. And I think we'll see a mix Talking about monetization, I wanna talk about the supply chain mm-hmm.

And the investments that people will make in infrastructure in hardware. Um, what are some things to consider? You're starting some projects, maybe you get past a buy stage. Mm-hmm. And maybe some of the folks that are listening are, are getting close to now thinking, well, we might need to start building some of our own.

What do they need to consider there? Because you start investing and all of a sudden we have a major supply chain disruption that could sink a project immediately. What are some concerns? And I, I guess when you're talking about supply chain, you're talking about more sort of the classic supply chain chip shortage, Getting the GPUs that you need, getting any of those Yeah.

Materials that are fundamental to this growth. Yeah. No, look, that's a, that's a fair, fair ex, you know, challenge. And you're right. The, the folks at the front end of the supply chain line are usually the hyperscalers, right. Because they buy in such volume. Uh, even when we, in the past had certain challenges around chip side, you, you often found the cloud providers be the ones that had less of a supply chain issue.

Oh, they got it first. Yeah. Yeah. They got it first. That's how it goes. Um, so yes, I mean, if that's an important consideration, then perhaps a hybrid strategy makes the most sense with the ability to burst up and down. Mm-hmm. Depending upon your utilization and that sort of hedges your supply chain concerns, Safety, security, all of this, uh, AI, it's data driven.

Again, as companies start putting their data into, you know, bot systems or they start building their own, we wanna make sure that that data is not being used in a nefarious manner and it's protected. You and I, uh, have had conversations over the years about all things security. What are a couple things that, uh, need to be considered invested in to protect the investments that we're making in AI so we can make money not all of a sudden lose it, because now we're fighting this major data breach that we had.

Yeah. And again, the, the, my answer would vary depending on your maturity of utilization of tool sets and toolkits and so on and so forth. Um, you know, but there's, there's usually almost three stages to the effort. You know, and the first very well could be, uh, when you're in the early sort of stages, um, of developing some sort of a secure framework is to establish your inventory.

Right? That's such a basic thing to say. Mm-hmm. But what's your existing AI inventory? What are the tool sets that are being used within your corporate ecosystem? We talked about AI shadow there. I can almost guarantee you would be a significant AI shadow that exists within the organization that is out of sight, out of mind, could be in the form of products like perplexity and, you know, open AI, uh, chat, GPT and so on and so forth.

That's being used in potentially your proprietary data is Being fed, is being put in there. Yeah. Fed Into it. So establishing your baseline inventory is essentially the first step, and then would be very quickly followed by your policy, right? How do you define the policy? You can start enshrining some rules for everyone, because I'm not saying individuals make these decisions based out of any insidious, you know, purpose, that they just want to use tools.

Right? They wanna, and so you've gotta sort of give them a pathway of saying, okay, these are the right tools that you can use. Here's the policy guidelines that you should adopt. And if you do so, you are on the right side. Uh, while at the same time having a good sense of, um, you know, this, this ecosystem's moving fast and people's need for tooling and the democratization of tooling is, is important.

The second sort of aspect of evolution now becomes where maybe you become more mature, you're starting to use models. So then, you know, establishing the lineage, the structure, the authorization levels of which models can be used for which purposes would be an important, uh, construct. Having provenance on, on the, all the types of different AI systems used almost like an AI bill of materials is very important.

Auditability is very important. Having red teaming, that's very important. Um, so again, no one single answer. It all depends on how mature you are in that road to AI maturity. And based upon that, there are these different areas that you, you have to sort of drive. Let's get back to, I, I, maybe it's compiling a list because I want people to feel energized leaving this conversation, uh, with some type of hope.

And the way I define hope is hope is great to hope for good things, but you have to work towards making that happen. Mm-hmm. And I think there are a lot of teams that are working towards that, some examples of who has made the money? Is it service providers? Is it the individual application companies?

Is it just the big names that we see come across, uh, you know, the news wires and stuff like that because their stock mates this major jump one day. Right? Right. Is it smaller groups? Where is it being made right now? Yeah. So that's, that's a pretty long response. Big question. Yeah. Big question.

Because, and the reason for that is the AI ecosystem is so rich with so many constituents, right. That as I mentioned earlier as well, that AI is a teaming sport. So if you think about it, the hyperscalers absolutely right. They, the cloud consumption components related to artificial intelligence workloads is certainly going up.

Yes. So you've got the hyperscalers doing a phenomenal job. You've got the, uh, chip manufacturers, the folks like Nvidia and their ability to bring accelerated computing because we're now for the first time seeing a fundamental shift on how computing is being done. It's moving away from the legacy CPU architectures into now the GPU architectures.

And there's a whole set of applications that, that ride on top of that infrastructure. Certainly the Nvidia work, uh, types and other chip manufacturers that are, are absolutely making, um, money and driving revenue. You've got the software, um, providers, the, the folks like the service nows, the sales forces as they're sort of elevating their product lines and bringing in AI capabilities, they're likely to sort of be able to monetize them more effectively.

But it's not just the biggest software Development. No. There's, I, there's a small audio and video editing suite I use. Absolutely. They've now got a little subscription model for me to get new jingles. Right. Right. So that's one end of the spectrum. And you, you've got this whole ecosystem of smaller mid-size organizations, mid-size, we've got glean and you've got smaller organizations with simple subscription types, lovable and others, uh, from a code support perspective.

Uh, and so this ecosystem of applications that are exploding, that are AI applications, usually generative AI applications, that is, That's where there's a lot of little money to be made, but as more subscriptions come in Right. That's long-term revenue. Right. That ends up on the, the sheets. Absolutely. But at the same time, there's a risk that you've got some of these larger organizations like ServiceNow and Microsoft and others will be able to bring in that capacity and capability, those small ones, either through acquisitions Yeah.

Or just build that capacity. And so there, there's gonna be some level of consolidation that happens in that ecosystem. Then you've got the reseller, uh, world, uh, the solutions integrated world, right? And, and they are sort of helping organizations bring all of these disparate parts. And so it's now services reselling model that has the ability, uh, to certainly monetize AI.

And then you've got sort of the large, um, consulting firms, right? And, and they are sort of approaching this, again, in the classic way, one should expect this is a business re-engineering effort. It's a process redesign effort, it's an agentic framework effort. And so there's significant consulting engagements that happen.

So there's a lot of money a lot of folks are making, and because it's a function of the complexity of the constituents mm-hmm. That actually can drive value. The question that I keep asking, which is always a good North star question for any provider, even from our perspective, is are we providing business value to our customers?

Right? And, and because at the end of the day, whether this sort of drifts into, uh, uh, you know, a petering ebb of nothingness as that hype sort of devolves, right? Uh, or does this then amplify and people really see meaningful value in this increases? And that's always a concern that I have. What about outside of the big names?

Because what I'm hearing, right? And, and, and it, it's what I've seen as well, the individuals, the companies that are making the most right now are the biggest out there. Mm-hmm. How does that trickle down? And sometimes we hate using that terminology in finance, but uh, 'cause this is controversy there, but how, how does that trickle down to, uh, you know, I'm thinking maybe there's, uh, someone out there that owns, you know, 12 car dealerships, right?

How does the car dealership individual, can I start make Yeah. Before you go into the car dealership and I, I wanna just sort of put to bed a little bit of a fallacy. Okay. And that fallacy is that the big boys and girls out there, the big companies, right? The, the, the, the big frontier, the household Worldwide Names, the big frontier model companies are making money.

Yeah. They, they're, they're not, they, they are, they're significantly challenged, right? Because the amount of investment that's gone into building that core baseline infrastructure has been enormous and it's continues to be enormous. And so we'll find that the, that outside of somebody like an Nvidia and others, that that certainly is, is providing in some ways the, the shovel for that, that miner, right?

In the classic sense. But the, the one who actually owns the mine is having is, is pretty challenged the big model providers. Um, and that's Risk reward. They're putting it out there. They do have the potential to it. They have The potential of making a lot. Absolutely. And, and, and you'll see, you know, sort of the, the alignment of meta with and roll the alignment of OpenAI and Johnny Ivy and, and Sam Altman aligning you, you can see now that there's likely going to be a whole set of products that come out of these big model ecosystems.

So I wanna put that, that, that fallacy to bet that some of the big companies, they're making tremendous investments, right? They're not, they're Not quite printing money yet at The end. Okay. They have. But I think the promise is significant and in many ways this is the last sort of game in town. Whoever wins this game of owning a GI will essentially then be able to do a lot with it.

Yeah. Um, hopefully A lot of good with It, right? And so then comes the question again, back to the smaller organizations like the, uh, the car dealerships. And that's where the challenge is because AI still hasn't reached a level of democratization that you can effectively think that a single or three or four car dealership group will be able to leverage it effectively, perhaps not, but maybe a more centralized private equity backed car dealership organization will bring in a level of automation. So, Or maybe their monetization is the efficiencies, right?

Productivity, that's what their monetization, right? And that's what they should, or Their marketing outreach, their ab testing, their automation from, uh, marketing, their ability to have predictive maintenance, right? And, and sort of do call outs on that. Um, but that again comes from, uh, a set of investment and resources that usually smaller groups will have, will be challenged.

And I would argue a group like that probably doesn't have as much of that maturity, right? Especially in house. And that's where then talking to your team and talking to others out there and just learning more. And I think that's what we really, uh, at the end of the day, so many of us are just trying to learn so much more in everything we can to see where is our organization going to fit in here?

Are we going to make a new product with AI that will create new revenue streams, right? Or are we gonna save on some cost savings, right. By buying some technology To someone else's. And that's exactly where those three aspects that I initially started off with, the ROE, the return on employee, the return on infrastructure, uh, and the return on future for the smaller organizations, you're driving the ROE component more so, and for the more mature, larger organizations you're driving on the future ROF.

And that's where we would come in and we would sort of arbitrate or have that conversation say, okay, where is that right investment, um, that has a small blast radius that has, uh, an immediate re, you know, return on that investment. Um, that's kind of what we will sort of help, uh, get out of that.

I have more questions. I don't have more time, so we are gonna wrap things up. I do appreciate Sure. Thank you for everything that you, you shared. Folks, if you have additional questions, you want to connect with the Helix team and talk more about your AI journey, ways that you can monetize all the investments you are making now or will be making, then head online to cnxnhelix.com.

You can get more information, can get in touch with the team. Really appreciate you taking time to join us and we do look forward to talking to you all down the road.