AI Doesn’t Transform a Business. People Do. | WWT CEO Jim Kavanaugh

0:00Not only can we respond faster, we respond faster with better quality. We respond in a more creative way. And at the same time, we're scaling from our back office to our front office that is actually driving more revenue. And we're growing over 50 % last year. And then halfway through the year, we're up another 50%. But I truly don't believe we would be having the success that we're having today if we didn't make the investments that we were making in AI and digital transformation of the business. And overall, it's a business transformation. That was Worldwide Technology co-founder and CEO Jim Cavanaugh describing what AI looks like when it moves beyond experimentation and begins changing the economics of a business.

0:48But across the enterprise landscape, that outcome, 50 % growth, two years running, is far from guaranteed. In fact, while 74 % of the world's largest companies now have AI in production, half still can't consistently prove if it's creating value. So what separates investment from impact? In this episode, Jim shares what WWT is learning firsthand about controlling AI spend, scaling agents responsibly, connecting technical expertise with business knowledge, and turning individual breakthroughs into institutional capability. This is the AI Proving Ground podcast from Worldwide Technology. Let's jump in.

1:33we'll get right to it i mean you know i was just reading this morning a research article on forbes said 74 of large enterprises have ai in production we're talking about like fortune 500s forbes global 2000s enterprises like that big enterprises but only half of those they can't say if it's producing measurable ROI. This seems to be a little bit of a theme right now within the landscape. I'm curious from your perspective, you're very plugged in, what is the gap there? Why does that gap between spend and value seem to be widening or is it? Yeah, well, first, I guess my first comment would be, I'm actually surprised if we're talking about the Fortune, did you say 500?

2:12Yeah. Fortune 500, that it wouldn't be at 100 % of them are using AI. And then we would have the discussion around what level of return are they getting, and we can talk about that. But I am a bit surprised that not every Fortune 500 company has some type of AI investment that they're making and going through some process. I think the companies that aren't making those investments are the 25 % really put themselves at risk. So that's one point. The other point in regards to, you know, I would say the variations of outcomes that are happening today. You know, some companies are having great success with their AI implementations.

3:05But I would also say that every company that's implementing AI is having a level of learning as they go through this. And there's trial and error that you must be going through. This is not a perfect, you know, kind of line of sight of here are the top or here, you know, step one through 10 that we need to do and we will drive all the ROI benefits that we ever expected and then some. It's just not that easy. And the models are getting better, as we've talked about, the large language models. However, every company, the reason every company is going through, I would say, different stages of success and different stages of learning and trial and error is because of the complexity of every enterprise.

4:00And how well are they at unifying and aggregating their data and putting their data in, I would say, the most advanced data structure, data lake, you know, in are they writing APIs, you know, into these different, you know, SaaS products. and just overall, organizing their data, aggregating their data, governing their data, and then starting to enable it to be leveraged by AI. And as you do that, then you have to not only integrate your IT organization, but you need to align it with the line of business owners. And then you need to factor in how are you going to automate that and implement the AI platforms and tools to do that.

4:47There's a lot of complexity and there's a lot of complex processes that big enterprises, specifically Fortune 500 organizations have, that require time and significant work from both the business owners, call it advisors and consultants, and deep technologists to bring all this together. So to your point, there will be gradations of success. That being said, I'll go back to my original point. I'm a little bit surprised to see that there are any companies that are not leaning in to AI to some degree. Some are going to do it faster and more aggressively than others. And that's what we're doing here at Worldwide.

5:35And I believe that we're seeing great success as we go through this process. But I will also tell you along that journey, we are having our trial and error scenarios. And that's just the mindset and I think the cultural structure that you need to create, that not every project and program and use case that you look to implement is going to be an immediate success. There are going to be challenges. Yeah, and I should qualify that a little bit. The research was that 74 % of Fortune 500s or Forbes Global 2000s had AI in production. I'm sure the AI pilots, proofs of concepts, whatnot, I'm sure that's close to 100%, as you know.

6:23But to your point, we are all learning here kind of on the fly on how to handle AI and implement it and scale it within the enterprise. But at the same time, spend is rising quite rapidly. I know our spend here at WWT is rising. How do you guide your teams and, you know, by proxy, how would you guide other CEOs to balance responsible fiscal spend versus driving innovation when there's a lot of learning to be done, but at the same time, there's cost to control there? Yeah, it's going to be a bit of an art and science, I believe. You know, when I look at what we have done internally is by really just pushing the entire organization, you know, with messaging from myself, the CEO and the line of business owners and the other leaders throughout the organization with a goal to drive, you know, as close to 100 % adoption from the no coders, you know, the low coders and, you know, the pro development, pro coders, full stack developers.

7:27So everyone in the organization plays a role and needs to take accountability in regards to upgrading yourself and training yourself on how you are going to actually make yourself more valuable. How are you going to upgrade your own skills and capabilities to leverage these new tools and capabilities around AI initiatives that we have? So that is one that we have pushed in a very aggressive way. But to your point, we've also been very aggressive about rolling out these tools. And I would say we're probably more aggressive than most organizations because we have looked at rolling out, you know, different AI coding tools and we've got a, you know, a smorgasbord.

8:19What I would say, there's more concentrated in certain areas of these coding platforms that we're using internally. And then you have some of, I would say, the business use, whether it's a co-work, it's a co-pilot scenarios that we're using in here that could be on more of the business owner. So all that being said, we have distributed the tools and the access to the tools in a pretty aggressive and I would say free-flowing way. We will start as we see how they're being used and where we feel they're being used in the most impactful ways. We will continue to engage in that area and approve that kind of structure.

9:03There are areas that we're learning today that need to be curated. We need to pull back some of the tools. Not every coding application needs to have a fable or the most advanced frontier model. And so in that process, we're learning a lot and we're going through that as we speak today. And every customer that we talk to today is, you know, is moving forward around these AI initiatives internally, whether they build a complete AI factory on-prem like some of our large customers and neoclouds and large enterprises, or they're doing it through, you know, some of the frontier models and some of the open source models in the cloud.

9:47They're moving forward and they're looking for help in regards to how do they manage the tokenomics? How do they manage the spend that they have today? And again, I would say no one has this perfected. We are also spending a lot of time with our consulting teams, our AI advisory teams, our data scientists, software developers, and using ourself as our own guinea pig in regards to the routing. intelligent routing of the different tools and who gets access to what tools, where you're not leaving it up to the individual to make that choice, because that's a big decision and a heavy load. So a lot of that is happening as we speak, and we are learning a lot.

10:37And then as we share our learnings with our customer, we're also learning from what our customers are doing. So The spend is something that I would see will continue to get better. As the tools get better and as the automation gets better, the intelligent routing gets better, the spend will get better, and the AI use cases will become even more and more, I would say, meaningful from a return on investment standpoint. Yeah. It's interesting you said guinea pig because one of the questions that I had kind of teed up here was you seem somewhat unwilling to let us learn about AI secondhand. When somebody brings up a demo or just talks about a use case, I've heard you say multiple times, well, let's try that here at WWT.

11:25Why is that firsthand experience and experimentation, why is that valuable for our organization? Well, it's valuable for our organization. is also incredibly valuable for me because honestly, we can cut through where people may be blowing smoke and there may be a little BS in regards to what someone's doing or not doing, what tool actually works or doesn't work. And there's nothing like actually challenging our internal organization to put that into practice and to see, does it really work? Because there's a lot of things that may work in the cleanest and call it single-threaded, call it test cases, where it's not integrated into a fully, call it complex enterprise environment that needs to be integrated into other tools, other applications, additional data.

12:24So when you bring that in to actually apply it in a real enterprise situation, you get a real understanding of how these different things work. And so I believe worldwide as an advisor to our clients, we need to make sure that we're not just going out and regurgitating things that we have learned, you know, white papers that we've read. We are learning and providing guidance and advice from real world experiences that we are actually living and breathing internally and living and breathing with other customers. So that's where I've, you know, from day one with everything that we were doing with AI, I'm like, we're going to be an AI first organization.

13:13We are going to be AI first at every level of the organization. And we're going to learn and aggressively learn. And that's where even looking at where Worldwide is from a cultural standpoint is that if you don't believe that, you know, you need to individually and we need to as a collective organization lean into these tools to be the best that you can be and to actually improve your own capabilities and organizational capabilities, you're probably at the wrong place because that's the mentality we have here. Yeah, I mean, what you're talking about is essentially continuous learning. I know that's been a value forever here at Worldwide Technology.

13:57Is learning becoming an increasingly bigger differentiator for companies even more so than it was in the past? Or is this just always what's been? If you're learning, you're advancing. I think it's what has been, but not at the degree of, I would say, comprehension. And I mean, but comprehension is things are coming at us so fast and furiously. It is difficult to keep up. It's difficult for me to keep up. And I have the benefit of being able to pull in a lot of very, very talented people to get their insights. But it's a bit overwhelming to sit here and try to stay on top of everything that's going on because the technology is moving so fast and it is so compelling, more compelling than anything I've ever seen in my history here at Worldwide Technology, 35 plus years.

14:57But it's also equally exciting. So I think if so important to have the culture and the mindset that, you know, look at this as try not to be overwhelmed, but also to make sure that you have this level of excitement that it can be fun to learn and exciting to learn. But it does require a lot of hard work to stay out in front of this. And the other piece I would say, because it covers a lot of ground to actually make enterprise AI work within a customer environment and or build out an AI factory, it requires a lot of subject matter experts with deep skill sets in certain areas to come together as a collective team to build out these capabilities.

15:48So, yeah, that mindset of continuous learning, it's just not going to stop. I mean, if anything, just get your mind right that that is a space that you're in. And I think we all have to also have this mindset of giving us collectively, organizationally, individually, a little grace in regards to you're not going to get everything right. And if you expect to get everything right, you will kind of put yourself in a state of paralysis through analysis. You just don't have that much time. So you need to go try things and through trial and error, you will get better and then course correct as you make those mistakes and learnings.

16:35So if you're talking about frontier AI models that are rising in capability, but also rising in risk, we're talking about, you know, agents that are, quote unquote, jumping out of their sandbox and going in, hugging face. You know, we've all seen those those headlines. So how do you, as a leader of an organization, determine who gets access to those frontier AI models when you're saying you got to learn, you got to learn fast, but there's also a level of added risk here? Yeah, it's a very real deal. You know, that I think there is this blend and it kind of goes back to the point, you know, responding to your question is even more so the need of team of teams.

17:16And where I'm going with this is that we need to be promoting a team when you talk about pro coders and you look at like there's those basically pontificating two years ago that software developers, because of AI, are going to be out of a job and all the coding is going to be done by the platforms. Well, one part of that was 100 % correct that a lot of coding is being done by the AI platforms that are going on. However, to actually make a lot of the code work in, I would say, complex enterprises and actually bring it together to really deliver at scale, you need software developers. And you need developers that are not developing the way they used to develop, that they're developing using these frontier models and they're using these AI coding tools and platforms, IDEs.

18:13So they need to be trained. And where I'm going is that at the same time, those models that were built to actually solve problems and write code and create capability around incredible AI innovations have also been able to hack into companies. And so where I'm going on these team of teams is the developers and the data scientists also need to be working with our cyber teams. And they need to be working very collaboratively in regards to how to move fast and write code and think about, you know, software development lifecycle process that you're going through. But also, how do you do that with the right guardrails, with the right security in mind.

19:00And that's where that kind of collaboration, when you bring those team of teams together, you can do some pretty amazing things, you know, thinking about how you're going forward and innovating, but how you're also doing it in a very, I would say, secure way with the proper guardrails and also using these models to help protect against the bad guys, the bad actors. So that's where I think so much of this is bringing individual experts together and teams of teams together. And the better you can do that, we're going to create capability that are going to be incredibly beneficial for large enterprises because there's so much overlap.

19:42And the other piece, it's very similar to if you think about, since you brought up some of these things around some of the challenges around cyber, it's also the need to go in and look at the tech debt and the infrastructure of organizations and going in and looking at, so how do you come in and upgrade potentially legacy network infrastructure, data center infrastructure storage, but also looking at and using the tools to go out and assess, you know, what about some of the software that you have that you need to be, you know, thinking about patch management and upgrading, you know, a lot of the software.

20:22So if you're trying to do that the way that you've done it in the past without leveraging the tools, it's really – it's not a fair fight because you're going to have bad actors coming at you with tools and capabilities that you're just not going to be able to defend and protect yourself by using tools that were used six months ago, nine months ago, and five years ago is definitely not going to work. Does that change any of the math in your head on how quickly we should be adopting or deploying agents? Because, you know, by the last 12 to 18 months was like agents, agents, agents. And now this happens and it's like, okay, well, does that change the math at all?

21:05Yeah, we were having quite a discussion around, I don't think all agents are created equal. And there will be agents that are given a lot more access, call it, to compute and authority to go do things. I think when you're building agents, you need to build them very thoughtfully and they need to be managed in the right way with the right guardrails internally. There can be agents that you have that may be single-threaded, an agent that is kind of task-oriented to something. There may be agents that actually are more sophisticated that are not only task-oriented but actually can do reasoning that is a bit, I would say, more complex in the scheme of things.

21:57And then you may have agents working with other agents and swarms of agents that are actually, you know, doing tasks, but also you're linking tasks together and you're having them collaborate in a way. And that is obviously a more advanced capability. So depending on what you are doing with these agents and how you are activating them, I think has a big impact on the situation or the security guardrails and some of the cyber things that you need to be thinking about. Is that something that a CEO or an executive should be taking on to try to figure out that visibility? Or are you just very intensely interested in that right now?

22:43How would you guide other executives? Yeah, I would say no. Just general CEOs, no, that would not be your job to do that. But I believe what your job would be is to give the guidance to your organization in regards to making sure that you are doing everything you can to provide the right tools with the right people. and being on the front foot in regards to rolling those tools out and training and enabling all of your people from, you know, the deep technical experts within your organization. They need to have access to those tools all the way as you progress up into the middle management of the organization and the actual, you know, doers of your organization.

23:34They need to have access. And you need to be going through that process of evaluating these use cases all the way up to managers and leaders that are working with IT, then for them to really understand how are these being made? And my point being around why I'm digging into those is because I want to understand what those agents are doing and what are we learning from those agents that we can recreate in other use cases. And then how do you string those agents together to actually create a multiplier effect of capability and value and do it with additional efficiency? So no, I would not expect a CEO to be leaning in, but I would expect a CEO to be enabling his organization to make sure that the IT leaders and the line of business owners are coming together to understand what they're building and then being able to communicate that to other groups.

24:36Because that's part of what we're even doing internally is where we are having success in a number of areas is how are we making sure that we're communicating that to others and showing them how that is being done today and the value and the impact that can have. What is that playbook for taking what people have learned individually or what teams have learned on their own and then extrapolating that into institutional knowledge? What I try to communicate to CEOs and leaders that I'm talking to is this journey, this AI first, AI being an AI enabled organization, it's a business transformation. It's not a technology project.

25:17It's not a business project. It is a business transformation because you're bringing together so much here. And the reason I say that is because I truly believe it can have a monumental impact on the organization in regards to how you are actually growing and scaling your organization. So if you think about everything on the back office capability, how do you provide better tools and capabilities to the individuals that you have within your organization to allow them to do more, to do it faster, to do it with a higher level of quality outcomes, and to do it with scale? But also what happens there is when you do that, it also provides a capability that links to the revenue producers, the sales organization, business development organization, marketing organization, that they can start doing their job in a more cost-effective way, in a more scalable way, and they can start to innovate and create new products and new capabilities.

26:26When I step back and look at things that we're doing, and it's not all associated with just AI enablement, but it has a lot to do with it, AI enablement, digitizing our business. But we're growing over 50 % last year, and then halfway through the year, we're up another 50%. So the, you know, one, it's a big market that we're going after, but I truly don't believe we would be having the success that we're having today if we didn't make the investments that we were making in AI and digital transformation of the business. I feel like there's a lot of attention on what's being produced at that kind of cutting edge part of the AI landscape.

27:09If you were speaking to kind of our customer base that doesn't necessarily always have the time to understand what's unfolding there, what do they need to understand or know about what's coming out of these frontier labs or what's possible or what's more efficient potentially with open weight models? What's signal there that they should be paying attention to? Yeah, I guess I would first, with a client, just step back and say, I think it's so important for the level that you're speaking with that understand the importance of leaning in and this being CEO-led and executive-led from a communications, a messaging, and a methodology of how you're going forward.

27:49I think that is just so important to send that message throughout your organization that if you don't lean in, you know, individually, you know, you're going to put yourself at risk as an individual, your value, you know, even Brian with what you're doing and all that you're learning through, you know, the podcasting you do. And I mean, you've what you've learned in the last 12 months is probably mind boggling. You probably, you know, you don't even realize all that you're learning. But at the same time, we need to bring that together organizationally to create a multiplier effect of learnings that we have.

28:21So that mindset, that communication, I think is so important. And also just the recognizing because things are happening so fast right under our nose and our eyes that we're seeing that I think we underestimate how significant the investments around AI are happening. And meaning that, you know, I've talked, I think, even on the last podcast, is just the amount of CapEx investment that's going into the build out of the infrastructure, even from the hyperscalers. It just dwarfs anything that we've seen, you know, society, you know, or any enterprise businesses, you know, even think about or contemplate.

28:59And just when you think, you know, wow, last year coming out of 2025, going into 26, the CapEx so big it could not, you know, my God, are they actually going to execute against that? And it's got to, you know, just start going down. It's going up. You know, you're looking at the hyperscalers. They're saying it's going up. We're actually compute deprived. You know, they want more GPUs. They want more CPUs. They want more capacity. and the model makers and the hyperclouds and neoclouds, hyperscalers and neoclouds are trying to build out as fast as they can. So there's not enough energy, there's not enough compute to keep up with the demands that are happening.

29:41So I think understanding that. And then as recent as yesterday, you have Jensen bringing together five or six of the largest, call it financial institutions along. And these are just some that he brought on to CNBC to talk about this$500 billion investment to help further enable the build out when you're talking about working with, you know, organizations, you know, you know, like Apollo, like BlackRock, Blackstone, you're looking at Brookfield. These are, you know, organizations that are, you know, that are helping enable the build out, the financial backing to enable this. to see those coming together.

30:26And do keep in mind when I talk about the CapEx on the hyperscalers, and then you think about the organizations I've just mentioned, Goldman Sachs, you know, these are some of the most intelligent minds in, you know, the financial ecosystem and technology ecosystem that are so far leaning into everything AI that everybody's constantly looking for this bubble. And rightfully so, it's a healthy thing to do, but we're just not seeing it. And what I am seeing is more and more doubling and tripling down on what we're doing. So I think stepping back and understanding that is a very important piece. And then the messaging of, you know, how you're communicating to your organization, how important it is to be an AI kind of led organization will then help, you know, then as you go down and you think about the different models.

31:25And that's where, you know, it will take, I believe this takes an ecosystem, you know, as we work with our partners and we help consult and advise to our customers, the customer cannot do it, it just can't do this on their own. They need to bring in very, very solid, knowledgeable advisors and partners that they can trust. They can help work through this process to make decisions on whether you go frontier model or you're actually looking at an open source model. And some of those things, they will continue to evolve and we'll figure those things out. But right now, it's not an easy button just to push and say, oh, I'm going to go this route or I'm going to go this route.

32:15It's evolving as we speak, and it takes, I would say, a collaboration of a smart, trusted team to actually drive those best outcomes. I saw a fun graphic. I think it was on LinkedIn or something. It was somewhat tongue-in-cheek, but it always feels too early until it's too late. So the margin for error on either side is very slim. Well, I mean, Jim, it's easy for us to focus or we're an AI podcast. It's easy to focus on systems and models and bubbles and things like that. But we haven't really talked about people yet. You've been consistent on in the AI world, the value of human expertise, judgment, taste and so on is rising.

32:56But I am curious, have you have you changed your thought on kind of the value that a human brings to the table over the last 12 or 18 months? You know, I don't know if I've changed. If anything, I think the human in the loop is more important than ever. Because if anything that I've continued to learn that this stuff is complex, it's complicated. It requires really smart, passionate people to figure out how to put the tools to work and actually to start enabling these capabilities like agents of agents. We're in a very early stages of that. The aggregation of, you know, data, it's really complex.

33:47You know, Brian, you know, I've asked our CFO, which I've gone to and I've been working with Tom for 30 plus years, Tom Strunk, and I've asked him to take on the role not only as a CFO, and he's not getting paid twice, but also the job as our chief data officer to work with our AI teams and our AI business transformation office. because he can uniquely help accelerate the aggregation of data and he can help, I would say, be the tiebreaker in situations where it's not necessarily a technology challenge of enabling this AI use case. It's a decision around who do we give access to that data and somebody that really understands the business and the positive ramifications of enabling that data and potentially the risk of enabling that data and making sure we do it with speed but with the proper guidance and guardrails.

34:57So he is spending a lot of time and his own personal words is coming back to me after three or four months. He goes, I didn't know what you were even thinking about asking me to do this with the demands he already had. But he has come back and said, it's actually very exciting because he is in the middle of some things that he would not have been in the middle of that can really help advance the business forward. and it does require somebody like that that can help make those decisions. So that's, I would say, looking at it, even at the highest level, you gotta think about what is the role that people play and they play a very, very valuable role.

35:41And I also see that within people in very operational areas, as you know, I'll go through and get an update, multiple updates on Monday through three or four different meetings I have, but on different use cases and to see individual business operations people articulate what they're doing around AI and the automation and the capabilities that are being built out to scale out their organization. In a lot of cases, I know we would have had to add significant incremental people in specific areas, whether it's in finance or it's in marketing or it's and sales operations that have just allowed us to scale those groups out.

36:25And I'm learning from those groups. So people are incredibly important to this. And that's where I think the ones that lean in, they're actually improving their resume and their own personal net worth because they are doing that. The ones that are sitting back and not really engaging are the ones that will put themselves at risk. Yeah. To go back to Tom Strunk and him taking on the chief data officer role here, what's the lesson learned for organizations out there right now? Is it that you need to have somebody that's just finely tuned and aware of what the business is doing and is well-versed and been there for a long time?

37:06Or is it more somebody that has an operations mind? Or what is some of the value or the lessons learned that you can say like, hey, here's why we put Tom in this space. And here's the, here's the lesson that you can learn from that. Yeah. I, I, I, I believe it's a, it's a combination of things on one hand, you know, I even look at it and I'll digress a little bit here, but on one hand I'm like, you know, we had a group of 250 interns in and I'm like, I like to exercise those interns. And we had capstone projects by giving them access to the coding platforms and a lot of the different things that we're doing because they don't have any type of, you know, call it tech debt or mind debt in regards to how something should or should not be done.

37:54So you want some individuals that are coming in with kind of a greenfield, you know, clean paint a glass in regards to how they're looking at they're not held back with anything. Then at the same time, when you look at someone like Tom, you need somebody that truly understands how the business runs, what's important to the business, where is the data, what systems are the data, is the data locked in, what data structures, and then it's who should get access to that data and how should we set that up. And there are business decisions that could be made. It could be around pricing information. Do you deploy pricing information across everybody or does it need to be siloed in specific areas?

38:43Or can you spread some of that to provide better insights, but the data isn't anonymized, you know, in a way? So there's all kinds. I mean, as you dig layers deep, complexities that go into it. But as I go back, I think you need a blend of all of those types of individuals. And that's where I think the blending is more important than ever, that this blending of business owners need to take the time to understand this AI transformation and what these tools can do. You don't need to be the deep expert in every area. You bring in people. At the same time, your software developers and your experts need to understand what the business needs from them and the impact it can have.

39:31because the better both of them can learn more about each other's job, the faster we can come together and the better solution. Jim, we're taking up plenty of your time here, so we'll wrap it here. Thank you so much for taking the time joining us in the studio. I'm sure we'll have you on again soon. You got it. It's always a pleasure and it's exciting, you know, and just sitting here and talking to you and listening to the different guests that you've had on. It's another one. It's like finding the time to do, you know, all the things. And I need to get back and listen to every one of your podcasts.

40:00I think you can listen up to two times speed. That's what I do. Sometimes it gets a little crazy. But good seeing you. Thanks, Jim. Okay, as always, thanks to Jim for joining. My takeaway, enterprise AI creates meaningful value when leaders treat it as a CEO-led business transformation. Combining aggressive experimentation, continuous learning, governed data, and human judgment to reshape how the organization operates and grows. This episode of the AI Proving Ground podcast was co-produced by Nas Baker and Kara Kuhn. Our audio video engineer is John Knobloch. My name is Brian Felt. Thanks for listening, and we'll see you next time.