#180 Thomas Lah: Navigating AI Adoption in Tech Businesses

0:00Tech professionals, they're trying to figure out how do I apply AI. And one of the muscles they have to build is how do you evaluate this plethora of tools that have come out of the woodwork? And, you know, just talking to different folks, that is, again, a skill that you have to mature so you can kind of poke through and say, okay, that tool looks really cool, but is not going to scale for us, not going to be practical. This tool over here, much more practical. So that's still early, early days, but it is pretty amazing how rapidly it's maturing. Hi, I'm Craig Smith, and this is Eye on AI. In this episode, I speak with Thomas Law, the Executive Director for the Technology and Services Industry Association, TSIA.

0:41While you assume tech companies offering AI-powered products are way ahead of everybody else in using AI in their operations, it turns out they're having as much trouble as non-tech companies. Thomas delves into how AI is transforming internal processes such as content development, support services, and field services, and discusses the challenges and opportunities tech companies face in integrating AI into their workflows. I hope you find the conversation as informative as I did. AI might be the most important new computer technology ever. It's storming every industry and literally billions of dollars are being invested.

1:28So buckle up. The problem is that AI needs a lot of speed and processing power. So how do you compete without costs spiraling out of control? It's time to upgrade to the next generation of the cloud, Oracle Cloud Infrastructure, or OCI. OCI is a single platform for your infrastructure, database, application development, and AI needs. OCI has four to eight times the bandwidth of other clouds, offers one consistent price instead of variable regional pricing, and of course, nobody does data better than Oracle. So now you can train your AI models at twice the speed and less than half the cost of other clouds.

2:13If you want to do more and spend less, like Uber 8x8 and Databricks Mosaic, take a free test drive of OCI at oracle.com slash IonAI. That's E-Y-E-O-N-A-I, all run together. So, Thomas, go ahead. Can you introduce yourself? Yeah, absolutely. So I am Thomas Law. I serve as the Executive Director for the Technology and Services Industry Association. We are a for-profit research institute. We've been around almost 20 years. We focus on technology business models. We get companies basically under NDA, and we do a lot of benchmarking and research on their operating models. Yeah. And you're talking about companies that are using technology or companies that are providing tech infrastructure?

3:02Companies that are providing basically enterprise technology. So if you think about a Microsoft, if you think about a Cisco, a Dell, all those companies are members, Salesforce. And what's also interesting in that is more of, you know, older companies, think of somebody like a John Deere as they start to get into AI and software. Rockwell Automation, Siemens, all those companies now are members of TSIA. And again, our lens is if you're providing hardware or software, what is the operating model around that, how you monetize, selling product versus services, all that kind of fun stuff. Yeah. And so do you guys operate as kind of a consulting company or just as a place to exchange information among members?

3:47Yeah, it's a great question. So our model is to do operational research, play that, you know, those insights back to members to help them implement best practices, optimize performance metrics. We do light advisory work. We very deliberately do not get into, you know, big, heavy consulting. That's not our model. We want to have a one-to-many, get people to answers as quickly as possible, and quite frankly, avoid some of the exploratory consulting you have to do to get to a right answer. We were like, hey, we already know the right answer. We've talked to tons of companies on this. Here's what you should be pursuing.

4:22Yeah. And what's your background? How long have you been at TSIA? Yeah, yeah. So I was at Silicon Valley company called Silicon Graphics, where I worked for about eight years within their service organization. So I really kind of learned directly what it meant to build services when you're also providing technology and the good and the bad and the ugly of that. And that inspired me to write a book. And since then, I've written seven books on technology business models and doing research on it ever since. Yeah. And a lot of tech companies are both builders and service providers these days. I mean, as a matter of fact, I'm not sure if I can think of one that is not.

5:05Is that an accurate perception? Yeah, I mean, the big transformation that has occurred with enterprise technology companies over the last decade or so, and you can pick on a Cisco, you can pick on a Microsoft. These companies, their preferred model is to create great technology, whether it's a piece of hardware, a piece of software, launch it into the world, have partners worried about implementing it and helping customers optimize it, etc. They want to focus on creating the core technology. What's happened over the years is, especially when you're dealing with complex enterprises, is that these customers want to make sure they're really adopting and getting the business value from the technology.

5:49So the services component has become important in terms of really making sure that customers can adopt, get the value realization. So all those companies have support services. They have education services. They have some level of consulting. They have work that they do to enable partners. So it's definitely more of a blended model than it has ever been. Yeah. And I would guess that AI and generative AI in particular is sort of like a bull in the China shop of TSIA. Literally every tech company, every technology services company is trying to rewrite their product from a JPI perspective. respect it.

6:30Well, and so there's two, there's sort of two lenses on this when you think about AI within the tech industry. So one lens is if, again, if I'm a Microsoft, if I'm a ServiceNow, if I'm a Salesforce, they all want to basically have AI capabilities that are implementing into their products and promoting out into the world, right, and placing their bets there. And so that's one thing that everybody's, you know, working to do and scrambling to do. We're actually focused on the other lens, which is how is AI changing the way they are operating internally? Everything from how they develop the products to, okay, now we're going to, you know, service the products, educate customers, all those workflows.

7:13How are they being impacted right now by the potential of AI? Because, you know, we think they're, we don't think we know there's a massive there there already. And you use the term bull in the China shop. It's exactly what's going on. Everyone's like, whoa, this is really, you know, a game changer. And so that's what we are really laser focused on is helping people understand where the most compelling use cases are, separate the hype from the reality, all that kind of fun stuff is this really becomes something that is going to just be ubiquitous within operating models. And it's going to get even more disruptive or confusing.

7:54People I've been talking to about agents, you know, AI agents, AI models that can take actions even in the physical world. I would guess that all tech companies are crawling through their operations to see where that can be applied. So tech, your members are focused on getting their technology into the hands of enterprise and having enterprise adopt their technology. But at the same time, there's this foundational technology that's appearing and your members have to figure out how to integrate that into their own operations. And how do you go about advising on that or how are people thinking about that?

8:46I mean, is it? Yeah. So this is the way that we've approached this for our member companies. So, you know, I'm sure you're familiar with the AI framework. It talks about AI capabilities that are below the waterline. You know, they've become very common at the waterline versus they're way up on the mountain still. Right. And so I think one of the first things that people are struggling with leadership teams is, you know, understanding that landscape. What is really mature? What are the use cases that are right here versus stuff that really is futures? And so we track that. We continue to track that.

9:24We did our first cut last year. And we think about AI capabilities below the waterline all the way up to way above across seven different areas. areas like customer success, areas like support, are people using support agents, areas like education services, etc. So we take snapshots for the members. And it was amazing. The first snapshot we did last year, we identified, the research team identified over 70 use cases that were already out there, right, across these areas. And then what we've been doing is just clicking into those use cases in more detail to understand, you know, what's working and what's not.

10:03And so I'll just give you some very practical real world examples, right? We've been identifying over the past couple of months. So this concept of co-pilots, right? So Microsoft has a co-pilot. Well, the real there there for enterprise cases is where people, you know, enterprises make that unique to them. So Nokia has a co-pilot targeted at telecom engineers. So it's very specific to their language, helps with support. Dell has the same thing for support people. OpenTex is a member that's done just incredible work on leveraging AI to generate educational service content. And one of the biggest issues these tech companies have is, you know, technology can be complex.

10:48Members don't adopt all the features, right? There's too much. And so if you can do a much better job of building education materials that are more persona driven, more customized, local language, all that fun stuff. So that's another example. So we are just finding use case after use case that there is real ROI there, real impact already. And so that's, you know, what we're super focused on. Yeah. And you mentioned Copilot. But is there, I would guess that there's no bias within TSIA toward one company's tech or another, but there is Code Whisperer and some others. How do you, do you make an effort to talk generically about these things or do some tools become so dominant that you're talking?

11:51Yeah, yeah. And we are, we are, because we have pretty much all the, the tool providers on the platform, we're agnostic, right? What we do for members when we're, we're doing these case studies is we, is, we will ask them, you know, what tools, AI tools are you leveraging? Is it off the shelf? Are you taking something open source and modifying? And then we will play that back to folks so they can start to see the pattern recognition. And it's still, you know, it's interesting. It's still, you know, early days, a lot of the tools are, are, you know, immature. I think one of the interesting things we're finding for, you know, enterprise companies, for, you know, tech professionals, they're, they're trying to figure out how do I apply AI.

12:32And one of the muscles they have to build is how do you evaluate this plethora of tools that have come out of the woodwork? And what, you know, you know, where, and just talking to different folks, that is, again, a skill that you have to, to mature. so you can kind of poke through and say, okay, that tool looks really cool, but is not going to scale for us, not going to be practical. This tool over here, much more practical. So that's still early days, but it is pretty amazing how rapidly it's maturing. And I'll just pick on this area of education services. This case study, I was talking to the woman that runs education services at OpenText.

13:06And then they started this journey of how they were going to leverage AI for content development. It was probably about, I don't know, a year and a half, two years ago, she said they started. And they were literally like just beta testing a tool with their Canadian base, with a Canadian based company, and really almost like co-developing the capabilities of that tool with them. You know, go to today. She said, you know, what we were, you know, doing with them co-developing two years ago, that is now just all off the shelf capability that another education service, you know, organization can just onboard immediately and take off running.

13:40So it is maturing rapidly for sure. Yeah. And that's kind of a cautionary tale. I mean, I talk to not necessarily tech enterprises, but enterprises about developing, about adopting AI. And, you know, it's C-suite executives are kind of deer in the headlights because there's, in any category, there are a dozen offerings and you don't want to invest in training up people and buying the tech and everything. and find out that you bet on the losing horse. So a lot of people are waiting. Is the same thing happening in the tech services industry? Or is it? I think that, well, I think there's a couple different flavors of this, right?

14:26So if you're a company that has your own AI capability that you're already investing in, Heavenly, if you pick a Microsoft, you pick a ServiceNow, then those organizations are aggressively applying that for internal use cases, right? They don't have to blink. They know that they're committed to that. and they're, you know, getting the benefits that they can. So that, I think that's one use case. I think the other use cases are some of the tools that are already proving to be mature. They're out there. And so I think, you know, people can jump on those. And there's a third use case, which is, you're right, that's still maybe that particular, you know, area of AI is still maturing.

15:02My cautionary tale, though, to these executive teams, because I agree with what you're saying. I think when you go in and you speak to, you know, the more senior, the bigger this gap is, I mean, these people have really zero experience with AI because it is new, right? So think about being an executive your whole life. You know, you're building a business. You know what works. And this whole new thing comes along. And you're trying to figure out, okay, what does that really mean to my company? And I have no experience. And I'm looking left and right. And none of us have any experience. So the common reaction I'm seeing is this sort of Mignogna strategy, right?

15:40this is let me just take a deep breath you know let's wait until everybody else figures out the tools and then we'll kind of jump on it and i i understand and appreciate the caution but my greater concern is i don't think that these executive teams are internalizing how massively disruptive this is going to be to their operating models and my my concern is if they're not starting to lean in now and saying, Hey, you know, we got to start understanding some of the use cases, we've got to start piloting some of the more proven use cases, we've got to start getting basically experience with what it means to use AI and change.

16:19If we're not doing any of that right now, every month, every quarter that goes by, there's a bigger gap between your operating costs, and somebody who's who's figured this out. And again, I think it's going to move fast. And so, you know, pick a horizon, two, three years, you could wake up and say, my cost structure is 20, 30, 40 % higher than my competitors because I had this wait and see mentality. And I think there's a real risk there. Yeah. Although there's also kind of an art to timing. I mean, that example you gave of this, of the education company spending - A year and a half, two years, yeah, working on it.

16:58Yeah. There's a big cost associated with that. And if by the time you're through those two years you've committed to this technology, suddenly there are all these other off the shelf solutions. Other, your competitors can pick up and be right where you are without having invested that time and money. Well, and I think, but if you think about, so, so, but let's think about the components of implementing an AI solution, right? So one of them is clearly just, what am I going to spend on a core piece of software or technology? The other components are understanding, you know, how my professionals, how my, you know, my employees are going to work differently.

17:36And there's a curve on that. Right? So let's keep just stay on this education thread. So one of the big advantages of AI in education is around content development, and how you can use AI. So if you're, let's say you have a team of content developers who are professionals have been doing this for a while, they're very good at it. But they have a certain way of developing content that they've been doing for years and years and years. And you don't snap your fingers and say, well, here's this new tool, and voila, you've changed your workflow and you're proficient on... There's a curve there, right?

18:11So again, my nervousness is if you're not building any experience with what it means to integrate AI into your workflows and you're waiting, waiting, waiting to place the safest bet possible on the technology, there's still a gap there, which I think is a little concerning. So, yeah, you would assume the technology services companies are quicker on the uptake than than not. I spent a lot of time briefing these executive teams on the state of technology business models. Right. What's driving profitability or where we see headwinds? What are and in every one of these briefings for the past 12 months, I put this topic of AI on the table.

18:54Right. And we do a lot of survey work in this. So you ask questions like, you know, do you have a senior executive assigned to AI, your AI strategy? Do you have a clear budget? Do you have processes to share best practices of implementing AI across departments? So you can test on these type of practices, right? And, you know, tech, I think, has the same challenges that almost any industry does on this right now, right? Again, it's such a new thing. They're not sure how to organize around it. They have data problems, like everybody has internally, just because they're tech doesn't mean that their data is clean.

19:31And so, you know, there's sort of this paradox when you read something in the business press. If you look at the industries that are always shown to be most aggressive with AI, it's technology companies and financial services comes up pretty high there. And then, you know, you go from there. Right. So they're probably doing a better job than a lot of industries. But I think there's this misnomer that just because they're, you know, you're a tech company, that you're just, you're totally on this thing and you got it. I am not seeing that. I think that there's a lot of executive teams that are still just as flat-footed as any retail executive or, you know, other industry.

20:08Yeah. And what percentage of just off the cuff, not a hard data point, but what percentage would you say of your membership are, if not AI native, have been in the AI space for 10 years or so? Yeah, that's a fantastic question. Here's the lens we use on that. So we think of a spectrum of tech companies right now that goes from being what we call AI advantaged. And so, you know, what are the attributes there? They probably have AI products they're monetizing, they're using AI internally. So they're pretty savvy, right? And the opposite end of that spectrum is what we call severe AI laggers. And they have no AI offerings.

20:58They're not using AI internally. They're not well positioned to leverage it because they don't have good data, et cetera, et cetera. So here's this spectrum. And right now, it's sort of almost like this classic bell curve of life. You have about 10 % to 14 % of the tech companies we look at that really, I would say, are AI advantaged. But on the opposite end, you have 10, you know, 15 % that are severely disadvantaged. And then you have, you know, the folks, the folks in the middle. So it's all, it is almost like this classic bell curve of, of, of life. But again, I think my observation looking at these companies operate, these AI advantage companies, you know, to me, this is like the internet on steroids.

21:38You know, like when the internet came out and companies that were able to jump on that fast created some real advantage, right? this AI thing is going to move faster than that. And the advantages that get created are even going to be more massive. So if you're down here and you, again, are severely disadvantaged, not well positioned to really even leverage AI in your operating model, that is going to be a problem. You were talking about the different, looking at the different places in the operations where AI could be applied to immediate ROI. or do you have a distribution there that you should start with?

22:16I'll give you two responses to that and what we see in the data. So one response is if we just think about different activities within a tech company, where are the folks that are leaning in, where we see the use cases that are mature and you could definitely should be leveraging right now. The classic is support services. Everything from, you're talking about agents, but just helping with self-support all the way to more predictive support. I mean, they're, you know, basically they're using AI to prevent things from even happening, right? Outages, et cetera. So that is becoming very mature. I would say that, you know, field services, you know, using AI so you don't have to deploy hardware and equipment on site, that's a big use case.

22:56Education services, as I mentioned, for content development, that's a big use case. And then you start to see, you know, a fall off. So, you know, if you look at the area of customer success, which is very popular, how to help customers adopt. Customer success organizations are kind of lagging on, you know, applying AI. There's good use cases, they're just lagging. And then when you come to the revenue generation side, so if you think about any technology company, you have to sell and market your products. And there are absolutely use cases for AI to help you do that more effectively. And what we're seeing is sales and marketing organizations are severe laggards right now in using AI.

23:37Yeah. And again, there are some that are, you know, but I'm just saying in general, right? You say, hey, how are you using AI to do, you know, renewal management or your forecasting or to better understand, you know, opportunities. Still, sales organizations are typically not technology, you know, technology forward, you know, organizations. But I will tell you the other view on this, like where are the use cases? I think the companies that get an ROI, proven ROI, you focus on workflows actually that you understand really well, that you have really solid performance metrics on. Again, content development, an education service organization can tell you, hey, to generate an hour of training materials, it takes make up a number, 10 hours of labor.

24:21If I apply AI to that, I can see exactly where I'm reducing. Same with how much time does a support engineer spend on a call? How many times do I deploy equipment replacement hardware that I didn't need to? Those are all clear performance KPIs where you can put AI to it and you can see the benefits. That's a winning attribute. As opposed to, and it's interesting, there's just an article in the Wall Street Journal. It was about Microsoft's co-pilot. And they were saying, well, we're paying whatever,$30 a month. And we're not sure if there's benefit and we've got it. But I read that and I'm like, well, but what's the use case?

24:57I mean, just giving employees co-pilot and throwing it out there and paying for it. And they go, well, I don't know if there's an ROI. Because you haven't defined what you expect to get out of that. So, you know, that's where you really do see ROI is when people are crystal clear on what they're trying to optimize. Yeah. You mentioned a few times now the education content generation. Are you talking about companies that are creating training programs for other companies? Or are you talking about their own? for their, yeah, their internal, if you, so if you think about this challenge, so there's something we wrote about years ago and we called it the consumption gap and we, and it was a book called complexity avalanche.

25:42And our argument was that enterprise tech companies just keep throwing out all of this feature functionality, you know, wave after wave and the gap between what customers can, you know, enterprise customers can actually consume and apply, you know, it just gets bigger and bigger. Right. So that's always a challenge for, you know, the education, internal education folks is getting the materials, what's effective education. Again, is it, is it targeted toward different user types and personas? And so using AI to create personalized content at scale is amazing. And I know you had the, you know, the Khan Academy founder on talking about, you know, it's that thought, right, is how does AI really change and make it easier for people to consume?

26:26And, you know, I'll give you just, you know, simple examples of the biggest challenge with these education service departments is keeping their content up to date with new releases. So if it takes me all these man hours, you know, to do this every time, I'm always falling behind. If I can use AI to compress that and get fresh releases out there in a super timely manner, it's a game changer for them. And so it's just, you know, just one example of, I think, you know, it speaks to not only saving real costs, but creating a better customer experience, you know, as well. So it's a twofer. Yeah. And do you guys as an association provide training, any kind of training?

27:08I mean, it seems like that would be a use case where, you know, you say, look, you have this tool co-pilot, you have a, depending on your, but these are tech companies, you have all these developers. You know, how do you get it applied? Well, so our use case on this, and it's a great question. So if you think about a company that creates research, right? So we go out, we study things, and what you typically do, you create papers on that, right? Research artifacts that people can consume. And then we will also, we will workshop around that. So if somebody goes, well, hey, I read the paper, but I want my team to really get this framework or really understand these lessons learned.