The AI Visibility Gap: Why Enterprises Still Cannot Measure What They Are Using

[00:00:00] What happens when the fastest moving technology wave in decades crashes into enterprise reporting governance and leadership processes that were never built for this kind of speed? Well, on today's episode of the Business Tech Perspectives podcast, I'm joined by the CEO of a company called Laredin.

[00:00:24] And he's someone who's had a front-road seat to several major technology shifts over the last three decades, from the early days of the web to mobile, cloud and now AI. My guests have seen what change looks like when it first arrives, when people underestimate it and when it suddenly becomes impossible to ignore. But what makes this moment different is the pace.

[00:00:50] AI is not arriving in one function, one geography or one neat little pilot. It's showing up everywhere, all at once, pulled into businesses by employees who are all trying to work faster, think better and frankly survive the modern workday with a little bit more help. But this creates opportunities. It also creates a problem, though, that a lot of leaders are still wrestling with.

[00:01:16] Because if AI is still being adopted faster than any previous technology wave, how do you actually measure what is happening? What is working? What is being wasted? And what is slipping under the radar? As the old belts and braces approach to IT, that mantra of you can only improve what you measure, I think that's never been more relevant than it is now, right? And this is where today's conversation gets especially interesting.

[00:01:45] Because my guest will talk about what he calls the AI visibility gap, the growing disconnection between what leaders think is happening across their organisation and what is actually taking place on the ground. And once you start hearing some of the numbers, I think it's impossible to ignore because it's no longer a story about just a few teams testing a chatbot or a note taking app.

[00:02:09] This is about real AI spend, real productivity shifts, governance questions and a growing need for businesses to understand their AI estate in real time. So today we're going to cover measurement, accountability, productivity and what it really takes to turn AI excitement into something useful across an organisation.

[00:02:32] So if you've ever wondered whether your business has a real handle on its AI usage or whether leaders are being asked questions they cannot yet answer, this one is well worth your time. But enough for me. Let me introduce you to my guest right now. Thank you for joining me on the podcast today. Can you tell everyone listening a little about who you are and what you do? My name is Russ Frieden. I am one of the founders of a company called Laridin.

[00:03:01] I've been in the Bay Area in the US. I've been in the Bay Area for the last 30 years doing start-ups really since 1996. And so, you know, I had a whole run of start-ups. A couple went public, a couple got sold. You know, been around the tech ecosystem for a long time. And my company Laridin that I started with my very long time partner of 25 plus years, Jim. And then we have another partner who joined us, Amaya.

[00:03:26] We started that company about 15 months ago and raised 17 million dollars from Andreessen Horowitz and from Gradient, from Bloomberg, from Homebrew and Hayes, a bunch of great folks. And so I've been kind of working on start-ups and trying to do something amazing for the last 30 years. Wow. And in 30 years, if you look back there, I mean, you've been through multiple waves of technological change from advertising to mobile and cloud and obviously now AI.

[00:03:55] I mean, I'm curious if I was to ask you, what feels different about this current moment, especially in how fast AI and now agentic AI is entering the enterprise? Does anything feel different this time around? What I would say is, you know, I was lucky enough back in, you know, the first startup I ever joined was before Netscape went public. Like this was, you know, I was the first guy at the first online ad network in the mid-90s. I was much younger then, although I did not have hair then either.

[00:04:17] And the web advertising clearly was exciting and the dot-com boom in the late 90s was clearly exciting. The issue there was it was really just about media and e-commerce and it was mostly in America. I mean, yes, obviously, you know, you know, the UK back then was, you know, 18 months behind continental Europe was another 18 months behind Asia was another, you know, two years behind that in the late 90s.

[00:04:44] Now you have something that is touching every aspect of the economy in every region of the world all at once. You know, chat GPT is wildly popular in India. Claude's wildly popular in France. Like these are not this isn't oh, it's just in the US just in media. This is changing the way work is done for a very not all jobs, obviously, but for a very large portion of jobs all around the world. And we haven't even gotten to robotics yet.

[00:05:11] So this is what's happened now is just what happened in the 90s. But but, you know, two orders times a thousand. I mean, it's not it's similar in terms of excitement, but in terms of global nature, you know, in 1996, 1997, you did not find people, you know, working on startups in Sweden. Right. You don't find people working on startups in Berlin. I mean, of course you did, but not not in the way that today.

[00:05:38] Yes. You know, the Bay Area is the center of the kind of at least Western world AI ecosystem as the Bay Area has been kind of the center for a long time. But, you know, there's amazing things happening all around the world now. And that's just because, you know, the Internet's been around a long time. Mobile's been around a long time and AI is impacting everything. Yeah, it really has. And despite all this success and its ubiquity, you've also spoken about what you call the a visibility gap.

[00:06:07] So for a business leader that's listening to us today, how would you describe that gap in real practical terms and on why it's become such a pressing issue? Sure. So, I'd say it this way. Right. One of the things that was true in prior sets of technology is they would all actually, even though it felt very quick when you were an entrepreneur, they would actually roll out quite slowly. Right. British telecom was actually quite slow to adopt all of these new tools internally.

[00:06:33] And, you know, as were, you know, as was Deutsch Telekom, as was T-Mobile, as was AT&T, right, as was all of these companies. Right. And so one of the things that has happened with AI is because there is such tremendous consumer pull. All of these tools and by the way, because they're so obviously economically valuable, these tools have just been pulled into the enterprise very quickly.

[00:06:56] So the traditional world where, honestly, for larger companies, IT moved relatively slowly and relatively relatively deliberately, they've just been pulled forward. And as that is pulled forward, because everyone knows it's going to be useful, all of the normal scaffolding and infrastructure around security, around measurement, around governance, these are normal things that very large companies, you know, tend to take years to put in place as they're adapting to the cloud or as they're adapting to mobile.

[00:07:27] That's just all being done in real time. Right. We're effectively kind of building the scaffolding while we're putting up the building. And so if you think about that analogy, sure, all of these gaps are really just obvious things there will be in the future. So today, for instance, one of the things we find with a lot of enterprises while they've deployed AI technology, if you ask them about what tools are being used internally, they probably only know about 30 percent of the tools their employees are using.

[00:07:55] And that's not to say their employees are doing anything wrong. It's because these tools are exciting. Most companies are taking this let a thousand flowers bloom approach, which is great. But what that means is you think you have 70 AI tools being used and you have 200. Now, other people would come on your podcast and talk to you about the security risks there. Obviously, those exist. We're not a security company. I don't spend much time talking about that, but obviously those are real.

[00:08:20] Well, there's governance issues, right? Who's allowed to use what, where there's data governance issues, where there's privacy issues. But then for us, we worry just a lot about measurement and visibility, which is if you believe you're going to get a lot of value out of these tools and you are going to let a thousand flowers bloom. We think that's amazing. But we also think you should measure what's happening.

[00:08:43] Right. We think if you're going to have 200 tools in your org and you know about 70 of them, it's not that there's the other 130 should be shut down, but they should certainly be known about. They should certainly be measured. You should certainly know what's happening. So really, we talk a lot about I think a lot of this is if you just think about this fact that, heck, these tools you and I spend a lot of time with every day.

[00:09:06] They didn't exist two or three years ago. And now we have this world where enterprises are adopting them at a faster rate than they've ever adopted any new technology ever. Then obviously, we don't yet have the tools to measure ROI, to measure performance, all the security, all the data, all the governance. It's all being built very quickly. So it's a very exciting time. But that's really what we're talking about. It's not that it's not that people don't know this, by the way. When we go talk to CIOs of large companies, they know they're not tracking this. They're looking for solutions here.

[00:09:33] And I think many organizations are already using dozens, if not hundreds of AI tools across teams. And your study recently showed that 45% of AI adoption actually happens outside of IT's view. Yeah, that's exactly my point. And by the way, if you talk to CIOs about this, obviously, they could lock this stuff down. But they don't want to. These tools are great.

[00:10:00] Your employees aren't spies from some sinister country trying to most of the time. I mean, that's cool in the movies. Your employees are actually working hard and they want to be successful. They want to be productive. They're using these new tools that exist. So I think it's great that people are allowing it. My perspective is, but if you're going to allow it, you should measure it. Because not every tool is useful. We had an early customer of ours where their employees had adopted AI note-taking apps like you're using right now.

[00:10:30] And those are great and those are powerful, except it turned out in their company of 2,000 employees, they had five separate AI note-taking apps. Now, that's dumb. Obviously, you should pick one. I'm not going to say which one you should pick, but obviously, you should pick one. First of all, there's just cost savings, let alone the fact that obviously you'd like all the notes to be in one spot. You don't want to split across five different vendors. And so that's an amusing version of this is obviously scarier versions with data, residency, and privacy, and all of that.

[00:10:59] So my perspective is you absolutely need visibility. If you're going to let there be a 500 tools or 5,000 tools, it'll take years before this space settles down. So run all the experiments you're going to run. But it's only an experiment if you know you're doing it and you measure it. Yeah. And another stat from your report there was 62% of senior leaders report they lack a comprehensive inventory on those AI tools. Again, no big surprise. And by the way, 62% said it.

[00:11:28] The other 38% just hadn't thought about it. Yeah. And like I said, I'm sure out there in the world is another entrepreneur like myself who's selling a security solution. And he's going to talk about or she's going to talk about how scary that is. It might be. I generally am the view that CIOs and CFOs are very smart people. They want to protect their companies and they want to enable work to get done efficiently.

[00:11:55] And so making the decision to allow lots of experimentation is a decision. And it might be a very smart decision. My point is just you should be measuring the experiments. And bear in mind some of the big stats we've both mentioned today. How realistic is it for a CIO to have that ability to produce a real-time accurate inventory of every AI? I mean, this is what our tool does, right? So, I mean, all of our customers have this, right?

[00:12:21] I mean, among the many other things our tool does, you know, and I'm not, I don't want to turn this into a sales pitch for you. But look, one of the reasons we talk to CIOs about this all day long is like, this is not an unsolvable problem. We have solved it, right? It's one of the reasons we've attracted a lot of capital and have a lot of customers. So here we are in the present day. You've built Lared in to bridge that gap between AI spending and measurement, what we're talking about here.

[00:12:44] But what does that better visibility actually unlock for organizations when it comes to ROI, workforce productivity, smarter decision-making, all that stuff? I really think about it this way. What's the – so first of all, a lot of our customers just want reporting. Like what's actually happening, right? We have all of these tools. Who's using the tools? Where are they using them? Are they using them well? Are they using them proficiently?

[00:13:09] How are we doing in terms of driving adoption and driving intelligent adoption and driving productivity? So for a decent set of our customers, that's it. They're happy to have that info. They didn't have it. Now they use us. Now they have it. That's fine. The next set of things people care a lot about that they're also going to get from us is – I think about it this way. What's the economic term? Emergent behavior. So you have 5,000 people at your company. If you roll out, I don't know, Claude – I mean, you roll out a bunch of AI tools.

[00:13:39] You roll out ChatGPT. You roll out whatever to 5,000 people. 50 of them to 150 of them, 300, whatever. Pick your number. Are going to use them to become superwomen, supermen, right? The most productive they've ever been in their career. And I'm sure you see this, Neil. Your friends that are very high agency, very productive are working harder now than they've ever worked because the amount of work you can get done, frankly, it's intoxicating the amount you can get done. Yeah.

[00:14:06] And so some subset of your employee base, under 100%, are using these tools. They know exactly what to do. But a lot of people, when they stare there looking at the, you know, Gemini prompt or the ChatGPT prompt or whatever, they basically look at it to see when's the next time Fulham is playing a Premier League match. And, you know, they rewrite an email or two and then they move on with their day. And that's okay. Yeah. Because the truth is we don't have good training materials.

[00:14:35] This stuff has only existed for a couple of years. So the way I think about our tool is when you have this workforce intelligence around what's happening with AI adoption, I talk about our tool as let's help you kind of, you have these, you have this emergent behavior in your company or another way to say it, you have these trailblazers. And so some subset of your employees are very high agency, have taken the bull by the horns, whatever analogy you want to use, and they've blazed a trail of how to be hyperproductive.

[00:15:02] If I'm a leader in HR or if I'm a manager, what I want is a tool to help me draw a circle around those people, not at the employee level, not Neil, not Russ, but what is the behavior of the hyperproductive people? So I can give it to leaders so that if I'm a leader, I can sit down with you, Neil, and just say, hey, Neil, just do these three things. You don't have to be a trailblazer. Here's a map. The 5% of your employees, they blaze the trail. Great. Great. Here's a map. Great.

[00:15:31] This is so that we can iteratively figure out together and train what people. So in terms of ROI, for a lot of our customers, it's how do I drive that kind of large percent of my workforce? Some percent of my workforce has defined what great looks like. Great. How do I uncover what great looks like so I can drive organizational change? Because if you've ever worked in a company, I don't know your background, Neil.

[00:15:56] If you've ever worked in a company with 5,000, 50,000, 100,000 employees, change management is a huge deal. Yeah. And so how does HR, how do leaders sit down with folks who are a little worried about AI? The word is going to take their job. They're not really sure what to do with it. They're feeling a little apprehensive that they're not hyperproductive. And just say, hey, look, Neil, when we look across the sales org, this is what's making the people hyperproductive. Do these three things. You'll be more productive. One, two, three. Do you want training? Do you want help?

[00:16:26] Here's videos. Just do this. You don't don't sit there and look at a blank chat GPT box. Not sure what to do. Yeah. Do these three things and you'll be better. And the constant iteration that. So that's a big deal. And that's a big use case of our tool for ROI is great. I've rolled out these tools. Let's figure out what's happening. Let's have an inventory. What's happening. Let's figure out what's driving productivity and then let's quadruple down on those productive behaviors to drive meaningful kind of workforce change across the company.

[00:16:54] And on top of shadow AI and all the different AI tools out there, there's also that growing population of non-human identities or AI agents. Do you guys track things like that or have any interest in tracking? So we certainly we do some today. We will over time. It's not a particular huge area today. I the way I think about it this way is. The future of the workforce.

[00:17:19] Is going to be some amount of work will be done by humans using AI tools. And some amount of work will be done by AI agents without human interaction. Yeah. I don't know what the workforce of the future will look like, but it won't be 100% humans, 0% agents. It won't be 100% agents, 0% humans. Right. So this isn't I'm not talking about jobs or anything like that. I'm just fundamentally in terms of work to be done. Some will be done by each group.

[00:17:48] The fact is today, as much as there's a lot of excitement around agents, almost all of the AI work today, all of that spend, all of those revenue numbers you read about is around AI tools used by humans at companies.

[00:18:36] So that's where we're really focused today. Call centers, agents are mostly coming in the future versus reality today. That's not to say they're overhyped. Like I'm not making a I'm not saying entrepreneurs lie. I'm saying anything like that. What I'm saying is obviously the first area for AI tools is how do I help humans become more productive? At some point, there will be tasks that can just be done totally by agents.

[00:19:01] And when those are big chunks of the business tasks, when you actually have agent sellers making calls and driving pipeline, obviously, you'll have to measure their productivity alongside the humans that are doing that for you. Right. You already see that today in call centers. If you look at the AI agent work in the call center, right, those tools are all about, relatively speaking, how productive are we with the entry level customer service calls when, you know, the agent vendor picks up versus when a human picks up. Of course.

[00:19:33] I'm curious, based on everything that you're seeing in the market right now, are there any practical steps that business and tech leaders could take to move from those fragmented AI experimentations, AI pilot purgatives, to a more measurable, accountable, and scalable approach? I actually think pilots are amazing. I think the purgatory issue is the issue. The thing that pilots are fantastic.

[00:19:56] Frankly, if I were in a large enterprise, I think the idea of making a long-term multi-year commitment today is kind of silly. You know, 18 months ago, it seemed, two years ago, if you were in the UK, it seemed that Microsoft co-pilot was everywhere. A year ago, it seemed like people were less excited about co-pilot and much more excited about chat GPT. And three months ago, they're much more excited about Claude. That doesn't mean Claude is one. I have no doubt.

[00:20:26] Oh, and by the way, I'm sorry. In between that, there was the Gemini moment where everybody was very, very excited about Gemini. So frankly, I don't think the way, I don't think pilot purgatory is the right way to think about it. I think the way to think about this honestly is how do you set up your measurement, governance, and data layers to support constant experimentation with these tools? I don't know why any large company or any mid-sized company would pick one vendor and that would be their vendor for years to come.

[00:20:55] Now, traditionally, that's how the world worked, right? Traditionally, you decided on Microsoft or ServiceNow or, you know, traditionally, but that's because the pace of change of those tools was not that great. Like, it is the case that last October, the best AI coding tools were, you know, cursor. But since December, it's pretty clearly been Claude if you talk to developer friends. Now, will that be true in April? Probably not. Probably Codex will be now.

[00:21:22] So my point is, I think it would be silly as an enterprise to say we have decided this is our platform for the next three years. What I would say, and when we talk about people, is get your security in place, get your governance in place, get your data strategy in place, and get your measurement in place. And then experiment like crazy, right? Of course you should try out 100 tools, right? This is the most exciting kind of time of new product creation ever. Of course you should be able to try out a bunch of tools.

[00:21:51] Now, there will be some areas that lend themselves to, hey, you just pick one vendor and you stick with it and you have economies of scale over time. Of course there'll be areas of the org where it's like that. But frankly, I think for most areas of the org, if you're in a marketing department and your goal is to create tons of great content and tons of great creative, if the best tool today is Canva, which let's say it is, why would you pick that for multiple years? Maybe a year from today, Adobe has leapfrogged them, right?

[00:22:18] Your goal, yes, it's more annoying for CIOs, of course. So I think, you know, and when I talk to CIOs who are customers, they're all thinking like this. They're all thinking about how can I be nimble? What is my data strategy? What's my measurement? But by the way, I was talking to CIO of a very large public company and we were doing it on a podcast, but I don't want to say the wrong company name.

[00:22:43] So I won't say the name, but it's a big public company and it's, you know, maybe I'll tweet it out later for if your listeners want to find it. And he said, we're not signing any vendor for more than a one year deal. The stuff's changing too quickly, but he had his data in place, his security in place, his governance in place, and his measurement in place. So great, let's try a lot of things. Absolutely love that approach. And for any business leader that's listening, maybe the thing that keeps them awake at night is that fear of AI sprawl.

[00:23:13] And maybe the dream is to get to that verified AI inventory that satisfies their board. I know this is your wheelhouse. This is what you do. What advice would you give to those people? And Phil, you can mention that. Like I said, my flat out perspective is as long as you have a framework to measure these things and to govern these things, you should not be afraid of having 20 tools, 50 tools or 100 tools. Now the spend is going to matter at some point. Right. So what I would much more worry about, like I said, is am I set on the security side? Am I set on a data ownership side?

[00:23:42] And am I set on a measurement side so that my CFO has the tools to understand if that $20 million of spend is worth it or not? Right. Because these tools are quite expensive. Right. We've gotten to the point with our developers at Larrantin where, you know, some of our developers, their monthly token bill is more than their salary. Now, that is OK as long as they're productive. I'm not I'm not annoyed with that.

[00:24:04] But if you're spending tens of thousands of dollars a month in tokens as an engineer, you sure as heck better make sure that the CFO and the head of engineering have tools to understand and track that spend and make sure it's driving productivity. So as long as you have kind of measurement and governance in place, some of that's from Larrantin, but some of it isn't right. As long as you have that in place, I think embracing the fact that it's an exciting time is great.

[00:24:31] And so I just can't imagine why you'd tie yourself to any one decision right now, because the platforms are just changing so quickly. They really are. And for anybody listening, maybe they want to check out your Twitter to see if you did name the public company that you mentioned or alluded to. Yeah, I will find it. Yeah. So I'm Russ Fraden. It's not hard to find me. The company is Larrantin. My Twitter is probably R Fraden. LinkedIn. I don't know. It's not hard to find me.

[00:24:58] But I will find the clip of the CIO, the discussion we were having, and I'll tweet it out at some point in the next 24 hours. Beautiful. Well, I'll add a link to that, the website, and indeed the report that we've referenced today. But before I let you go, anything else that excites you about the future, what you're working on, any teasers you can leave us with? You know, we're spending a lot of time going deeper at the department level. So I talked a little bit about developer productivity and going much deeper there, not just tools for CIOs and CFOs, which is our whole product, but tools for heads of engineering.

[00:25:27] You will see us do the same thing for heads of sales and for heads of marketing, for CFOs, right? So we're obviously going to work our way across the enterprise. We just started with very deep tools around developer productivity. And so, you know, hopefully I'll come back on and tell you about that in the future. Oh, you've left us with a tease. I love it. Well, I will get all the links added to everything that you mentioned, and I look forward to our next conversation. I've got to think it's going to be a good one. But thanks for joining me today and sharing your story. Thank you very much. Having listened to my guest there, where does that leave us? Where does it leave you?

[00:25:57] For me, one of the strongest takeaways from that conversation is that AI experimentation is not a problem. In many ways, it is exactly what businesses should be doing right now. The issue begins when experimentation happens without visibility, without measurement, and without any real understanding of which tools are helping, which are overlapping, and which are quietly burning cash without moving the needle.

[00:26:25] As Russ said there, we don't need five different note-takers in meeting rooms. And I wonder how many of you listening have exactly that right now. So I think what we need is a different mindset, less obsession with locking in one long-term answer, and more focus on building the measurement, the governance, the data foundations. And then we've got all these things in place. Then you can test. You can learn. You can adapt. You can move with confidence.

[00:26:51] And I liked his point about productivity inside the workforce, because in every company, there will be a small group of people who seem to figure out all this stuff very early and suddenly start operating at a completely different level to everyone else. And the challenge is not for leadership to sit back and admire them like they're some kind of AI Olympians. I think it's actually to understand what they are doing, identify their behaviours that are delivering and driving results,

[00:27:19] and help the rest of the organisation get there too. That is where the value starts becoming very real. And there was also a much wider message in all of our conversation today, and I think that is AI visibility is simply not an IT issue. It's a leadership issue. It touches spend, performance, governance, workforce enablement, and board-level confidence.

[00:27:43] So if you cannot see what is happening, and if you cannot measure it, you're left making bets in the dark, ultimately. So hopefully today's conversation raised a few questions and lit up a few light bulb moments for you about AI sprawl, productivity, or whether your organisation can produce a real-time inventory of its AI usage. And if it did, I'd love to hear your thoughts.

[00:28:10] Are you or your business moving too fast with AI, or are you moving at the speed that the moment demands? And how are you measuring it? As always, techtalksnetwork.com. You'll find myself, 4,000 interviews, 9 podcasts, and a whole heap of ways of contacting me. And I want to hear from you. This is a dialogue, not a monologue. You've heard from me, you've heard from my guest. Now let me know your thoughts. But that is it for today.

[00:28:39] I will be back again very soon with another guest. So please, subscribe to the podcast, leave a rating and review. It really does help the podcast. And I'll speak to you again soon. Bye for now.