3511: BCG on Closing the Gap Between AI Experiments and Real Business Impact

[00:00:04] How do you lead a workforce through a moment when the speed of change outpaces the average worker's ability to keep up? This is one of the questions that I hear everywhere I go, from boardrooms to front lines to conference halls. And today's guest, he's someone that sits right at the heart of this conversation. His name's David Martin. He's a senior partner at BCG and the leader of their people and organisation practice,

[00:00:30] where he spends his days helping companies rethink skills, roles, culture and working patterns, all as AI is reshaping every corner of the workplace. And he also serves on BCG's global AI leadership team, which gives him a rare vantage point over the adoption divide that is winding between those who instinctively fold AI into their daily rhythm and those that just keep hitting friction at every step along the way.

[00:00:59] Sound familiar? Well, our conversation today will move through the patterns BCG is seeing across more than 10,000 survey respondents. He'll explain why AI fluency is increasing in some pockets while new joiners are stalling at the starting line, why middle managers are quietly becoming the heavy users and why frontline workers still lack that hands-on learning that they need to thrive.

[00:01:25] But we'll also explore together the reality behind those big headline claims about productivity jumps in engineering and customer service. And look at how the biggest gains only arrive when companies rewire the whole workflow rather than just drop a tool on top of an existing process. So, yeah, that means we're going to be talking about leadership behaviours, the psychology of learning under constant disruption,

[00:01:49] and the honest fears around shadow AI that many leaders are quietly trying to contain right now. So, the theme that we'll run through everything David will share today is this phase of AI change has very little to do with shiny new models and everything to do with people, trust, skills and confidence. So, how do you lead through all of that with clarity and care?

[00:02:15] Here at the Tech Talks Network, we now have nine podcasts and approaching 4,000 interviews, and that is only possible with some of the great friendships that I've developed over 10 years of podcasting. And a company I'm proud to call friends of the show is Denodo, because not only have they been on this podcast multiple times, they also help make sense of the AI data chaos that we're seeing now. Because the data world is louder than ever.

[00:02:41] AI hype, lake house complexity and pressure to deliver more with less. These are things that I talk about every day on this show. But Denodo is helping businesses make sense of it all, because they provide a unified data foundation for trustworthy AI, lake house optimisation and data products to finally bring service to life. So, whether you are a CIO or a builder, Denodo helps you activate your data with speed and governance.

[00:03:08] And their global partner network also helps you accelerate every step of the way. So, if you're ready to unlock real outcomes, simply visit denodo.com today. But now, it's time for today's interview. Let me introduce you to today's guest. So, a massive warm welcome to the show. Can you tell everyone listening a little about who you are and what you do? Yeah, sure. So, my name is David Martin. I'm a senior partner at BCG.

[00:03:36] I lead our people and organisation business unit. So, everything around talent and skills, organisational and operating model design and culture and change management. And then I serve with a cross-functional group of folks on our AI leadership team globally. Awesome. And of course, AI continues to be a huge topic. It's been, what, three years since ChatGPT first arrived. We have the early adopters, those that are caught in pilot phase not getting out.

[00:04:04] I've got to ask, what do you see as the main reason behind this growing divide between those AI fluent teams and those that are just struggling to keep up and struggling to get out of those pilot phases and into life? You know, it's funny because I was really impressed with how quickly and broadly companies started to attack this.

[00:04:24] I think what we see with folks who are getting value out of it is they're actually doing fewer things and attacking it bigger and more holistically, which was surprising to us. Companies that are still struggling are the ones who, you know, we're approaching it by what we call a thousand flowers blooming. And they took on, you know, hundreds of use cases across the organisation, but thinking about it very narrowly.

[00:04:50] And because of that, it's really hard to capture value if you save, you know, five minutes of someone's time a week. Translating that into dollar impact is really difficult. So they're doing fewer things bigger. And I appreciate this is a question that everybody wants the answer to right now, but how can an organisation move beyond simply providing AI access to actually enabling their employees to use it confidently and effectively?

[00:05:16] Because we're three years down the line now, so you think we'd start to see more happening. Yeah, for sure. Sure. It's funny because, and we especially see adoption lagging on the front lines. And part of it is because people have so much day-to-day work to do. You ask employees and they say, I just don't have time to learn the tools. And so one thing that organisations are starting to do well is full immersive training.

[00:05:42] I think what we've also seen is an hour workshop is just not sufficient to help someone understand how to use it effectively in their day-to-day life. So they're doing more immersive training. They're building awareness. They're providing access to good quality tools. And we're also seeing leadership behaviour being a big contributor. And so I think our recent data showed that supportive leaders have employees who are four times more likely to use the tools effectively.

[00:06:12] And so leadership behaviour, obviously, another big piece there. Wow. I never saw that one coming, I must admit, because I've heard so many stories of some bosses, some teams outright banning issues because of the fear of the data, etc., which is a valid point. But there's workarounds around that now. But training is a great point. I mean, why is AI training still such a gap for frontline workers? And what kind of hands-on learning is making the biggest difference?

[00:06:38] And so one of the other reasons I asked that, I was speaking to somebody recently, and they said inside the average organisation, it's the people at mid-tier at an organisation. They're the heavy users, not the junior users, not the C-suites. So what's going wrong with training here? Well, OK, so maybe a couple of things. And your stat, I think, is spot on. We saw 90% of middle managers using it at least once a week, I think it was, were frontline lagged significantly.

[00:07:05] I would say that for the frontline folks using it, they are much more frequent users and probably more effective users than the middle managers. And so we do look at volume and effectiveness of usage and adoption, not just, you know, weekly usage. At the frontline, I think one big piece is, look, they have, you know, their everyday work to do. And so the tools need to be relevant for the work they're doing.

[00:07:34] It's not just using ChatGPT to summarize emails or to write emails. It's actually, you know, if you're a software engineer, it's using Cloud Code or Cursor to write your code more effectively. And, you know, marketing, same thing on content creation. If you're an HR professional, you're using some of the embedded tools inside of your HRIS. And it's really difficult for employees to keep up with just how fast the ecosystem is changing.

[00:08:01] And we saw this in software engineering. We were working with a large tech company and their engineers were using it about once a day, but they had no idea the capability of the tools that had recently rolled out. And so their ability to use things like the model context protocol and some of those tools that that Claude has rolled out is just far lacking relative to what the capabilities are.

[00:08:27] So that immersive training helps, I think, really educate people on what the quality of the capabilities are and how impactful it could be for your day to day life. And middle managers, I would say it's slightly less so. It's more in that chat GPT world, which is great. And we do know that leaders and managers who are using tools have their frontline employees using them more effectively. But they're seeing a little bit less value at that layer of the organization than the frontline employees for using it well.

[00:08:58] And you mentioned everyone from tech teams, software engineering to marketing teams are largely unaware of some of the capabilities of this technology. So on that side of things, what role does leadership play in helping accelerate AI adoption across a company? And I'm curious as well, what behaviors separate those successful leaders from the more cautious and hesitant ones?

[00:09:20] Yeah. So I think leaders generally embrace the notion that they need to communicate to their employees that the employees need to be using the tools. But a good leader is actually using the tools themselves so they can understand, you know, the capabilities and the disruptive change. But I think above that is curiosity. I think there's a lot of apprehension for leaders to actually reinvent themselves.

[00:09:46] And what you'll see, I think, going forward at all layers of an organization is that curiosity and the always on learning mindset is critical. And leaders who exhibit that we do see are much more effective at helping their frontline drive it. I would say the other thing is communicating a vision for how they view work to be in the future.

[00:10:10] We've seen a lot of leaders get overwhelmed by just how fast the world is changing and the uncertainty. And employees can pick up on that. And leaders who are able to actually craft a vision of where they think their function is going, or if you're a CEO, where you think the organization is going, and communicating to employees on how they fit into that vision, you see much larger uptake of the tools.

[00:10:35] Whereas if you don't do that, you see increased amount of fear of job loss. And obviously there's just a lot of uncertainty that employees, you know, aren't very comfortable with if you don't have a good leader communicating that vision. And we started 2025 talking about the ROI, the big questions that organizations were asking as they were struggling with scattered AI experiments caught in pilot phase.

[00:11:02] Then it evolved or the conversation evolved into agentic AI and the work that agents were doing. We're recording this at the end of the year now and begin to look back and look ahead at what could be happening next year. But how have you seen organizations shift from those AI experiments at the beginning of the year to true workflow transformation that's driving that real measurable business impact? Are you seeing that this year? Yeah, you are.

[00:11:27] And I do think that customer service, marketing, software engineering, some in the back office functions like FP&A and some HR functions, you are actually starting to see not just experimentation and pilots, but really strong value coming from the investment there. And it is to your point, you have to start with thinking about the workflow end to end and identify not just places where you can use AI,

[00:11:54] but identify what that new workflow looks like and what the role of agents are in that workflow. And then consequently, what the role of the humans are. Sometimes that includes redefining roles. And that's why companies who you see have done it really well in 2025 have also redefined roles. And they have redefined competencies in that as well. But the biggest thing they've done is think about the workflow end to end.

[00:12:23] If you go back to software engineering as a good example, the tech company I mentioned earlier, when you started just by looking at the workflow, you saw the engineers were only spending 35% of their time with their hands on the keyboard. And the other 65% was going back and forth with the product manager on getting clarity on the requirements and working downstream with quality assurance and deployment. And so if you don't address some of those upstream and downstream parts of the workflow,

[00:12:52] then even if you're driving huge productivity increases on their time spent coding, you're still only at a fraction of what the opportunity is. So start with the workflow and then think about how Gen.AI fits into that, not the other way around. I love that. And as I said this year, AI agents has been talked about everywhere. Yeah, adoption has remained relatively low. I think companies try and get to grips with the capabilities there. But anything you're seeing that's holding companies back and how do you see their role evolving next year?

[00:13:24] Yeah, a couple of things. You mentioned it earlier. I do think data quality to some degree is holding some companies back. And so they're having to address just some of their core infrastructure. Maybe more importantly is apprehension or risk. I do think there's like, and maybe well-justified reservation from companies about turning over too much of the workflow to be run by agents because there's this fear of, you know, either hallucinations or alignment risk or a variety of risks

[00:13:54] that they want to figure out before they embrace that too much. And the last one would just be, I don't, I think a lot of companies just aren't familiar with the, you know, the possibilities or the capabilities of some of those tools. I think what we saw, 70% of employees said they had heard about agents and only 15% could describe what they actually are. And I think that, you know, mirrors a lot of companies too, which certainly has inhibited adoption.

[00:14:22] And another topic this year is how IT has added shadow AI to the list of losing battles. Others on the list, of course, would be shadow IT and before it was BYOD. But I mean, with more than half of employees more than willing to use unauthorized AI tools on their machines and devices, how can businesses strike that right balance between the innovation

[00:14:47] and the increased capabilities and productivity and, and as an ex-IT guy, that governance, you know, the boring stuff. Yeah, it's incredible, right? Like one of companies' biggest concerns is not being able to get employees to adopt it. And then we see the data that says 50% are using it for work outside of their IT infrastructure, to your point, which is kind of ironic. One is access to quality tools.

[00:15:14] I think we see if companies aren't providing quality tools to their employees, then they are finding ways to go outside of the enterprise ecosystem and use tools. And then the other thing I would say is just really, really good quality training. And I think specifically on that one, there's huge risk for an enterprise that needs to be

[00:15:35] managed, I think, by both IT and HR about using tools outside of the enterprise IT infrastructure. Obviously, there's, you know, huge risk with that. It's not just an IT job, though, to help navigate that risk. It is an operating model and a culture change that some companies have to have to make sure that their employees understand the risks and that they practice the right behaviors.

[00:16:03] The other piece on that shadow IT, which is maybe not surprising, is you see, while 50% of employees are using tools outside of the ecosystem, for Gen Z, it is far higher. And so you would expect that trend to only increase if companies don't get on top of it quickly. Yeah, such a great point. And when I was doing a little research on you before you came on the podcast, one of the

[00:16:28] things that I found was that you said that the secret to AI success is 10% algorithms, 20% technology, and 70% people and processes. And it is so true. But for people that may disagree with you, why does that human factor matter so much in this equation? Because I think it often gets lost in large organizations. But it's more important than ever, isn't it? Yeah, well, and this is based on empirical data. We run consistent tracking surveys.

[00:16:56] And this data that you mentioned, the 10-20-70, as we call it internally, has been consistent really since companies started investing in digital transformations, even prior to AI. And the people process is so important. I mentioned earlier just reimagining the workflows and how critical that is. But clearly, talent and culture are also just really important drivers of a company's ability to get a return on investment.

[00:17:26] And so if 10% is data, then 20% is the tech infrastructure, as you mentioned. You still have to have really good, high-quality talent managing that other 30%. And we do see companies who have been early adopters and companies who are what we define as leaders. They're actually extending their lead. And a lot of it is because of the people in process. You see those companies attracting better talent.

[00:17:52] And so consequently, they're able to realize value more effectively and more quickly in their investments. But yeah, the 70% on people in process is huge. And you are in a fantastic position here. You get to not only speak to customers and clients all around the world, but you also get access to a lot of information, a lot of data, a lot of reports. So when you put all this into your own little AI in your brain, what excites you about the road ahead from everything that you're seeing and hearing?

[00:18:23] I think a lot of us are in the consulting business because there's a little bit of ADD in us, and we love change. And we love to, you know, just to be a part of what's at the bleeding edge. And I think what excites me is just the amount of change we have in front of us and really the quality of life improvements that will come out of this as well. I think there's really exciting opportunities in the field of healthcare. Like back to your point on us having, you know, pretty good data and pretty good visibility across a lot of industries.

[00:18:53] I'm really excited about the impact AI can have on humanity and solving some of our most complex challenges, whether it be healthcare or food scarcity or things like that. And then I'm just fascinated to see where the technology takes us. I think I've, you know, seen some of the tech leaders over the past couple of weeks or even year just talking about how quickly some of the breakthrough technologies are, you know, changing and are going to be in front of us.

[00:19:23] And as I said, you've got access to so much information now, you might find it easy. It might even be more challenging for you as someone leading the road ahead. But for many people listening, that pace of technological change and how do they continuously learn and keep up to speed with everything? I think that's something many people are going to be thinking about as they hit 2026. And this is what I'm going to do differently. This is what I'm going to learn. And all those kind of resolutions before we resort back to our usual self by end of January.

[00:19:50] But for anyone that is thinking that, how do you self-educate? How do you keep up to speed? Any tips or anything you'd be doing there? Well, I do listen to a lot of podcasts. And I guess I'll start by saying another piece of data from the survey was asking people if they're saving time by using generative AI, what are they doing with that time? And you could say, how much of that productivity improvement do you get to invest into your own learning?

[00:20:16] And I think right now the data showed, well, companies aren't really demanding a lot of different things. And so they're spending it more on leisure activities and things like that. And I do think that spending time investing in yourself and learning is going to be critical to the specific question. I spend a lot of time listening to podcasts. I read as much as I can. And I have children who are, one, just started college.

[00:20:44] And I have twins who are just about to start high school. And I'm convinced that they need to also be learning about it. And so I'm trying to be a good role model for them. But yeah, lots of reading and lots of podcasts, primarily. Brilliant. And I think that's a brilliant moment to end on. But before I do let you go, for anyone listening wanting to find out more information about you, your work at BCG, et cetera, or where they can keep up to speed with anything, anywhere in particular you'd like to point everyone? Yeah.

[00:21:13] So we've been pretty prolific in publishing. We put all of that on our bcg.com website. We, as I've alluded to a couple of times, have a variety of different research reports that are out. One is called AI at Work, which I'd encourage people to check out. It covers a lot of what we talked today about with respect to training and adoption. We have research called Build for the Future, which is some of what I alluded to around value and return on investment.

[00:21:43] And then we actually just this week published research in concert with MIT about kind of the impact of AI on the workforce. And so those would be a couple of specific ones to check out. But like I said, we publish significantly. Secondly, people can also, of course, hit me up on LinkedIn and we can connect that way as well. And you can see some of what I've been posting about this topic. Brilliant. Well, I'll have links to everything you mentioned there. Make it easy for people to find out more information.

[00:22:13] Love chatting with you today. We had a little bit of fun along the way. But more than anything, just thank you for sharing your insights today. Pure gold. Thanks for joining. Thanks, Neil. And thanks for all you do. So this is great. I think what stood out most in that conversation was just how much of the real progress happens far away from those headlines in our news feeds. Yes, AI capability is accelerating at a wild pace. Yet the human readiness that unlocks its value is often uneven and overlooked.

[00:22:42] And David's point about immersive learning landed strongly for me because one hour workshop scattered pilots might look neat on a slide and ticking a few boxes, but they don't move the dial. It's when people receive focused hands-on coaching and see leaders using the tools themselves. These are the moments, the magic moments where confidence rises and real change takes hold.

[00:23:06] So it's a reminder that adoption is built through rhythm and repetition rather than announcements and slogans. So thank you to him for bringing some much needed clarity to the debate around AI agents as well, because I think plenty of companies are curious right now. Very few feel fully prepared and that is okay. But many still need to strengthen their data foundations before they go.

[00:23:32] One of the repeated phrases I'm hearing at conferences all over the world is no data, no AI. And despite all this though, optimism from workers who understand the tools is quite striking. When people see how these agents fit into the wider workflow and understand where human judgment remains essential, the fear fades. And that tension between excitement and caution, I think it's something that could shape 2026 in a very real way.

[00:24:01] And leaders who guide their teams with transparent communications, rather than just trying to replace people with technology. These are the ones that will benefit the most. So as we wrap up, I'm keen to hear your thoughts. Where do you see yourself in this adoption story? Are you feeling energized by the pace of change or overwhelmed by it? And if you had to choose one area to upskill next year in 2026, what would it be?

[00:24:28] Share those reflections with me because your experiences matter just as much as the experts, as the expert insights on the show. So please head over to techtalksnetwork.com. You can leave me an audio message there or go to LinkedIn X Instagram, just at Neil C. Hughes. But that's it for today. So huge thank you to everyone at BCG, my guests, and indeed you for tuning in every single day.

[00:24:55] I'm conscious I do throw a lot of interviews your way. So thank you for coming back. And hopefully you'll join me again tomorrow. Speak with you then. Bye for now.