speaker-0: Hey everyone, welcome to Chat B2B, the podcast that gives you a front row seat to AI and B2B marketing. I'm your host, Andrew Au, co-CEO of Intercept. And on this show, we talk with enterprise marketing leaders who are scaling AI initiatives. Each episode we explore what are they learning and what we can apply within our own organizations. Today's guest is Shohab Shaikh. He's the senior manager of global campaigns and strategic programs at Lenovo, specifically the Solutions and Services Group. And his team is responsible for some of Lenovo's biggest partnerships, including FIFA and F1. We get into the real experiments that his team is running from AI-powered avatars and videos to AI-powered research. What is working? What are they iterating on? And finally, we get into a shift that every B2B marketer should be paying attention to. which is AI is no longer just a tool, but a member of the B2B Buying Committee. If you like what you hear, don't forget to subscribe wherever you get your podcasts. This is Chat B2B. Let's get into it. You're the senior manager for global campaigns and strategic programs at Lenovo, specifically the solutions and services group. ⁓ For those people who maybe find themselves more familiar with Lenovo devices versus the SSG side, can you just give folks a quick overview? What is SSG and just your role within the mix?
speaker-1: Yeah, for sure I do. So Lenovo has been around for close to 30 years. And for a large part of those years, we've been really good at selling PCs. Recently, last four or five years, we've really gotten more powerful at selling end-to-end solutions. Right. that is redefining how digital workplaces are done, how AI is actually implemented. We are users of AI, but in the back end, a lot of the technology, a lot of the solutions are being driven by companies like ours. And one of the goals of my role over here is to help the market, help the perception of Lenovo to move forward from being, hey, they're just the PC guys. Wow, they have all these solutions. And they're really my one-stop shop. So if I want to go from the hardware to the entire infrastructure or the entire solution to run my business, so I can essentially future-proof us. And there's been a lot of our investment and energy is going. So even in some of our recent ⁓ tech world announcements, we're talking much, more about solutions that are transforming businesses around the world and how Lenoho can actually bring that to you. That is my role here. It's to make that exciting story a lot more available to everybody who's our customer and our prospects as well.
speaker-0: Yeah, I'm hearing more more again, obviously Lenovo is an organization is doing very well, but the shift where people kind of see Lenovo is more of a holistic technology partner, right? Like a trusted technology partner, but one that can kind of help them in this whole AI era, which we're going to dive into more about what that means. I think that's the new background you got is actually from Tech World, right? And that's what that resembles is the AI era.
speaker-1: Absolutely. this is a tech world background. I hope it's coming through. It's our logo. It's about how we want to help usher in or empower further the AI era. And Lenovo is actually that portal. It is that one technology company that can do the hardware, the software, the solutions, the services to empower AI in your organizing team. And that is how we are trying to work with a lot of our customers, a lot of our partners to bring AI in a more meaningful, scalable way for our customers. So it's quick. We want to all get to the ROI of AI and we'll have to get there.
speaker-0: which I think is actually very topical. want to dive into that too in a minute around that piece because I think there's a lot of discussion around how do I actually get impact? And there's a lot of talk about organizations struggling, right, in that regard. coming back to, I mean, your role, mean, global campaigns, strategic programs, you cover obviously a lot of ground. When you think about, you know, identifying what's the right use case or what do you prioritize, right, with bringing AI into your workflows, your team's workflows. How do you go about doing that? Because you've got a pretty expansive landscape that you're kind of, you know, you're accountable for.
speaker-1: So as a part of my role, I do all the global campaigns with some of our big partnerships with FIFA and F1, as well as some strategic motions that we have, ABM, research. And we always have to look at AI solutions that are not only quick to deploy in one of these things, but are also result-oriented. So we have to get from, AI is great. It can do these things. But let's see how. Let's test it out. And the other things that I'm definitely trying to actively do with the team is, if you've got an idea that AI can implement, how do we make sure we try it quick and easy so that we can see the results of how it works? Because in a lot of cases, the show real to the actual reality vastly different. And that is something that even we at Lenovo try to do for our solutions, right? Push it forward. Make sure it's real. And that is where we're seeing a lot of use cases. One of the ones that I've ⁓ I really like to talk about is we talk about AI's ability to do ads quickly. So we picked a task that was pretty easy. said, let's give AI an ad and let us ask AI to make adaptations of it. Like super simple size adapts, one part of what we do, we said, OK, here's the ad. Here are the mean five sizes. So here are the mean five sizes, but.
speaker-0: Yep.
speaker-1: In most cases, the letters were outside of the required size. were stuff all over the place. And like a human designer would not have done that. So it is pretty awful. You still need to have a human in the middle. But you definitely need to keep trying out and testing some of these things. And then it is a decision about how much effort do we want to do to teach this AI so that it is useful in the long run. We'll say, it a quick test. see if it's working, cut your losses and move on. And I think that is an important use case here.
speaker-0: Yeah, it's actually, interesting because, you know, one of my, uh, one of my favorite quotes is, uh, you know, our problem well stated is a problem half solved, right? And that's one of the ways that we look at this too of there's so many things that we can do, but what are the things that actually add value? Right. And it's, it a problem we're solving? Like, what is the quantifiable, you know, impact of that? Like obviously the ad variations. has a quantifiable impact. There's a direct cost to what you would have paid a designer or an agency partner to do. And so it's like, OK, well, how much do we do that? What's the volume of that? And then you can start to quantify, OK, is this a worthwhile problem to solve?
speaker-1: And take the risk, right? I mean, we want to be able to empower our teams to take the risk. Because it's that leap between, hey, look at this cool video that AI can do. How do make it real for my organization, for me? How do I add value to it? And then see if it's actually viable or not. We would have all thought making ad adapts, size adapts, right? It's pretty easy. Get a designer to do it. It's easy, when you lob it over to AI to do, you actually have to train the model to say, OK, think inside this. your spot. And then you've got to decide, is this an effort or an exercise that I really want to do to train the model and really see what the long-term impact of it is. But that's real use case.
speaker-0: Yeah. Exactly. And that's it. It's like, if you know the workflows really well, you can start to pick off these individual use cases that over time accumulate to a lot of work, right? Or a lot of costs. Outside of the adaptations, there other pilots that you're either running or maybe thinking about running as an experiment?
speaker-1: Absolutely. So we have actually taken the challenge within Lenovo to embrace AI and show more of the measurable outcomes of using AI. So as a marketing organization, we're trying to implement it in a lot of different things. So the AdAdapter is just a quick example where we're definitely using them in research. So we are trying to figure out how AI can help us either scale or draw parallels. from the data a lot faster. That is ⁓ using AI to both test what we feel about our customers using AI to act as a customer and see how the customer would respond so that we can help train our sellers, our marketers to do better. That's one aspect. The other aspect is how we can start to do more personalization at scale. We know that our customers like to hear a solution that's baked for them. And growing customers, actually, people like to have the answer to their questions. with AI, you are able to scale up and say, hey, I can answer your specific question. Let me show you how it is unique. Then there is another exciting bit. And then the third one that I actually like is about being able to tell stories a lot faster. In a video world that we are in, there is a lot more video consumption. So if we are able to distill a 20 page white paper into a minute and a half of video that has a compelling outcome-driven story. But we've got a lot of insights to share, right? So we use AI to really operationalize this and have a lot more content that resonates a lot quicker, a lot faster. In fact, the most recent AI video-generated gen AI use case, we've actually been able to do that at almost 25 % of the cost. So we know people are consuming video. We're now able to create that at 25 % of the cost and really use it to scale how we tell more meaningful stories to our audiences. It's a lot of use cases.
speaker-0: this powerful, I mean, that's a 75 % reduction in cost, right? On the video production. And are those like the more AI avatar styles or are they just like illustrative? they?
speaker-1: They are more the AI avatar styles that we're using right now. It's very interesting because we could definitely do some of these videos with the animation styles and stuff like that. But we took a decision to say, hey, let's make a generative AI person kind of tell the story. It showcases a person in action. And we're trying to balance this out with real world story. So we take a real world example from our customer, and then we try to showcase what it would mean for a different customer in any different industry. So really bringing together personalization, storytelling, and to be able to do this at scale is ⁓ a phenomenal example of how we're trying to bring AI. ⁓
speaker-0: I love that. think that I think avatars have come a long way, you know, like when they were, as people say, it's sort of that uncanny valley of like, it's not quite there in terms of realism. And I think that in some extent, maybe where you have a little bit more way to go. But in our experience with the AI avatars, I mean, they speak 180 languages. You can create custom, like, you you can create your own, your own personas. And I know that for example, at Qualcomm, one of their use cases is actually taking a video that's produced in English and actually dubbing over with the, obviously the mouth movements, but taking the actor itself and actress themselves. So I think you're right. think video, there's a ton of opportunity there. And that's, something that is not just more cost efficient, but something that is a new capability to be able to do that in however many languages in real time.
speaker-1: I think the nuance here is we like to consume content that is personalized to us. So as we have this great tool that adds value and customization and personalization, that's when we use it to tell compelling stories. There is still an element of video that needs to be produced at scale and all of that. And that is a part that is working. But for more impactful, more personalized storytelling, This is a great value addition. And I resonate with it too. I watch a video, like, I wonder what would happen if I was in this scenario. Now you can find out. The tools or the capabilities to find out. Make this for me. It's phenomenal.
speaker-0: Yeah. personalization is the ultimate pursuit that is not a new pursuit, but one that we couldn't really have done well before. Like it's not possible, like it's cost prohibitive to do that. But with AI, like, you know, as you and I have talked about before is you can create these one-of-one engines, right? So we're even pushing that into content as you think about white papers and eBooks. How do you take a single asset and make it a one-of-one where If know, show up global mandate, ⁓ campaign strategic programs, you download it. How do you get a different asset as opposed to somebody is, you know, myself ⁓ on strategy, right? And within a marketing services organization. So that I think is really the unlock. And I think you're right. And that's multiple modalities, right? That's not just written in this video, as you said. So.
speaker-1: Yeah, in fact, because of this, and we're seeing this with our solutions buyers as well, right? They want to be able to tell their story to their stakeholders. What you are now doing is empowering your customers, empowering your storytellers, right? To be able to take your content and then add to that in their own personalized way. We all have great stories that we like to tell and now I'm allowing you to use content
speaker-0: Yes.
speaker-1: to tell that story within your organization too, because of the power of personalizing this. So it's a lot more personal. I look at the impact of research to me, to my company, to my buying habits. And with the power of AI now, you are able to customize it like never before. So it is a use case that is now vastly exploding. It is something that we are using. And I think the potential is immense because the value that it is giving to the buyer, to the buying group, phenomenal. I can tell my own story.
speaker-0: Yeah, from your perspective and depending on where you sit on the buying committee, you could tell the right story to the CFO, the COO, the chief legal officer, right? It's a bit of a different version of that story. Do you find as like a technology organization that's really sort of at the frontier of this, do you feel added pressure to be able to bring in these, these like, you know, as people say, dog food or drink your own champagne? Do you feel that pressure?
speaker-1: So yes, we have embraced the pressure and we've kind of challenged ourself. Because even as users, we now consume a lot of AI content. We use AI to make a lot of decisions, validate decisions, inquire and understand. We also try to use that to then empower how we turn up in the marketplace. We're trying to take that more as a challenge of how do I drive efficiency scale improvements by using some of the AI. And even if you look across your personal habits, a lot of our discovery journeys are now starting with AI. Hey, what do you think about this? How do I do this? That's right. One of the five questions I should ask.
speaker-0: Yes. Yeah.
speaker-1: And if we're doing that in our own personal capacities, that is soon going to expand into our organizational decision-making. So it is an interesting challenge as buying groups become more complex, as the discovery of what your solutions could be become more AI-driven, as a marketer, our challenge is to now turn up where our customers are looking for information. So I was just rounding it off. We have to turn up where people are searching for things. So if people are searching for things using AI, how do I now convince not only the customer, the human, that, this is a good prospect. Let's start exploring. But I now have to additionally convince the AI to view my content in the same way. So my customers have increased. My prospects have increased. And the way that I have to utilize my content, is an interesting, a phenomenal challenge. that is influencing buying decisions today more than ever before. And you've got to get ahead of it.
speaker-0: It's completely transforming. know we're going to want to dig into that one ⁓ shortly because you're right. It's like AI is becoming sort of the intermediaries between a lot of our clients and the decision makers right across the mind committee. ⁓ To add some sort of data to this, Watchtower is we've got an AI powered research platform. so at this point, our B2B marketing audience is around 550,000 B2B marketers specifically around the world. And so we looked at what were the conversations related to initiatives that they're going to scale in 2026. And so the top five were AI research assistance, which you mentioned research was one of your workflows, multi-agent workflows, AI search. AI campaign orchestration and agentic campaign orchestration. a lot of automation, obviously you mentioned research, ⁓ search, obviously that's transforming with GEO. How does that kind of list compare to maybe what you're prioritizing within your team? I mean, you mentioned research already, so that's kind of one for one. What are your thoughts on that?
speaker-1: So we're trying a lot of different things. We're definitely trying the research. We're trying the campaign orchestration. We're trying the personalization. But I think the challenge that we are kind of getting over is there's a lot of different types of AI that you can deploy and implement. There's a lot of different verticals where some have gotten really good, some are getting better. And then you as an organization have to make a choice about, do we want to put all of our fingers in the honey jar? or just 100,000 at a time. So that is a balance that we're trying to do. And I believe the role of the human in all of this is still important to the human in the middle. If nobody looks at the outcomes of some of our research, we're just saying AI said it and it's a we want to be sure that there is a human in the middle. It is looking at some of the aspects. And that becomes a scale question. Can I hand you everything with AI? Sure. Probably not today, right? So, we still have some time to get there.
speaker-0: Yeah. It's, it's your right, because it's ultimately as humans were accountable for the decisions and the outputs. so, you know, you can't blame Bob the agent for messing that up. Right. It's, it's, it's on us, right. It's on us. Yeah. You know, it's the, things too, that I, when I see this list, it's, shows me that I think, as you had mentioned, A lot of us are thinking about how do we create this sort of 10 X impact of how do you create like workflows, like end end workflows that can actually do these outputs. Right. And there's sort of the different two modes of thinking. There's like the 10 % thinking, which is like, how do I get 10 % better versus 10 X, which is how do I completely transform these workflows? And I think that's kind of where a lot of our clients are kind of like their head is at how do we transform workflows, but they are also handcuffed to some degree. with obviously, you know, the pace that internally they can move. you know, we're in, you know, compliance and privacy obviously is really important data protection, particularly for global enterprises. ⁓ but there is that, that tension that, that I do see, which is I want to move really, really fast, but internally we're not built to move that fast. And that's, that's a struggle for these global. mean, I'm preaching to the choir here, but You know, you live it every day in some respects.
speaker-1: It's an interesting challenge, right? As organizations become larger, there is a much bigger risk factor involved. So I recall we wanted to try an AI bot or an AI agent to do something. We thought, hey, this is quite easy. Let's pick it up. Let's see what happens. And let's deploy it. And then we bounced it over to procurement and said, nice, simple project. Let's get this out the door.
speaker-0: Yep.
speaker-1: actually looking at it in a lot more scrutiny because the impact of deploying something like that for a larger organization has wide-ranging reach. So they have to make sure that a lot of their checks and balances are in place. oftentimes, you find the early innovators who take a technology and drive 10x improvement. Like, let's do a 10 factors. And then you have parts of the organization whose job is to ensure continuity, security, safety. And then you've got to figure out how do we do, how do we kind of work together in this? Because you can move fast and break things. And sometimes it's OK. It's OK to break things and fail and figure it out and always take that risk. But some of these existing systems are in place to make sure that we don't break the whole system.
speaker-0: So great, the important things.
speaker-1: Don't create the important things. So I think that balance is important. being in marketing, being in that side of the organization where people say, hey, can we try this out? Let's see what happens. I love to push boundaries and say, let's do it. Let's give it a shot. And then we need the functions to look at it in larger pictures. This is great, but the risk is too big. Yeah. Because you have to see the larger picture. And I think as organizations implement or adopt AI more, you will start to see the use case for pre-existing solutions. Yeah. Right? So if somebody has used it once and you have seen ROI from it, can I buy that? Is it repeatable? Is it scalable? And then it kind of circles back to your initial point. it's repeatable and scalable, it's probably not getting you 10x. It's getting to 10%. And you've got to figure out where you are or what project it is that you are working on, where you want to go. All the way, figure out incremental needs and bring the entire organization along.
speaker-0: Yeah. And, know, you got to experiment, right? And, and, ⁓ you know, my, ⁓ my second favorite quote is on, on experiments, which is, you know, if you know that, if you know it's going to work, it's not an experiment. So you have to be able to take a real risk. So I, I, I wanted to ask you, how do you build that sort of risk appetite internally or, or, or navigate some of these conversations and just your thoughts on not just. fail fast and break stuff, but failing smart. know that you've talked about before about everything doesn't have to be successful. Let's be honest and transparent on this is an experiment. We don't know how it's going to go. What's your perspective on that? And then how do you have these conversations internally?
speaker-1: So I think as a goal, I've kind of taken it up to try things knowing well within the risk appetite that the organization is in. Right. Because if something works well, like you rightly said, it's probably not an expert. Right. And if it works well, you've got to scale it. But if it doesn't work, your opportunity to learn, to grow, to improve is far more. So you're within the team here. Let's try something risky. Let's break the barriers to what everybody thinks is normal and let's try something different. And in fact, there are a couple of examples that we've implemented AI. It's not worked. We've recognized it's not worked. And now we're in a position to say, okay, if you want to make it work for this, this is the kind of effort that is required to put into it. And this is the ultimate long-term vision of So now you've got this interesting point where you selected, we know the capabilities of the system. So either we have to invest to upscale, up train the people, the system, the AI, the data, and therefore here's the outcome. Or now we can start to look at different alternatives, maybe different solutions, different AI, in-house, app-soc. But we wouldn't have got here if we didn't try. We were just studying the, oh, hey, here's the potential. the unaddressed potential is always much larger than if you tried out. And the conference is trying out some of these things.
speaker-0: Yeah, and you have to try. I I was just chatting with a global MSP called Adastra. And I'm hosting one of their fireside chats later this week. And they are a global MSP. But they had one of their clients, which was the Mark Anthony group, who makes food and beverage. They're a global food and beverage company. And one of their heads of AI was talking about, look, we actually had to set up a real incubator to test a lot of these things. We had to sandbox it because we had to set the expectation that a lot of these things are not going to work. so it was also like a mindset shift that the organization needed to have in order to really embrace it because you have to embrace, you know, the way they put it is like, you know, not necessarily fail fast, but learn fast. Right. It's like, it's a failure, but you're to learn something. and I've seen organizations, ⁓ like Fannie Mae is an example that actually this was sort of, I'm going to go back. This is almost six years ago where they introduced this concept of, they called it Friday failures, which was, ⁓ I know it was brave, right? You had. leaders that would gather with their teams on Fridays and share the things that didn't work and what they learned from it. And this is all in the context of transformation and modernization and so forth. And so I think that we're going to have to learn a lot. We're going to to stub our toe. But you have to have the risk appetite to at least stub your toe and it's okay. You're going to learn.
speaker-1: And this is a great example. So in the early days of using AI for research, we asked our AI model, here's all the research that we've done on our buying group, our personas, consumer. And it did. And then we started asking the model, OK, tell me, what do you think the conversation groups will be for these people?
speaker-0: Yeah.
speaker-1: And in lot of cases, the results were pretty generic. Whereas the data showed that people would actually like to get a little bit more specific. And that's an interesting point where we had to look at this and say, I think it understands the essence of the research. It understands how to dissect the data and give us different variations of it. But if I had to use the data, the model as it is right now,
speaker-0: Interesting.
speaker-1: And if I were to throw a question saying, knowing what you know, can you tell me would this pitch work? we would have to take it with a grain of salt. Right? course. And that is an interesting exercise because for us, we had to train the model. We had to invest the time. We had to put a lot of the research in it to figure out that actually we've got something out of it that is not what we intended. Super useful. Yes. It did not drive the original purpose, but we were still able to experiment with it enough to figure out, OK, now we've got this different tool that we can use to
speaker-0: this.
speaker-1: to drive more meaningful conversations. And what we've been able to do with that kind of output is whenever we have new offerings and new products come through, we always take it back to the data and say, we know our segment. We understand this offering. Where do think we should start the conversation? And it is able to give us from the data saying, why don't we consider these three or four routes now? Because it's obviously learned and it's gotten better at it. But that is a time investment with AI that that I think everybody has to be ready to make. It's the investment in the future outcomes.
speaker-0: Yeah. I think it's a great example because you know that there's potential there. You know that if you continue to train the model, you're going to get more value. So you have conviction there. And I think the other thing too, is you think about it. If it's better than you, the team sort of like flipping a coin, if it's still directional, there's some directional guidance there that is valid. It's still valuable. It's just, do you get it even more specific and your use cases and all that comes with training and time. But I think if people expect, and this goes back to, think your ROI point earlier is like, what's the ROI of AI? I think that the, the revenue side is a, is a lagging indicator. If you do things right and you've got the conviction and you, you know, you, you, you're, you're working on your pilots, you're iterating and you're proving them out. It will generate revenue, but it. There's not, if you expect that to come so easily and immediately, you're never going to get past this first, first base, so to speak.
speaker-1: That's a great way to look at it. When you look at AI, revenue is one way of measuring. In marketing in an organization, we've picked up how we define the ROI of the AI that we're implementing. We're trying to look at it from a couple of different things. Revenue is the most important one, but we also want to look at it as efficiency. How are we able to drive efficiency using AI? Are we able to drive skill using AI? How are we able to drive personalization using AI? All the things that allow us to turn up more meaningfully in front of customers, and how are you measuring them? Now, we're figuring it out, right? We've not got the answers. But at least by being able to expand how our organization looks at the benefits of AI implementation, we are establishing parameters that we can improve. So we have some projects that purely drive efficiency, ultimately leading to revenue. We have some projects that drive purely personalization, ultimately leading to revenue. But we are able to experiment on more parameters than just revenue alone, which allows us to expand on the types of experiments, expand on the types of use cases. Because if somebody turned around and asked me how much revenue did that one AI-generated video generate,
speaker-0: But I would say you go back to any sort of marketing activity or even sales activity and say, well, hey, what was the direct response from that? And in many cases, we can't quantify it. It is a correlation, not necessarily direct causation. even if it's the last, you use a last touch attribution. this is the last thing. Well, events was the last thing that they went to. Does that mean that events? is what's driving the revenue. Not necessarily. It's all this stuff that we did leading up to it. Right. So attribution I think is such a, is such a tricky one. Obviously something that is very topical in the news that's always coming up, which is AI's impact on headcount. And so ⁓ in the Watchtower study, like I said, over half a million B2B marketers specifically, when we asked them how do they expect their headcount to change this year in 2026, a lot of it was, ⁓ I'll read these out, she said, reduce roles. The AI was 17%, centralized also AI role 17%, freezing new hiring 15%. So maybe speaking on behalf of yourself and not necessarily in Lenovo, how do you think AI impacts team design? Like, what is your perspective on that one? Cause that's such a tricky one and lots of news on related to this coming up, obviously. What's your perspective?
speaker-1: I think this is an interesting one, right? AI as a tool has now allowed us to generate a lot more value within the same amount of resources. So even you as a user, you now have access to a lot of tools, of course, within your organization guidance that you can use to generate a lot more value. And now you have this opportunity where you can become a value generator.
speaker-0: Got any magris? Yes.
speaker-1: And if you are on the side of the organization and you're pitching up a role that allows you to use AI to generate value, you have a far more valuable tool set at your disposal. Your value as an employee within an organization exponential, thanks to AI. The ones where there is a little bit more risk is if you're looking at the opportunity in front of you and you're saying, want to have happier
speaker-0: Yeah, so you
speaker-1: to continue to take orders to execute, to execute faster. I think that is the side that I would encourage to embrace AI to drive efficiency, drive scale. Because ultimately, AI is a tool that is going to help us drive efficiency and scale. And that is how users, employees need to use it. yeah. Overall, it impact?
speaker-0: Yes.
speaker-1: the shape of the organization, whether it back the usage or the organization structure, I think it will drive a lot more value to the people who are able to embrace it and use it. And it is not that you need to become a coder overnight and figure out stuff. I'm not either, but I'm a user. And I think with the right mentality, you can get ahead of it. So I think organizations will see the impact of AI and will take the benefit of the value that it drives. You just gotta decide on which side of that value coin as a user you want to
speaker-0: I like that perspective and I believe that because I think that, you know, rather than sort of victimizing oneself, you can say, this is, this can absolutely empower me. And so when I think about, and I mentioned this before, but when I think about previous technologies and I know AI is very unique in the acceleration, but you go back to the seventies where they launched the ATM. right, automated teller machine. And, you know, the, the prediction or the forecast was that we are not going to need tellers in a matter of years. And so when they did the study to look at what, how did ATMs impact tellers employed and over, was about, ⁓ I think 20 or 30 year span, the compound annual growth rate was 43%. So we actually grew the number of tellers by as a society, as a workforce by 43%. There was an initial ⁓ state where it contracted and then it grew. And so when you, when you look at the data, the thing that I see is that what people as tellers weren't automated out of a job, they were automated out of bank telling, which is like giving money, receiving money. Like you go to a teller.
speaker-1: Yep.
speaker-0: You can go at, like you're talking about your mortgages and you're like more complex things, right? Like to your point, they're elevated. So I think we will see a net positive, but to your point, it's a mindset. Are you going to look at yourself as a victim or are you going to look at yourself as I'm empowered to create value?
speaker-1: Absolutely. I think that's a great example, right? Like old history, you've always seen a lot of these new technologies and there's a lot of fear when you have new technologies come to market. The credit card, ⁓ my God, we're not going to need cash ever again. Like cars, what are all these horses going to do? Right? Like we've had a lot of these technology, technologies that have really changed the way humans get up. Get out, do work, spend money, spend time. And in lot of cases, fear around what is going to happen to the old way of life quickly transforms into how am I going to use this to add value to the old way of life. And we have a lot of these tools now that are unlocking exponential value potentially. So we're able to do so much more in
speaker-0: Yes.
speaker-1: almost the same amount of time with almost the same amount of resources. And it is so much more meaningful to our customers, to our prospects, to our company, right? It is about embracing that and being able to derive value from that. Everything on the right side of that choice. Because at the same time, you have detractors. Like with every piece of technology, you say, hey, this is not going to work out. Which is fine. Everybody has their place. You need the cynics to kind of challenge you to do more work you have.
speaker-0: Yeah.
speaker-1: And I think that balance is important. And it's also different teams in the organization adopt AI at different bases. Absolutely fine, right? Which is a learning curve that the entire organization can go through. And that is essentially what a lot of us at Lenovo are doing too. Like different organizations have picked up AI for different sizes and scales. And I think marketing has been one that has picked it up and said, we're going to do it all.
speaker-0: Yes.
speaker-1: We're going to try it all. Sure, not all of it will work out. We're going to take those learnings and we're going to allow the teams that work with us to be able to embrace those learnings, figure out what to do, and then move forward from there. But you have to have one of those groups that is ahead of the curve and really pushing the limits to what AI can do for you. And marketing has taken up that challenge.
speaker-0: Well, kudos to you and the team on that because I think that there's a lot of, I would say that that's more of an exception than the standard right now. I think a lot of people are a little bit hesitant and it sounds like there's the right culture to say, let's experiment, let's try, truly experiment. Obviously we want this to be successful, but not everything can be. You your point on the value generation, it reminded me of there's a pretty big AI conference called Maycon that happens in the fall in Cleveland. And so we're partners in terms of, know, I'll do either a keynote or we've got sort of a session there. And one of the keynotes from the few months ago was one of the founders, Jeremiah O. Yang from Blitz Scale Ventures. And so he talked about how before it was how many employees you had and all that. The metric is now changing to sort of revenue per head. And so they looked at uh, SaaS companies, which I think was around the, like an average SaaS companies around 250,000 revenue per per per employee. said a Salesforce or somebody that obviously is like a dominant player in the space is closer to 400,000, 450,000. And do you know what that number is for SF startups, like San Francisco based startups? Like I didn't know either. Like he just used to tell 2 million per head. And so I think to your point on value generator, this is an example how if you're an AI native organization and it doesn't mean you don't have, you know, you could still be human first, right? But if you are AI native or AI forward, that's the kind of value that each one of us can have in the, like that's pretty, that's pretty incredible. So just the numbers on it.
speaker-1: Yeah. It kind of transcends across, right? So you now have a workforce that can add value to your work. And it is, and you have to decide how you want to use it. It can be doing a lot of the repetitive work. It can be questioning all the thought work that you're doing, asking the tough questions, right? Challenge me on these three things.
speaker-0: Yes,
speaker-1: And how can I make this better? And you have another capability using some of the existing AI models to pretty much try out and challenge yourself to do better. And you can pick, right?
speaker-0: Yeah. And I was just going to say, speaking of challenges, in the study, we asked these B2B marketers, what type of, like, what are your biggest organizational blockers to adopting AI and embracing AI? And I'll read you out some of the top responses here. Limited AI skills, no training program, no AI governance, too many tools. and limited IT support. to me, what was interesting is that none of these were technology problems. They're all people problems. So I just curious, what's your reaction to that on that finding?
speaker-1: I think it's unique, right? As organizations are being born in the AI world, their challenges are unique too. A lot of these organizations are born AI ready, so they're able to ramp up from get-go to 100 people, 50 people pretty fast. That's when scale has a lot wider reach. And I think I'll you an example for our organization.
speaker-0: Yeah.
speaker-1: ⁓ a couple of these AI opportunities that we wanted to explore. So it was a pretty quick route to market. We said, let's try it out. And our biggest blocker for making it happen was procurement. Procurement wanted to look at it and say, wait, what exactly is the outcome of this? And how is it going to impact the rest of the business? So it was less about the technology behind it. was less doubt in the implementation. It was more about what else. I think as our organization is going up, as AI adoption use cases is going faster, I think you start to see more of these types of challenges come out. You have to experiment. We would have never known that the biggest challenge to implementing AI is not the model, it's not the infrastructure, it's our procurement guidelines. So even if we had all of those things checked off and procurement was saying, whoa, hold on a second. Right. So it is pretty interesting. In fact, this reminds me of ⁓ GDPR. I think with GDPR, was a thing. ⁓ Everybody was like, OK, this is great. It's going to change the way we do business. And there was this guy who were always saying, where is all of our data? So I think we're now getting into that wave where everyone's saying, let's get AI. And then people are asking more meaningful questions of AI for what? And how do you categorize this? Is this internal? Is it confidential? Is the data, what is it learning from? And I think a lot of that is how real challenges are going to be dealt with as we're kind of going through this AI wave, as we're in the middle of this AI wave.
speaker-0: Yeah. It's, you raised such a good point because yeah, I mean, just, just on data, like data assets are, are becoming obviously even more valuable. Right. And so I think as, as the models become somewhat commoditized, where like the leaderboard's changing, you know, month to month, a quarter of a month, really months a month at this point, it is like, well, you know, how do you, how do you protect that data too? And I do, I do, I don't know what your thoughts are on this, but I feel things are moving back on prem in a way of like, are, they're pumping a ton of very confidential stuff into some of these baseline models that you just no longer have ownership of a lot of that data. just don't see how there's going to be so much information sprawl and data sprawl that companies are going to want to protect that and bring that back in. mean, obviously with your infrastructure group would be able to see this, but Is that also like, that, is that what you're seeing too? Like there's a trend to move back to on-prem in a way or.
speaker-1: So I think the answer to this is kind of twofold. One is as models get smarter, we want to be able to train them better. And a lot of these models rely on being able to understand a data source that is unique to your business. And as we get into more data, more AI models, businesses will start to realize
speaker-0: Yeah. Yes.
speaker-1: If they have a store of data that is unique to them, that is their business IP, they want to be able to protect that as opposed to give it out to the cloud. They are everybody tree. So I think the investment on hybrid AI data, databases, AI models that work both in the cloud, where you can store a lot, as well as on-prem.
speaker-0: Yes.
speaker-1: is the next wave of AI average for marketing. We want to be able to make quick AI decisions based on what we have, what is available, what is accessible to us, and protect all of our customer information, keep it close to us in well-protected servers, as well as take advantage of the larger models. So we're always moving between these two types of data deployments, these two types of models, right? And essentially, that is how That is how we see more organizations embracing AI. Think of it this way, if you've ever been to a doctor or hospital recently, a lot of the ability to take a lot of that data and run it through an AI model and come up with scenarios, you don't want that data to take a round trip to the cloud and come back. it's not only expensive, it's bad for the environment. You can do a lot of that compute.
speaker-0: because
speaker-1: within the four walls of your hospital. And that is the technology that we as a company are now developing and implementing in a lot of hospitals as well. Keeping the data close to us, keeping the data close to the users, keeping the data close to the people who need it. it is an interesting time because we're moving away from send everything to the cloud. Let's figure out what is the best use case for this. It needs to be on a PC.
speaker-0: Yes. Yeah. Well, that's it, right? Now you've got on-device AI capabilities directly. And so I do think that as the models advance, they get more efficient and they will be able to run directly on the device. Like today's, you know, just, use JetGBT as an example, cause it's the most well-known, right? But, or, or Claude, the, the frontier model today, give it a six months or a year. You could probably run that locally. on the device because it's going to get more more efficient. I think paired with our energy crisis, we don't have enough energy to be able to, if you wanted to send everything to the cloud and put this in these massive data centers, you're just not going to be able to do that. I think the device and on-prem is actually making this comeback in a way, I feel, I think it was really going to, to your point on hybrid. You're going to need both. it's not like before, which it was just push everything to the cloud, which I felt that that was the direction.
speaker-1: Yeah, I think like we're going through this wave now even even our our devices have a lot of AI capabilities on it, right? Like it's not just organizations want to keep data close to them, but users too. Do creators now like to keep a lot of their content close? Yeah, they want to use their on Prem on their device. They they want to be able to do that and I think as as our portfolio of.
speaker-0: Yeah, that's true.
speaker-1: connections into the AI sphere expect, right? You want to be able to decide as a user, as an organization, what do I need? Do I need the speed? Do I need the safety? I need the security? Do I need the scale? How do I plan for both my data, my organization's data to make sure it's giving me the right solution? And as an organization, we want to be able to give you the option
speaker-0: Yes.
speaker-1: It's not one size fits all. Everybody sent it to the cloud. It's not there. Let's talk about your use cases. A lot of our solutions engineers do exactly that. What are you going to do with this data? What's your implementation look like? Here are our solutions that help you do that. So it's keeping a lot of it close by. And these are not easy decisions. In fact, AI is now even involved in the decision making.
speaker-0: Yeah.
speaker-1: which is phenomenal, right? AI is part of the decision-making journey, and that is really spurring how we turn up as customers, too. How do we engage with our buying group in a meaningful way with meaningful content, with relevant content, and start to drive that conversation?
speaker-0: One of the points that caught my attention in the studies that were running is that marketers are clearly looking to produce content for AI. In fact, when we looked at how B2B buying is going to change in 2026 because of AI, I think the number one response was producing content for AI. Now, I know before you had mentioned something like A couple years ago, that number was 40 some odd percent or something like that. Tell me, tell us more on that because I find this really interesting and how we're going to, how are we, what are we going to do about that? How are we going to pivot, you know, given that AI is becoming the intermediary.
speaker-1: We did some research about two years ago. We looked at our buyers and we looked at how they are making decision juggies. ⁓ What is their decision points? And in fact, AI and generative AI or search-based AI was one of the fastest growing categories for discovery. Now, two years since that research has come, so I'm sure it's much more important now. But two years ago, that was on the uptake. And two years ago, when we had that challenge and you look at what we could do, the options for us to work with AI to empower it so that it has all the right content to provide to its users the right decisions were pretty limited. Today we're getting into a space where we can talk to AI more meaningfully. We can say, here's our point of view. You can take it from a lot of different places. You can optimize the way you want to. generate content for a person who's searching for stuff. But two years ago, we had no way of doing that. So I think this has two impacts for us. One, as AI is becoming more involved in our customer's decision process, it is turning up at the discovery stage. It is turning up at the evaluation stage. It is turning up as a co-buyer. Yes. What is the role as marketers that we can take up to influence everybody in the buying committee? which now also includes an AI agent. So how do we empower not only the humans in the middle, but how do we empower the AI bots? And this is where we use generative engine optimization. How do we optimize our content to become relevant to models like this? And I think as we grow up, as AI becomes more ubiquitous across everything that we do, I think that is the next frontier. Back when... SEO was a thing. What are people really searching for? That was a thing. How do I have my website, talk about Italian food, but everyone's hungry? Nobody searches for I am hungry. They search for what kind of food do I work in? We're now getting into that phase with AI too. How do we tell the AI model, listen, you might be hungry, but you won't come to us for the subs. That's the model.
speaker-0: Always the food metaphors, right? Always the food metaphors. ⁓ No, I love it. I actually, think that this is, I feel like we could do a whole episode on this because one, now we're seeing obviously a paid entry model with chat GPT with ads. And obviously there's a lot of debate on that. Then there's the organic path, as you had mentioned, which I think is so powerful because I think that people are relying on AI heavily. in B2B buying decisions. And AI is basically rewriting perception in a way because if AI has conviction that, let's say Lenovo makes the best servers to run local AI models, which in many cases it does, I think that the buying committee, that is going to influence the buying committee. I think there's more trust that they put on these outputs. And actually it might even be even more impactful for the line of business folks, like obviously as we target IT and line of business, where the line of business are increasingly being involved, right? As we know in these buying decisions, but maybe they don't have the technical acumen or insight and they might be leaning more heavily on these models. So it is, I think you're right. It is such a large frontier opportunity to be able to win.
speaker-1: Yeah. think, I think the role of AI and the buying group is, is a friend, is an advisor, is a consultant, right? And we all love our consultants, but we have to take the information they give us and we have to distill it down for ourselves. But even if we disagree with them, we always consider them. And I think that is an important, important stakeholder for us to deal with now.
speaker-0: Yes. Yes.
speaker-1: It's just like back in the day when we had to turn up on page number one of search to be full set. There's nobody's going to page 256 to say, let me click on link four on this page and see what the answer is. That's right. So I think that is the kind of world that we live in now where you have to make sure we turn up in a meaningful way. But the challenge is not the same as with search because people are not as simple with their products anymore. People are
speaker-0: Yeah. Exceria
speaker-1: people are getting a lot more complex and what they want to achieve out of this answer. So they are asking a lot more meaningful questions. And that is where our role of marketers is important. We don't want to say, hey, if the guy is searching for these three keywords, turn up this ad all the time. We've got to be meaningful in the context that the buyer is actually using to find the solution. And that is a challenge. And that is a challenge that we've got to tackle. Honestly, we have to figure it out. We are experimenting. This is one of the things that I'm happy to learn fast and improve it. But it is how we use AI to benefit what our buyers are doing.
speaker-0: can share a little bit about some of the insights that we've had so far on this, which is, first of all, before we built our GEO platform, we did look at Profound and some of the others. And what we found with them is while they're great platforms in general for GEO, they are really built for consumer. And so when we would look, our CTO and our team would look at the underlying prompts that's being generated. It's very consumer and these are not realistic prompts that people send. So then when you start to see some of the data back that says X percent on Reddit or expert, like you see the source breakdown and that will change drastically depending on what the prompt is and the specificity of the prompts. So one of the ways that we got around this was ⁓ we actually, when we built our GEO platform, we interviewed IT decision makers and business decision makers to say, hey, if this was the scenario, what would be the prompt that you would use? And so we consider that, OK, that's your ground truth data asset, right? So it's like, OK, you've got that. Now, what you need to do then is generate prompts that are, if you're an awareness or consideration or purchase, like depending on what stage, I mean, it's not always obviously linear, but depending on buying intent, how does that change? And you need to run thousands of prompts. So, so we would put in sort of five, 600 per model. Say you run that across the five major models, somewhere around 3000. ⁓ but the, the platforms that we saw, you would run 12 or 20. like, this is too small. So, so there's a couple of things that we found. there's that we abbreviated into this acronym called beta B E T A. And so B stands for best for, which is you have to produce when you think about content strategies, who is it best for? You have to make that declaration, right? It's like, we are the best model for da, da, da, da, da, da, da. The, the E is evidence. So need proof points, which what we find is that if the proof sounds like bullshit, The model knows it's bullshit. so is it, if it's like, this is 500%, you know, 20 times you actually seen the reasoning that the model says, this doesn't sound believable. So there's the believable evidence. should say the, the, the T stands for trade off. is the toughest part that we are having conversations with our clients, which is T is about who are you not best for and try and take that back to the product team to go. Who are we going to state that we are not the best use case for, and they should look for an alternative. And they're like, no, no, no, no. We're good for everybody. And so if you don't declare that the models will come up with an answer, but the folks that are declaring who they're not best for actually are ranking higher. Uh, and then the A's answerability, which is format making it really stackable. like we, we call it, we call it beta because not only does it line up from a word. from that word standpoint, but it is going to change. It's going to be different pillars as the models reason differently and all that. Anyway, I'm going off on a rant. I think that GEO, you're right, I think is going to be the frontier win.
speaker-1: Yes, I think that's a good model. That's a good way for a conversation to start, whether it's with a human or with an AI. That's a great way to establish credibility. And I think the challenge for us as marketers is it is extremely difficult to tell somebody, I don't think that's the right customer for this.
speaker-0: Yeah, right.
speaker-1: because you can change the prompt in a couple of ways. And it was like, okay, now me is kind of the right customer. Right. So, so it is extremely challenging to say, I understand who this is really good for and he was wine and also who this may not be that good for. here's what, but our, our, our product conversations, our seller conversations would be like, this is great for everybody. ⁓
speaker-0: Yeah.
speaker-1: You've got to be able to take that insight and translate it into what are people really searching for? Because if I had to take it back to my food analogy, everybody who's hungry could eat Italian food.
speaker-0: X.
speaker-1: But not everybody who is hungry is eating Italian food. Right. Like you've your audience for it based on their needs and their personas and all of that. and people will always search for what's good food near me.
speaker-0: Yes, that's right.
speaker-1: The answer is not always Italian food, right? So I like Italian food, but still.
speaker-0: Yeah, me too. I'm a huge fan. ⁓ No, I think there's so much to get into on that. I'm not sure you and I will chat more on this on the on GEO side. So, Shoaib, I know you're a busy guy. So I'll ask you one last question, which is, you you look ahead to the next 12 months, which sounds like a sounds like a decade, right? In this whole AI era. But let's say 12 months out. What are you most excited about? What are you most cautious of as you look forward?
speaker-1: think for me, 12 months is something I'm a little cautious of because the modeling can change so much in 12 months, right? So what we're definitely trying to do, what I'm trying to do is look at the newer term, as much focus as the long term. And a lot of times, you have to pick one, which is the toughest thing that we go through now. But we have to be able to establish wins for today and experiments for today that will impact the long term. And I think that is where we are really trying to dabble in what we do, how we adapt technology. But I think from a long term, it is an extremely exciting time. We now have the opportunity to train a customer, to talk to a customer, to understand a customer that, for the first time, can talk back to you and actually tell you what they're thinking. I will tell you why this is not working for them. And now the challenge for us is, OK, now I know exactly what you need to hear. How do I tell you and the people that you want to convey this to in a meaningful way? So the next 12 months, I think that journey of AI as my customer to AI as my channel is going to be an extremely interesting one for marketing, marketers across the world to up. How do I express my product solution challenge? to AI so that they can help educate further. And I think that is something that is going to really impact the way buying will change, definitely the next year.
speaker-0: show up great perspectives. And, and, and I love the genuine enthusiasm that I share too around. This is an exciting time to be in marketing. It's an exciting time to be in knowledge work. know that there's a lot of fear and uncertainty right now, but I do think that to your points on B take that perspective of being a value generator, you know, rethinking some of the traditional ways of doing things and having the realistic perspective on Not everything is going to work and that's okay. Right. Let's go figure it out together. So, uh, on, on that note, show up. Thank you so much for joining us. Great conversation. uh, yeah, excited to have you on.
speaker-1: Yes, for sure. Thank you so much. Thank you for the great conversation. I think this has been a phenomenal partnership and thank you for some challenging questions, right? Like it's always, it's always interesting to see what, what else we can do. So thank you for the opportunity.
speaker-1: Yeah, for sure I do. So Lenovo has been around for close to 30 years. And for a large part of those years, we've been really good at selling PCs. Recently, last four or five years, we've really gotten more powerful at selling end-to-end solutions. Right. that is redefining how digital workplaces are done, how AI is actually implemented. We are users of AI, but in the back end, a lot of the technology, a lot of the solutions are being driven by companies like ours. And one of the goals of my role over here is to help the market, help the perception of Lenovo to move forward from being, hey, they're just the PC guys. Wow, they have all these solutions. And they're really my one-stop shop. So if I want to go from the hardware to the entire infrastructure or the entire solution to run my business, so I can essentially future-proof us. And there's been a lot of our investment and energy is going. So even in some of our recent ⁓ tech world announcements, we're talking much, more about solutions that are transforming businesses around the world and how Lenoho can actually bring that to you. That is my role here. It's to make that exciting story a lot more available to everybody who's our customer and our prospects as well.
speaker-0: Yeah, I'm hearing more more again, obviously Lenovo is an organization is doing very well, but the shift where people kind of see Lenovo is more of a holistic technology partner, right? Like a trusted technology partner, but one that can kind of help them in this whole AI era, which we're going to dive into more about what that means. I think that's the new background you got is actually from Tech World, right? And that's what that resembles is the AI era.
speaker-1: Absolutely. this is a tech world background. I hope it's coming through. It's our logo. It's about how we want to help usher in or empower further the AI era. And Lenovo is actually that portal. It is that one technology company that can do the hardware, the software, the solutions, the services to empower AI in your organizing team. And that is how we are trying to work with a lot of our customers, a lot of our partners to bring AI in a more meaningful, scalable way for our customers. So it's quick. We want to all get to the ROI of AI and we'll have to get there.
speaker-0: which I think is actually very topical. want to dive into that too in a minute around that piece because I think there's a lot of discussion around how do I actually get impact? And there's a lot of talk about organizations struggling, right, in that regard. coming back to, I mean, your role, mean, global campaigns, strategic programs, you cover obviously a lot of ground. When you think about, you know, identifying what's the right use case or what do you prioritize, right, with bringing AI into your workflows, your team's workflows. How do you go about doing that? Because you've got a pretty expansive landscape that you're kind of, you know, you're accountable for.
speaker-1: So as a part of my role, I do all the global campaigns with some of our big partnerships with FIFA and F1, as well as some strategic motions that we have, ABM, research. And we always have to look at AI solutions that are not only quick to deploy in one of these things, but are also result-oriented. So we have to get from, AI is great. It can do these things. But let's see how. Let's test it out. And the other things that I'm definitely trying to actively do with the team is, if you've got an idea that AI can implement, how do we make sure we try it quick and easy so that we can see the results of how it works? Because in a lot of cases, the show real to the actual reality vastly different. And that is something that even we at Lenovo try to do for our solutions, right? Push it forward. Make sure it's real. And that is where we're seeing a lot of use cases. One of the ones that I've ⁓ I really like to talk about is we talk about AI's ability to do ads quickly. So we picked a task that was pretty easy. said, let's give AI an ad and let us ask AI to make adaptations of it. Like super simple size adapts, one part of what we do, we said, OK, here's the ad. Here are the mean five sizes. So here are the mean five sizes, but.
speaker-0: Yep.
speaker-1: In most cases, the letters were outside of the required size. were stuff all over the place. And like a human designer would not have done that. So it is pretty awful. You still need to have a human in the middle. But you definitely need to keep trying out and testing some of these things. And then it is a decision about how much effort do we want to do to teach this AI so that it is useful in the long run. We'll say, it a quick test. see if it's working, cut your losses and move on. And I think that is an important use case here.
speaker-0: Yeah, it's actually, interesting because, you know, one of my, uh, one of my favorite quotes is, uh, you know, our problem well stated is a problem half solved, right? And that's one of the ways that we look at this too of there's so many things that we can do, but what are the things that actually add value? Right. And it's, it a problem we're solving? Like, what is the quantifiable, you know, impact of that? Like obviously the ad variations. has a quantifiable impact. There's a direct cost to what you would have paid a designer or an agency partner to do. And so it's like, OK, well, how much do we do that? What's the volume of that? And then you can start to quantify, OK, is this a worthwhile problem to solve?
speaker-1: And take the risk, right? I mean, we want to be able to empower our teams to take the risk. Because it's that leap between, hey, look at this cool video that AI can do. How do make it real for my organization, for me? How do I add value to it? And then see if it's actually viable or not. We would have all thought making ad adapts, size adapts, right? It's pretty easy. Get a designer to do it. It's easy, when you lob it over to AI to do, you actually have to train the model to say, OK, think inside this. your spot. And then you've got to decide, is this an effort or an exercise that I really want to do to train the model and really see what the long-term impact of it is. But that's real use case.
speaker-0: Yeah. Exactly. And that's it. It's like, if you know the workflows really well, you can start to pick off these individual use cases that over time accumulate to a lot of work, right? Or a lot of costs. Outside of the adaptations, there other pilots that you're either running or maybe thinking about running as an experiment?
speaker-1: Absolutely. So we have actually taken the challenge within Lenovo to embrace AI and show more of the measurable outcomes of using AI. So as a marketing organization, we're trying to implement it in a lot of different things. So the AdAdapter is just a quick example where we're definitely using them in research. So we are trying to figure out how AI can help us either scale or draw parallels. from the data a lot faster. That is ⁓ using AI to both test what we feel about our customers using AI to act as a customer and see how the customer would respond so that we can help train our sellers, our marketers to do better. That's one aspect. The other aspect is how we can start to do more personalization at scale. We know that our customers like to hear a solution that's baked for them. And growing customers, actually, people like to have the answer to their questions. with AI, you are able to scale up and say, hey, I can answer your specific question. Let me show you how it is unique. Then there is another exciting bit. And then the third one that I actually like is about being able to tell stories a lot faster. In a video world that we are in, there is a lot more video consumption. So if we are able to distill a 20 page white paper into a minute and a half of video that has a compelling outcome-driven story. But we've got a lot of insights to share, right? So we use AI to really operationalize this and have a lot more content that resonates a lot quicker, a lot faster. In fact, the most recent AI video-generated gen AI use case, we've actually been able to do that at almost 25 % of the cost. So we know people are consuming video. We're now able to create that at 25 % of the cost and really use it to scale how we tell more meaningful stories to our audiences. It's a lot of use cases.
speaker-0: this powerful, I mean, that's a 75 % reduction in cost, right? On the video production. And are those like the more AI avatar styles or are they just like illustrative? they?
speaker-1: They are more the AI avatar styles that we're using right now. It's very interesting because we could definitely do some of these videos with the animation styles and stuff like that. But we took a decision to say, hey, let's make a generative AI person kind of tell the story. It showcases a person in action. And we're trying to balance this out with real world story. So we take a real world example from our customer, and then we try to showcase what it would mean for a different customer in any different industry. So really bringing together personalization, storytelling, and to be able to do this at scale is ⁓ a phenomenal example of how we're trying to bring AI. ⁓
speaker-0: I love that. think that I think avatars have come a long way, you know, like when they were, as people say, it's sort of that uncanny valley of like, it's not quite there in terms of realism. And I think that in some extent, maybe where you have a little bit more way to go. But in our experience with the AI avatars, I mean, they speak 180 languages. You can create custom, like, you you can create your own, your own personas. And I know that for example, at Qualcomm, one of their use cases is actually taking a video that's produced in English and actually dubbing over with the, obviously the mouth movements, but taking the actor itself and actress themselves. So I think you're right. think video, there's a ton of opportunity there. And that's, something that is not just more cost efficient, but something that is a new capability to be able to do that in however many languages in real time.
speaker-1: I think the nuance here is we like to consume content that is personalized to us. So as we have this great tool that adds value and customization and personalization, that's when we use it to tell compelling stories. There is still an element of video that needs to be produced at scale and all of that. And that is a part that is working. But for more impactful, more personalized storytelling, This is a great value addition. And I resonate with it too. I watch a video, like, I wonder what would happen if I was in this scenario. Now you can find out. The tools or the capabilities to find out. Make this for me. It's phenomenal.
speaker-0: Yeah. personalization is the ultimate pursuit that is not a new pursuit, but one that we couldn't really have done well before. Like it's not possible, like it's cost prohibitive to do that. But with AI, like, you know, as you and I have talked about before is you can create these one-of-one engines, right? So we're even pushing that into content as you think about white papers and eBooks. How do you take a single asset and make it a one-of-one where If know, show up global mandate, ⁓ campaign strategic programs, you download it. How do you get a different asset as opposed to somebody is, you know, myself ⁓ on strategy, right? And within a marketing services organization. So that I think is really the unlock. And I think you're right. And that's multiple modalities, right? That's not just written in this video, as you said. So.
speaker-1: Yeah, in fact, because of this, and we're seeing this with our solutions buyers as well, right? They want to be able to tell their story to their stakeholders. What you are now doing is empowering your customers, empowering your storytellers, right? To be able to take your content and then add to that in their own personalized way. We all have great stories that we like to tell and now I'm allowing you to use content
speaker-0: Yes.
speaker-1: to tell that story within your organization too, because of the power of personalizing this. So it's a lot more personal. I look at the impact of research to me, to my company, to my buying habits. And with the power of AI now, you are able to customize it like never before. So it is a use case that is now vastly exploding. It is something that we are using. And I think the potential is immense because the value that it is giving to the buyer, to the buying group, phenomenal. I can tell my own story.
speaker-0: Yeah, from your perspective and depending on where you sit on the buying committee, you could tell the right story to the CFO, the COO, the chief legal officer, right? It's a bit of a different version of that story. Do you find as like a technology organization that's really sort of at the frontier of this, do you feel added pressure to be able to bring in these, these like, you know, as people say, dog food or drink your own champagne? Do you feel that pressure?
speaker-1: So yes, we have embraced the pressure and we've kind of challenged ourself. Because even as users, we now consume a lot of AI content. We use AI to make a lot of decisions, validate decisions, inquire and understand. We also try to use that to then empower how we turn up in the marketplace. We're trying to take that more as a challenge of how do I drive efficiency scale improvements by using some of the AI. And even if you look across your personal habits, a lot of our discovery journeys are now starting with AI. Hey, what do you think about this? How do I do this? That's right. One of the five questions I should ask.
speaker-0: Yes. Yeah.
speaker-1: And if we're doing that in our own personal capacities, that is soon going to expand into our organizational decision-making. So it is an interesting challenge as buying groups become more complex, as the discovery of what your solutions could be become more AI-driven, as a marketer, our challenge is to now turn up where our customers are looking for information. So I was just rounding it off. We have to turn up where people are searching for things. So if people are searching for things using AI, how do I now convince not only the customer, the human, that, this is a good prospect. Let's start exploring. But I now have to additionally convince the AI to view my content in the same way. So my customers have increased. My prospects have increased. And the way that I have to utilize my content, is an interesting, a phenomenal challenge. that is influencing buying decisions today more than ever before. And you've got to get ahead of it.
speaker-0: It's completely transforming. know we're going to want to dig into that one ⁓ shortly because you're right. It's like AI is becoming sort of the intermediaries between a lot of our clients and the decision makers right across the mind committee. ⁓ To add some sort of data to this, Watchtower is we've got an AI powered research platform. so at this point, our B2B marketing audience is around 550,000 B2B marketers specifically around the world. And so we looked at what were the conversations related to initiatives that they're going to scale in 2026. And so the top five were AI research assistance, which you mentioned research was one of your workflows, multi-agent workflows, AI search. AI campaign orchestration and agentic campaign orchestration. a lot of automation, obviously you mentioned research, ⁓ search, obviously that's transforming with GEO. How does that kind of list compare to maybe what you're prioritizing within your team? I mean, you mentioned research already, so that's kind of one for one. What are your thoughts on that?
speaker-1: So we're trying a lot of different things. We're definitely trying the research. We're trying the campaign orchestration. We're trying the personalization. But I think the challenge that we are kind of getting over is there's a lot of different types of AI that you can deploy and implement. There's a lot of different verticals where some have gotten really good, some are getting better. And then you as an organization have to make a choice about, do we want to put all of our fingers in the honey jar? or just 100,000 at a time. So that is a balance that we're trying to do. And I believe the role of the human in all of this is still important to the human in the middle. If nobody looks at the outcomes of some of our research, we're just saying AI said it and it's a we want to be sure that there is a human in the middle. It is looking at some of the aspects. And that becomes a scale question. Can I hand you everything with AI? Sure. Probably not today, right? So, we still have some time to get there.
speaker-0: Yeah. It's, it's your right, because it's ultimately as humans were accountable for the decisions and the outputs. so, you know, you can't blame Bob the agent for messing that up. Right. It's, it's, it's on us, right. It's on us. Yeah. You know, it's the, things too, that I, when I see this list, it's, shows me that I think, as you had mentioned, A lot of us are thinking about how do we create this sort of 10 X impact of how do you create like workflows, like end end workflows that can actually do these outputs. Right. And there's sort of the different two modes of thinking. There's like the 10 % thinking, which is like, how do I get 10 % better versus 10 X, which is how do I completely transform these workflows? And I think that's kind of where a lot of our clients are kind of like their head is at how do we transform workflows, but they are also handcuffed to some degree. with obviously, you know, the pace that internally they can move. you know, we're in, you know, compliance and privacy obviously is really important data protection, particularly for global enterprises. ⁓ but there is that, that tension that, that I do see, which is I want to move really, really fast, but internally we're not built to move that fast. And that's, that's a struggle for these global. mean, I'm preaching to the choir here, but You know, you live it every day in some respects.
speaker-1: It's an interesting challenge, right? As organizations become larger, there is a much bigger risk factor involved. So I recall we wanted to try an AI bot or an AI agent to do something. We thought, hey, this is quite easy. Let's pick it up. Let's see what happens. And let's deploy it. And then we bounced it over to procurement and said, nice, simple project. Let's get this out the door.
speaker-0: Yep.
speaker-1: actually looking at it in a lot more scrutiny because the impact of deploying something like that for a larger organization has wide-ranging reach. So they have to make sure that a lot of their checks and balances are in place. oftentimes, you find the early innovators who take a technology and drive 10x improvement. Like, let's do a 10 factors. And then you have parts of the organization whose job is to ensure continuity, security, safety. And then you've got to figure out how do we do, how do we kind of work together in this? Because you can move fast and break things. And sometimes it's OK. It's OK to break things and fail and figure it out and always take that risk. But some of these existing systems are in place to make sure that we don't break the whole system.
speaker-0: So great, the important things.
speaker-1: Don't create the important things. So I think that balance is important. being in marketing, being in that side of the organization where people say, hey, can we try this out? Let's see what happens. I love to push boundaries and say, let's do it. Let's give it a shot. And then we need the functions to look at it in larger pictures. This is great, but the risk is too big. Yeah. Because you have to see the larger picture. And I think as organizations implement or adopt AI more, you will start to see the use case for pre-existing solutions. Yeah. Right? So if somebody has used it once and you have seen ROI from it, can I buy that? Is it repeatable? Is it scalable? And then it kind of circles back to your initial point. it's repeatable and scalable, it's probably not getting you 10x. It's getting to 10%. And you've got to figure out where you are or what project it is that you are working on, where you want to go. All the way, figure out incremental needs and bring the entire organization along.
speaker-0: Yeah. And, know, you got to experiment, right? And, and, ⁓ you know, my, ⁓ my second favorite quote is on, on experiments, which is, you know, if you know that, if you know it's going to work, it's not an experiment. So you have to be able to take a real risk. So I, I, I wanted to ask you, how do you build that sort of risk appetite internally or, or, or navigate some of these conversations and just your thoughts on not just. fail fast and break stuff, but failing smart. know that you've talked about before about everything doesn't have to be successful. Let's be honest and transparent on this is an experiment. We don't know how it's going to go. What's your perspective on that? And then how do you have these conversations internally?
speaker-1: So I think as a goal, I've kind of taken it up to try things knowing well within the risk appetite that the organization is in. Right. Because if something works well, like you rightly said, it's probably not an expert. Right. And if it works well, you've got to scale it. But if it doesn't work, your opportunity to learn, to grow, to improve is far more. So you're within the team here. Let's try something risky. Let's break the barriers to what everybody thinks is normal and let's try something different. And in fact, there are a couple of examples that we've implemented AI. It's not worked. We've recognized it's not worked. And now we're in a position to say, okay, if you want to make it work for this, this is the kind of effort that is required to put into it. And this is the ultimate long-term vision of So now you've got this interesting point where you selected, we know the capabilities of the system. So either we have to invest to upscale, up train the people, the system, the AI, the data, and therefore here's the outcome. Or now we can start to look at different alternatives, maybe different solutions, different AI, in-house, app-soc. But we wouldn't have got here if we didn't try. We were just studying the, oh, hey, here's the potential. the unaddressed potential is always much larger than if you tried out. And the conference is trying out some of these things.
speaker-0: Yeah, and you have to try. I I was just chatting with a global MSP called Adastra. And I'm hosting one of their fireside chats later this week. And they are a global MSP. But they had one of their clients, which was the Mark Anthony group, who makes food and beverage. They're a global food and beverage company. And one of their heads of AI was talking about, look, we actually had to set up a real incubator to test a lot of these things. We had to sandbox it because we had to set the expectation that a lot of these things are not going to work. so it was also like a mindset shift that the organization needed to have in order to really embrace it because you have to embrace, you know, the way they put it is like, you know, not necessarily fail fast, but learn fast. Right. It's like, it's a failure, but you're to learn something. and I've seen organizations, ⁓ like Fannie Mae is an example that actually this was sort of, I'm going to go back. This is almost six years ago where they introduced this concept of, they called it Friday failures, which was, ⁓ I know it was brave, right? You had. leaders that would gather with their teams on Fridays and share the things that didn't work and what they learned from it. And this is all in the context of transformation and modernization and so forth. And so I think that we're going to have to learn a lot. We're going to to stub our toe. But you have to have the risk appetite to at least stub your toe and it's okay. You're going to learn.
speaker-1: And this is a great example. So in the early days of using AI for research, we asked our AI model, here's all the research that we've done on our buying group, our personas, consumer. And it did. And then we started asking the model, OK, tell me, what do you think the conversation groups will be for these people?
speaker-0: Yeah.
speaker-1: And in lot of cases, the results were pretty generic. Whereas the data showed that people would actually like to get a little bit more specific. And that's an interesting point where we had to look at this and say, I think it understands the essence of the research. It understands how to dissect the data and give us different variations of it. But if I had to use the data, the model as it is right now,
speaker-0: Interesting.
speaker-1: And if I were to throw a question saying, knowing what you know, can you tell me would this pitch work? we would have to take it with a grain of salt. Right? course. And that is an interesting exercise because for us, we had to train the model. We had to invest the time. We had to put a lot of the research in it to figure out that actually we've got something out of it that is not what we intended. Super useful. Yes. It did not drive the original purpose, but we were still able to experiment with it enough to figure out, OK, now we've got this different tool that we can use to
speaker-0: this.
speaker-1: to drive more meaningful conversations. And what we've been able to do with that kind of output is whenever we have new offerings and new products come through, we always take it back to the data and say, we know our segment. We understand this offering. Where do think we should start the conversation? And it is able to give us from the data saying, why don't we consider these three or four routes now? Because it's obviously learned and it's gotten better at it. But that is a time investment with AI that that I think everybody has to be ready to make. It's the investment in the future outcomes.
speaker-0: Yeah. I think it's a great example because you know that there's potential there. You know that if you continue to train the model, you're going to get more value. So you have conviction there. And I think the other thing too, is you think about it. If it's better than you, the team sort of like flipping a coin, if it's still directional, there's some directional guidance there that is valid. It's still valuable. It's just, do you get it even more specific and your use cases and all that comes with training and time. But I think if people expect, and this goes back to, think your ROI point earlier is like, what's the ROI of AI? I think that the, the revenue side is a, is a lagging indicator. If you do things right and you've got the conviction and you, you know, you, you, you're, you're working on your pilots, you're iterating and you're proving them out. It will generate revenue, but it. There's not, if you expect that to come so easily and immediately, you're never going to get past this first, first base, so to speak.
speaker-1: That's a great way to look at it. When you look at AI, revenue is one way of measuring. In marketing in an organization, we've picked up how we define the ROI of the AI that we're implementing. We're trying to look at it from a couple of different things. Revenue is the most important one, but we also want to look at it as efficiency. How are we able to drive efficiency using AI? Are we able to drive skill using AI? How are we able to drive personalization using AI? All the things that allow us to turn up more meaningfully in front of customers, and how are you measuring them? Now, we're figuring it out, right? We've not got the answers. But at least by being able to expand how our organization looks at the benefits of AI implementation, we are establishing parameters that we can improve. So we have some projects that purely drive efficiency, ultimately leading to revenue. We have some projects that drive purely personalization, ultimately leading to revenue. But we are able to experiment on more parameters than just revenue alone, which allows us to expand on the types of experiments, expand on the types of use cases. Because if somebody turned around and asked me how much revenue did that one AI-generated video generate,
speaker-0: But I would say you go back to any sort of marketing activity or even sales activity and say, well, hey, what was the direct response from that? And in many cases, we can't quantify it. It is a correlation, not necessarily direct causation. even if it's the last, you use a last touch attribution. this is the last thing. Well, events was the last thing that they went to. Does that mean that events? is what's driving the revenue. Not necessarily. It's all this stuff that we did leading up to it. Right. So attribution I think is such a, is such a tricky one. Obviously something that is very topical in the news that's always coming up, which is AI's impact on headcount. And so ⁓ in the Watchtower study, like I said, over half a million B2B marketers specifically, when we asked them how do they expect their headcount to change this year in 2026, a lot of it was, ⁓ I'll read these out, she said, reduce roles. The AI was 17%, centralized also AI role 17%, freezing new hiring 15%. So maybe speaking on behalf of yourself and not necessarily in Lenovo, how do you think AI impacts team design? Like, what is your perspective on that one? Cause that's such a tricky one and lots of news on related to this coming up, obviously. What's your perspective?
speaker-1: I think this is an interesting one, right? AI as a tool has now allowed us to generate a lot more value within the same amount of resources. So even you as a user, you now have access to a lot of tools, of course, within your organization guidance that you can use to generate a lot more value. And now you have this opportunity where you can become a value generator.
speaker-0: Got any magris? Yes.
speaker-1: And if you are on the side of the organization and you're pitching up a role that allows you to use AI to generate value, you have a far more valuable tool set at your disposal. Your value as an employee within an organization exponential, thanks to AI. The ones where there is a little bit more risk is if you're looking at the opportunity in front of you and you're saying, want to have happier
speaker-0: Yeah, so you
speaker-1: to continue to take orders to execute, to execute faster. I think that is the side that I would encourage to embrace AI to drive efficiency, drive scale. Because ultimately, AI is a tool that is going to help us drive efficiency and scale. And that is how users, employees need to use it. yeah. Overall, it impact?
speaker-0: Yes.
speaker-1: the shape of the organization, whether it back the usage or the organization structure, I think it will drive a lot more value to the people who are able to embrace it and use it. And it is not that you need to become a coder overnight and figure out stuff. I'm not either, but I'm a user. And I think with the right mentality, you can get ahead of it. So I think organizations will see the impact of AI and will take the benefit of the value that it drives. You just gotta decide on which side of that value coin as a user you want to
speaker-0: I like that perspective and I believe that because I think that, you know, rather than sort of victimizing oneself, you can say, this is, this can absolutely empower me. And so when I think about, and I mentioned this before, but when I think about previous technologies and I know AI is very unique in the acceleration, but you go back to the seventies where they launched the ATM. right, automated teller machine. And, you know, the, the prediction or the forecast was that we are not going to need tellers in a matter of years. And so when they did the study to look at what, how did ATMs impact tellers employed and over, was about, ⁓ I think 20 or 30 year span, the compound annual growth rate was 43%. So we actually grew the number of tellers by as a society, as a workforce by 43%. There was an initial ⁓ state where it contracted and then it grew. And so when you, when you look at the data, the thing that I see is that what people as tellers weren't automated out of a job, they were automated out of bank telling, which is like giving money, receiving money. Like you go to a teller.
speaker-1: Yep.
speaker-0: You can go at, like you're talking about your mortgages and you're like more complex things, right? Like to your point, they're elevated. So I think we will see a net positive, but to your point, it's a mindset. Are you going to look at yourself as a victim or are you going to look at yourself as I'm empowered to create value?
speaker-1: Absolutely. I think that's a great example, right? Like old history, you've always seen a lot of these new technologies and there's a lot of fear when you have new technologies come to market. The credit card, ⁓ my God, we're not going to need cash ever again. Like cars, what are all these horses going to do? Right? Like we've had a lot of these technology, technologies that have really changed the way humans get up. Get out, do work, spend money, spend time. And in lot of cases, fear around what is going to happen to the old way of life quickly transforms into how am I going to use this to add value to the old way of life. And we have a lot of these tools now that are unlocking exponential value potentially. So we're able to do so much more in
speaker-0: Yes.
speaker-1: almost the same amount of time with almost the same amount of resources. And it is so much more meaningful to our customers, to our prospects, to our company, right? It is about embracing that and being able to derive value from that. Everything on the right side of that choice. Because at the same time, you have detractors. Like with every piece of technology, you say, hey, this is not going to work out. Which is fine. Everybody has their place. You need the cynics to kind of challenge you to do more work you have.
speaker-0: Yeah.
speaker-1: And I think that balance is important. And it's also different teams in the organization adopt AI at different bases. Absolutely fine, right? Which is a learning curve that the entire organization can go through. And that is essentially what a lot of us at Lenovo are doing too. Like different organizations have picked up AI for different sizes and scales. And I think marketing has been one that has picked it up and said, we're going to do it all.
speaker-0: Yes.
speaker-1: We're going to try it all. Sure, not all of it will work out. We're going to take those learnings and we're going to allow the teams that work with us to be able to embrace those learnings, figure out what to do, and then move forward from there. But you have to have one of those groups that is ahead of the curve and really pushing the limits to what AI can do for you. And marketing has taken up that challenge.
speaker-0: Well, kudos to you and the team on that because I think that there's a lot of, I would say that that's more of an exception than the standard right now. I think a lot of people are a little bit hesitant and it sounds like there's the right culture to say, let's experiment, let's try, truly experiment. Obviously we want this to be successful, but not everything can be. You your point on the value generation, it reminded me of there's a pretty big AI conference called Maycon that happens in the fall in Cleveland. And so we're partners in terms of, know, I'll do either a keynote or we've got sort of a session there. And one of the keynotes from the few months ago was one of the founders, Jeremiah O. Yang from Blitz Scale Ventures. And so he talked about how before it was how many employees you had and all that. The metric is now changing to sort of revenue per head. And so they looked at uh, SaaS companies, which I think was around the, like an average SaaS companies around 250,000 revenue per per per employee. said a Salesforce or somebody that obviously is like a dominant player in the space is closer to 400,000, 450,000. And do you know what that number is for SF startups, like San Francisco based startups? Like I didn't know either. Like he just used to tell 2 million per head. And so I think to your point on value generator, this is an example how if you're an AI native organization and it doesn't mean you don't have, you know, you could still be human first, right? But if you are AI native or AI forward, that's the kind of value that each one of us can have in the, like that's pretty, that's pretty incredible. So just the numbers on it.
speaker-1: Yeah. It kind of transcends across, right? So you now have a workforce that can add value to your work. And it is, and you have to decide how you want to use it. It can be doing a lot of the repetitive work. It can be questioning all the thought work that you're doing, asking the tough questions, right? Challenge me on these three things.
speaker-0: Yes,
speaker-1: And how can I make this better? And you have another capability using some of the existing AI models to pretty much try out and challenge yourself to do better. And you can pick, right?
speaker-0: Yeah. And I was just going to say, speaking of challenges, in the study, we asked these B2B marketers, what type of, like, what are your biggest organizational blockers to adopting AI and embracing AI? And I'll read you out some of the top responses here. Limited AI skills, no training program, no AI governance, too many tools. and limited IT support. to me, what was interesting is that none of these were technology problems. They're all people problems. So I just curious, what's your reaction to that on that finding?
speaker-1: I think it's unique, right? As organizations are being born in the AI world, their challenges are unique too. A lot of these organizations are born AI ready, so they're able to ramp up from get-go to 100 people, 50 people pretty fast. That's when scale has a lot wider reach. And I think I'll you an example for our organization.
speaker-0: Yeah.
speaker-1: ⁓ a couple of these AI opportunities that we wanted to explore. So it was a pretty quick route to market. We said, let's try it out. And our biggest blocker for making it happen was procurement. Procurement wanted to look at it and say, wait, what exactly is the outcome of this? And how is it going to impact the rest of the business? So it was less about the technology behind it. was less doubt in the implementation. It was more about what else. I think as our organization is going up, as AI adoption use cases is going faster, I think you start to see more of these types of challenges come out. You have to experiment. We would have never known that the biggest challenge to implementing AI is not the model, it's not the infrastructure, it's our procurement guidelines. So even if we had all of those things checked off and procurement was saying, whoa, hold on a second. Right. So it is pretty interesting. In fact, this reminds me of ⁓ GDPR. I think with GDPR, was a thing. ⁓ Everybody was like, OK, this is great. It's going to change the way we do business. And there was this guy who were always saying, where is all of our data? So I think we're now getting into that wave where everyone's saying, let's get AI. And then people are asking more meaningful questions of AI for what? And how do you categorize this? Is this internal? Is it confidential? Is the data, what is it learning from? And I think a lot of that is how real challenges are going to be dealt with as we're kind of going through this AI wave, as we're in the middle of this AI wave.
speaker-0: Yeah. It's, you raised such a good point because yeah, I mean, just, just on data, like data assets are, are becoming obviously even more valuable. Right. And so I think as, as the models become somewhat commoditized, where like the leaderboard's changing, you know, month to month, a quarter of a month, really months a month at this point, it is like, well, you know, how do you, how do you protect that data too? And I do, I do, I don't know what your thoughts are on this, but I feel things are moving back on prem in a way of like, are, they're pumping a ton of very confidential stuff into some of these baseline models that you just no longer have ownership of a lot of that data. just don't see how there's going to be so much information sprawl and data sprawl that companies are going to want to protect that and bring that back in. mean, obviously with your infrastructure group would be able to see this, but Is that also like, that, is that what you're seeing too? Like there's a trend to move back to on-prem in a way or.
speaker-1: So I think the answer to this is kind of twofold. One is as models get smarter, we want to be able to train them better. And a lot of these models rely on being able to understand a data source that is unique to your business. And as we get into more data, more AI models, businesses will start to realize
speaker-0: Yeah. Yes.
speaker-1: If they have a store of data that is unique to them, that is their business IP, they want to be able to protect that as opposed to give it out to the cloud. They are everybody tree. So I think the investment on hybrid AI data, databases, AI models that work both in the cloud, where you can store a lot, as well as on-prem.
speaker-0: Yes.
speaker-1: is the next wave of AI average for marketing. We want to be able to make quick AI decisions based on what we have, what is available, what is accessible to us, and protect all of our customer information, keep it close to us in well-protected servers, as well as take advantage of the larger models. So we're always moving between these two types of data deployments, these two types of models, right? And essentially, that is how That is how we see more organizations embracing AI. Think of it this way, if you've ever been to a doctor or hospital recently, a lot of the ability to take a lot of that data and run it through an AI model and come up with scenarios, you don't want that data to take a round trip to the cloud and come back. it's not only expensive, it's bad for the environment. You can do a lot of that compute.
speaker-0: because
speaker-1: within the four walls of your hospital. And that is the technology that we as a company are now developing and implementing in a lot of hospitals as well. Keeping the data close to us, keeping the data close to the users, keeping the data close to the people who need it. it is an interesting time because we're moving away from send everything to the cloud. Let's figure out what is the best use case for this. It needs to be on a PC.
speaker-0: Yes. Yeah. Well, that's it, right? Now you've got on-device AI capabilities directly. And so I do think that as the models advance, they get more efficient and they will be able to run directly on the device. Like today's, you know, just, use JetGBT as an example, cause it's the most well-known, right? But, or, or Claude, the, the frontier model today, give it a six months or a year. You could probably run that locally. on the device because it's going to get more more efficient. I think paired with our energy crisis, we don't have enough energy to be able to, if you wanted to send everything to the cloud and put this in these massive data centers, you're just not going to be able to do that. I think the device and on-prem is actually making this comeback in a way, I feel, I think it was really going to, to your point on hybrid. You're going to need both. it's not like before, which it was just push everything to the cloud, which I felt that that was the direction.
speaker-1: Yeah, I think like we're going through this wave now even even our our devices have a lot of AI capabilities on it, right? Like it's not just organizations want to keep data close to them, but users too. Do creators now like to keep a lot of their content close? Yeah, they want to use their on Prem on their device. They they want to be able to do that and I think as as our portfolio of.
speaker-0: Yeah, that's true.
speaker-1: connections into the AI sphere expect, right? You want to be able to decide as a user, as an organization, what do I need? Do I need the speed? Do I need the safety? I need the security? Do I need the scale? How do I plan for both my data, my organization's data to make sure it's giving me the right solution? And as an organization, we want to be able to give you the option
speaker-0: Yes.
speaker-1: It's not one size fits all. Everybody sent it to the cloud. It's not there. Let's talk about your use cases. A lot of our solutions engineers do exactly that. What are you going to do with this data? What's your implementation look like? Here are our solutions that help you do that. So it's keeping a lot of it close by. And these are not easy decisions. In fact, AI is now even involved in the decision making.
speaker-0: Yeah.
speaker-1: which is phenomenal, right? AI is part of the decision-making journey, and that is really spurring how we turn up as customers, too. How do we engage with our buying group in a meaningful way with meaningful content, with relevant content, and start to drive that conversation?
speaker-0: One of the points that caught my attention in the studies that were running is that marketers are clearly looking to produce content for AI. In fact, when we looked at how B2B buying is going to change in 2026 because of AI, I think the number one response was producing content for AI. Now, I know before you had mentioned something like A couple years ago, that number was 40 some odd percent or something like that. Tell me, tell us more on that because I find this really interesting and how we're going to, how are we, what are we going to do about that? How are we going to pivot, you know, given that AI is becoming the intermediary.
speaker-1: We did some research about two years ago. We looked at our buyers and we looked at how they are making decision juggies. ⁓ What is their decision points? And in fact, AI and generative AI or search-based AI was one of the fastest growing categories for discovery. Now, two years since that research has come, so I'm sure it's much more important now. But two years ago, that was on the uptake. And two years ago, when we had that challenge and you look at what we could do, the options for us to work with AI to empower it so that it has all the right content to provide to its users the right decisions were pretty limited. Today we're getting into a space where we can talk to AI more meaningfully. We can say, here's our point of view. You can take it from a lot of different places. You can optimize the way you want to. generate content for a person who's searching for stuff. But two years ago, we had no way of doing that. So I think this has two impacts for us. One, as AI is becoming more involved in our customer's decision process, it is turning up at the discovery stage. It is turning up at the evaluation stage. It is turning up as a co-buyer. Yes. What is the role as marketers that we can take up to influence everybody in the buying committee? which now also includes an AI agent. So how do we empower not only the humans in the middle, but how do we empower the AI bots? And this is where we use generative engine optimization. How do we optimize our content to become relevant to models like this? And I think as we grow up, as AI becomes more ubiquitous across everything that we do, I think that is the next frontier. Back when... SEO was a thing. What are people really searching for? That was a thing. How do I have my website, talk about Italian food, but everyone's hungry? Nobody searches for I am hungry. They search for what kind of food do I work in? We're now getting into that phase with AI too. How do we tell the AI model, listen, you might be hungry, but you won't come to us for the subs. That's the model.
speaker-0: Always the food metaphors, right? Always the food metaphors. ⁓ No, I love it. I actually, think that this is, I feel like we could do a whole episode on this because one, now we're seeing obviously a paid entry model with chat GPT with ads. And obviously there's a lot of debate on that. Then there's the organic path, as you had mentioned, which I think is so powerful because I think that people are relying on AI heavily. in B2B buying decisions. And AI is basically rewriting perception in a way because if AI has conviction that, let's say Lenovo makes the best servers to run local AI models, which in many cases it does, I think that the buying committee, that is going to influence the buying committee. I think there's more trust that they put on these outputs. And actually it might even be even more impactful for the line of business folks, like obviously as we target IT and line of business, where the line of business are increasingly being involved, right? As we know in these buying decisions, but maybe they don't have the technical acumen or insight and they might be leaning more heavily on these models. So it is, I think you're right. It is such a large frontier opportunity to be able to win.
speaker-1: Yeah. think, I think the role of AI and the buying group is, is a friend, is an advisor, is a consultant, right? And we all love our consultants, but we have to take the information they give us and we have to distill it down for ourselves. But even if we disagree with them, we always consider them. And I think that is an important, important stakeholder for us to deal with now.
speaker-0: Yes. Yes.
speaker-1: It's just like back in the day when we had to turn up on page number one of search to be full set. There's nobody's going to page 256 to say, let me click on link four on this page and see what the answer is. That's right. So I think that is the kind of world that we live in now where you have to make sure we turn up in a meaningful way. But the challenge is not the same as with search because people are not as simple with their products anymore. People are
speaker-0: Yeah. Exceria
speaker-1: people are getting a lot more complex and what they want to achieve out of this answer. So they are asking a lot more meaningful questions. And that is where our role of marketers is important. We don't want to say, hey, if the guy is searching for these three keywords, turn up this ad all the time. We've got to be meaningful in the context that the buyer is actually using to find the solution. And that is a challenge. And that is a challenge that we've got to tackle. Honestly, we have to figure it out. We are experimenting. This is one of the things that I'm happy to learn fast and improve it. But it is how we use AI to benefit what our buyers are doing.
speaker-0: can share a little bit about some of the insights that we've had so far on this, which is, first of all, before we built our GEO platform, we did look at Profound and some of the others. And what we found with them is while they're great platforms in general for GEO, they are really built for consumer. And so when we would look, our CTO and our team would look at the underlying prompts that's being generated. It's very consumer and these are not realistic prompts that people send. So then when you start to see some of the data back that says X percent on Reddit or expert, like you see the source breakdown and that will change drastically depending on what the prompt is and the specificity of the prompts. So one of the ways that we got around this was ⁓ we actually, when we built our GEO platform, we interviewed IT decision makers and business decision makers to say, hey, if this was the scenario, what would be the prompt that you would use? And so we consider that, OK, that's your ground truth data asset, right? So it's like, OK, you've got that. Now, what you need to do then is generate prompts that are, if you're an awareness or consideration or purchase, like depending on what stage, I mean, it's not always obviously linear, but depending on buying intent, how does that change? And you need to run thousands of prompts. So, so we would put in sort of five, 600 per model. Say you run that across the five major models, somewhere around 3000. ⁓ but the, the platforms that we saw, you would run 12 or 20. like, this is too small. So, so there's a couple of things that we found. there's that we abbreviated into this acronym called beta B E T A. And so B stands for best for, which is you have to produce when you think about content strategies, who is it best for? You have to make that declaration, right? It's like, we are the best model for da, da, da, da, da, da, da. The, the E is evidence. So need proof points, which what we find is that if the proof sounds like bullshit, The model knows it's bullshit. so is it, if it's like, this is 500%, you know, 20 times you actually seen the reasoning that the model says, this doesn't sound believable. So there's the believable evidence. should say the, the, the T stands for trade off. is the toughest part that we are having conversations with our clients, which is T is about who are you not best for and try and take that back to the product team to go. Who are we going to state that we are not the best use case for, and they should look for an alternative. And they're like, no, no, no, no. We're good for everybody. And so if you don't declare that the models will come up with an answer, but the folks that are declaring who they're not best for actually are ranking higher. Uh, and then the A's answerability, which is format making it really stackable. like we, we call it, we call it beta because not only does it line up from a word. from that word standpoint, but it is going to change. It's going to be different pillars as the models reason differently and all that. Anyway, I'm going off on a rant. I think that GEO, you're right, I think is going to be the frontier win.
speaker-1: Yes, I think that's a good model. That's a good way for a conversation to start, whether it's with a human or with an AI. That's a great way to establish credibility. And I think the challenge for us as marketers is it is extremely difficult to tell somebody, I don't think that's the right customer for this.
speaker-0: Yeah, right.
speaker-1: because you can change the prompt in a couple of ways. And it was like, okay, now me is kind of the right customer. Right. So, so it is extremely challenging to say, I understand who this is really good for and he was wine and also who this may not be that good for. here's what, but our, our, our product conversations, our seller conversations would be like, this is great for everybody. ⁓
speaker-0: Yeah.
speaker-1: You've got to be able to take that insight and translate it into what are people really searching for? Because if I had to take it back to my food analogy, everybody who's hungry could eat Italian food.
speaker-0: X.
speaker-1: But not everybody who is hungry is eating Italian food. Right. Like you've your audience for it based on their needs and their personas and all of that. and people will always search for what's good food near me.
speaker-0: Yes, that's right.
speaker-1: The answer is not always Italian food, right? So I like Italian food, but still.
speaker-0: Yeah, me too. I'm a huge fan. ⁓ No, I think there's so much to get into on that. I'm not sure you and I will chat more on this on the on GEO side. So, Shoaib, I know you're a busy guy. So I'll ask you one last question, which is, you you look ahead to the next 12 months, which sounds like a sounds like a decade, right? In this whole AI era. But let's say 12 months out. What are you most excited about? What are you most cautious of as you look forward?
speaker-1: think for me, 12 months is something I'm a little cautious of because the modeling can change so much in 12 months, right? So what we're definitely trying to do, what I'm trying to do is look at the newer term, as much focus as the long term. And a lot of times, you have to pick one, which is the toughest thing that we go through now. But we have to be able to establish wins for today and experiments for today that will impact the long term. And I think that is where we are really trying to dabble in what we do, how we adapt technology. But I think from a long term, it is an extremely exciting time. We now have the opportunity to train a customer, to talk to a customer, to understand a customer that, for the first time, can talk back to you and actually tell you what they're thinking. I will tell you why this is not working for them. And now the challenge for us is, OK, now I know exactly what you need to hear. How do I tell you and the people that you want to convey this to in a meaningful way? So the next 12 months, I think that journey of AI as my customer to AI as my channel is going to be an extremely interesting one for marketing, marketers across the world to up. How do I express my product solution challenge? to AI so that they can help educate further. And I think that is something that is going to really impact the way buying will change, definitely the next year.
speaker-0: show up great perspectives. And, and, and I love the genuine enthusiasm that I share too around. This is an exciting time to be in marketing. It's an exciting time to be in knowledge work. know that there's a lot of fear and uncertainty right now, but I do think that to your points on B take that perspective of being a value generator, you know, rethinking some of the traditional ways of doing things and having the realistic perspective on Not everything is going to work and that's okay. Right. Let's go figure it out together. So, uh, on, on that note, show up. Thank you so much for joining us. Great conversation. uh, yeah, excited to have you on.
speaker-1: Yes, for sure. Thank you so much. Thank you for the great conversation. I think this has been a phenomenal partnership and thank you for some challenging questions, right? Like it's always, it's always interesting to see what, what else we can do. So thank you for the opportunity.
