S4 Capital plc (SFOR) Earnings Call Transcript & Summary

Okay. Well, it's 3:00 in the U.K. We will get started. So thank you, everybody, for joining today. Very much appreciate your time. And welcome to the S4 Capital Artificial Intelligence briefing session. AI has been a huge topic of conversation with our clients, but also obviously with you, our investors and the analysts that cover us. And that's not really surprising when you see a lot of the news flow, a lot of the consumer behavior that's been happening. We've seen reports like the recent one from Goldman Sachs that talks about AI, potentially adding 7% to global GDP over the next 10 years. So whilst there's a lot of hype around AI, there's also a lot of excitement, and we share that excitement at S4. So we wanted to take the chance today to give you an overview of how we think about AI, what we're currently doing with our clients and why we're excited and bullish about the opportunity that AI gives us in our industry. Today, Wes is going to share this presentation with you, it's essentially the deck that we've been using with clients. It's one we had a lot of traction with last week in a very busy Cannes Lions Festival. And we will be doing question-and-answer at the end of the session. If you have questions, please do put them into the chat box during the presentation, all feel free to e-mail me, [email protected] and then at the end, I'll collate them and be the quizmaster and ask them to Wes. We are recording this session, and we'll be posting it to the ESP Capital website after the fact. So with that, over to you, s.

Thank you, Scott. Hey, everyone. I'm Wesley ter Haar, Co-Founder MediaMonks and on the Board of S4. And to Scott's point, we have had a lot of conversation in Cannes about AI. I look myself into an apartment on the cost and really did top-to-top sessions with our clients, discuss our view on AI and really what it means for them. Before I dig into that, a quick introduction, I'm sure everybody is aware of who we are, but maybe a little bit of info about how we position ourselves because [Technical Difficulty] I think it's pretty key when we dig into this AI sort of wave First things first, we've been in business for about 9,000 people across the world, in deep subject and technology people. And we position a bit operate a single brand, which is MediaMonks operating brand. We run the business as Single P&L. I'll get to -- like that is important later on, but it just means there's flexibility in how we organize and operate around what our clients need. And then there's more of a mindset component to it, which we always go low to no ego. It really is our goal to be useful. I think it's reflected back in our portfolio. We are an innovator for the world's most innovative companies, and that doesn't mean it's innovation for the sake of being shiny. This is innovating how these companies turn up for their consumers, is innovating how they're able to organize internally. And if you look at the work that comes out of that, we just had Cannes, of course, the quality of that work is world class, which is a reflection of both our clients and the talent we have in our teams. So with that being said, what is our role and responsibility for our clients. I think if you look at the industry at large, it tends to really enjoy talking about the future. But they're doing that because it's easy, right? You don't have to solve the future. So our role and responsibility is in the now. What's happening now in culture, in commerce and technology and what can we do to unlock a fair -- unfair advantage for our clients. Looking at now, there really is only one thing happening now. [Presentation]

This very much reminds me of Cannes, but I do think it was the Cannes of AI platitude. It was very top level -- there's a lot of hand waving and relatively large amounts of smaller mirrors. I think the feedback that we received on our sessions was that it was real, right? We're calling this a race to reality. So what I'll do now, it will take about 40 minutes. We'll run you through that story. We'll start with our point of view. The rabbit is dead, then we become clear in about 10 to 15 minutes from now. Then we're showcasing how we're unlocking AI for our clients right now, right? There's real value in diving in technology at an early stage and being fast in first. But that's important because there's also some higher stakes to it. This idea of becoming an AI-first growth and marketing organization, and we'll share our thoughts on what that means. Then where do we start, give a bit of input on how we are packaging the our clients, how we're collaborating with our clients. Then I have a few slides that are really about our positioning and why we think this is a good moment of disruption for us. And to Scott's point, will be seeing our Q&A after if you have any questions, just open them in the chart and will follow along. So part one, unfortunately, the rabbit is dead. AI has upended the economics of advertising, canary in a coal mine, our output is easily recognizable through the lens of generative AI. But if you believe that it's a pending the economics of advertising, it's upending the economics of pretty much everything. This is where I would say something positive for our industry broadly. There is a bit of worry there. I would say, with the amount of disruption that's going to hit every category our clients operate in, I think we actually have massive amounts of opportunity, right? With that level of disruption, they need creative technology verse partners to help them through that moment of change. But the interesting piece here is we have to be honest about what's actually happening. And if you look at technology right now, what it's already pretty good at is what I would call decoupling hours from output. We were in Cannes. So, we showed my favorite at of all-time surfers again as ad. To the left, a beautiful piece of work. And then to the right, we saw images that were generated, purely prompted, no additional training sets, just people that understand the technology, and I have a decent AI supporting craft. The big question, of course, here is, is it perfect? My answer would be no. Would you notice it as a consumer as you're calling pass it in the 1 to 2 seconds you tend to spend on these types of activations, big question mark, right? And as we dug into that with our clients, there is definitely a conversation in the room with some people with different opinions, but it's important to understand what is already possible. So decoupling hours from output. And an important note is every time that you look at generative AI is going to be the worst it will ever be. And I look at my sort of history and our history with this technology, going all the way back to 2017, very bullish on it very excited by the opportunities. I've talked about it for years and years and years. Even we were caught by surprised with the speed this went from toy to tool, this idea that there is an exponential element to it, which can be difficult to forecast. So the images we see here is mid-journey, probably the most used, most famous sort of image-generative AI tool. Same point, 1 year apart, right? So the speed of change from toy to tool is really impressive. And I also feel we're seeing some of the same mistakes being made again when it comes to film, right? There's also really quirky AI-generated videos going around the Will Smith eating spaghetti is a personal pave that had its viral moment. The walk eating walks is probably the citizen Cannes of AI-generated video. But when I see what our team is doing, we're not that far away from what I would call end-to-end high-end film production, this idea that we can get beautiful composition in short and lighting and staging from the machines already. Doesn't mean there isn't some AI awareness right facial consistency hands always. Some of the dancing is slightly weird, but there's beautiful work already in that process. And again, this has been done by high-class well-trained teams but in fractions of the time. The question we have to answer for ourselves is where are we not timing, right? If we just saw mid journey with fill Imagery a year apart, is this month 5? Is it 6? Is it going to take another 6 months, 12 months, 2 years? t's coming. And there's a bigger societal conversation to be had what it needs to live in a landscape or everything and everyone can be generated at the press of a key. But for marketers, this becomes a really interesting element sort of built into their thinking, especially if you think about a space where content is what we call end cap, what does it take to stand out. So decoupling hours from output, it's there. We're doing it for our clients. If you start thinking about next, we've used different terminology here, real-time brands, we sponsored brands. It's going to feel like the age of resistance. And this really is the original intent of digital advertising. Everybody that was in time last week has either pitched this or has been pitched this personalization at massive scale in real-time, highly targeted, very sticky, conversational. That's possible already to some extent by the technology. And to Scott's point, I think that's what's going to unlock a lot of growth for our clients and brands. It's just going to get better and better and better. So if we already have really provable ways to implement the technology now and there's so much upside to come, why are we seeing clients be quite careful. Let's be clear, we call this the day 0 of AI for our clients. There is, #1, a bit of a cynicism about what I would call hype cycles. We, as an organization, look a bit broader than just marketing and advertising. But in marketing and advertising, we sort of live and die by the hype cycles, right? Something becomes massively hyped, people invest in it and then there is the disappointment on the other side, mostly related to region ROI. We just went through one of those cycles around Web3s, around NFTs. So a lot of clients until very recently were very much in, is this another hype cycle. Is this NFTs again? And the answer is no. Do you want some NFTs, but no. I think this quote from Sam Edelman, Co-Founder and CEO of OpenAI, which is really the company why we are having sort of a close conversation about generative AI at the moment. I think this quote is key. AI is a rare example of an extremely high technology that almost everyone still underestimates the impact on it. I personally could not agree more. but still uptick, right? We're seeing some of the legal complexity, book, some of the clients' enthusiasm to get started. Part of that is fair, although I think we're in a moment in time, and we're seeing some really good movements. We just had Adobe confirmed that they're going to indemnify all images generated from Adobe Firefly for instance. So you have no legal risk on that. I think 2 to 3 weeks ago, Japan already mentioned and implemented that there will be no legal risk on top of training data, right, now copyright issues. So we're seeing decisions made by companies and by countries that will get us past this moment. I think there's also a fundamental misunderstanding of what's happening, right? It's not called machine copying. It's called machine learning for a reason. Synthesizes massive amounts of information and data, learn from that. And then as you pointed, it gives you its best statistical approximation of what things you're asking for. It's not copy pasting from other parts of the web or data and trading sets. So some legal gray area, we're helping a lot of clients unlock that and unlock it. The image set at right from me was a bit of an Aha moment. So we've been tracking along diligently and with great excitement. This happened end of last year, October, November. And it's a bit of an AI mean now. This idea of everything in the style of West Anderson. This is designing in the sale of West Anderson. When I saw this for me, it was a toy to tool moment, right? Understanding that there's readability, there's art direction, there's nuanced. This is a really nuanced take on that reform design and production perspective. When we saw that, we went into a bit of a huddle and overdrive on our side and sort of sped up some of our plans for the people that follow that, we had quite a clear message that we possibly had on LinkedIn, which is we are all in on AI. We see that as a fundamental shift in our business, and we're pledging to be fast and first. It's interesting when you start digging into what's already available, right? Because it has synthesized such massive amounts of input, it's not random, right? It's art directed. If you look at these images, again, is this perfect advertising? No. But if we are art directorial archetypes, right, the typical beauty ad, the typical fast food ad, the typical car ad. All of that information is in the machines, right? It is in what has been synthesized. In Cannes, this was blasting. So Cannes is the festival creativity, and we're big fans of the festival in general and, of course, a very creative organization. But there was a lot of platitude around human creativity will never be replaced. It's interesting to see how uncomfortable a lot of our industry is just to be almost no there is creatively already available. This isn't some ChatGPT, this is from another LLM, but this is a great example of how idea generation is actually part of the tool set that's already now available that our team is using continuously. Example here, IKEA, I think, is a fun one, right? So this was briefed with the worst reach you can give credit or credit theme, which is we would like a viral video for IKEA. And the output is actually really interesting. Going back to Cannes. I've been on enough of those juries to know that well executed, this could pick up an all right? There's not winning a grand pre, but it will pick up a bones well executed in the right category. And there's quite a bit of nuance here, right? The idea is group of people go into the escape room door closes, there's a voice over. There's a time element to it, right? There's an understanding will make something sort of a video that we're sharing. And then the idea is you go into the escape room and with this group of people, you have to assemble IKEA furniture to escape the room, which is interesting, right? It even has a bit of the humble tone that IKEA sort of has, some of the friction where people know it can be a bit clunky to put the furniture together, it's a pretty strong initial idea, right? So it's not just content. Creativity is part of this process as well. So what's really interesting elements already available. There will be blockers though. And this is where it becomes interesting also from a client conversation perspective. So some blockers, the idea of an AI winter, we've seen a massive amount of change in a relatively short amount of time, and AI winter means we're plateauing, right? This was a big, big push, and now we're going to have a few years where it's flat. Having been at a conference a few weeks ago, the people within the industry don't believe that's happening. They think this is the start of that post exponential curve. The other part is what I would call the neutralizing of products. This 100% is happening, right? So our access combined agency brands, consumers to GPT4 isn't to the sort of core powerful technology that's already available and with good news, right? It's guardrail, there's new freeing that happens to it. If somebody is super interested in. I'll share some of the research, which is called Red testing, where you can see some of the scenarios that are testing, we said being made clear that it's smart, that they're not openly providing access to everyone. We have enterprise clients that are careful. We talked a bit about this already. But at the enterprise level, it's about mitigating and managing risk. And there's definitely a fear of the legal component. There's also quite a bit of potential ethical complexity, right? I think every brand has to have a real point of view on some of the ethics in the space. We'll see some unforeseen blow-ups create backlog, which will calm down some people's sort of willingness to adopt. And I think something that hasn't been discussed enough, but will be really key moving forward is this idea of Nation states and political book. We're in Europe at the moment, Scott and myself, realistically, we do not have a big player in the AI space, hidden AI races. So we only stand to connect the downside that was at, right, the idea of potential global losses. So I'm expecting super strict and quite speedy regulation in the EU. I know the U.K. AI Act is one it's way. I think it's rumored to be quite strict. It's actually interesting to think about some of the original ideas. Bill Gates had in this space. He talked about this, I think, about a decade ago, the idea that if somebody loses their job, the company that built a technology that sort of meant our personnel on how the job gets taxed. So we'll see if that comes to bear. It would be interesting to see that be some of the tax ideas towards OpenAI, for instance. One more note that isn't here, but just the access to GPUs. NVIDIA, I suspect, is on everybody's radar, a key company in this space, being able to build enough of them quickly enough will actually be a potential blocker as well. And then we have something that I call the magic button problem. Exponential curves are difficult to forecast. The reality is you can sit down today, plan your road map for the next 6 months, next 9 months, next 12 months. But there's probably a company out there that's already trying to solve that problem. There are currently 17,000 funded AI start-ups in the U.S. alone. I'm sure if you get added every single day. They're trying to solve pretty much every problem that you can think of. So there's this balancing act in where you invest your time and resources because there might be a magic button on the way. So with all of that being said, we have actually seen this movie before. And this, of course, is Casper, clearly human genius losing to deep milestone in the AI industry. I think it's a good reflection of what we saw in Cannes. This initial moment when AI starts to encroach on a human endeavor and in human expertise, we instinctively really want to humans to be better at it, right? We really, really, really want the humans to win. I think it was sort of reflected very well by Maurice Ashley, who was a commentator on this match. I wish he could pull wrap it out of his hat. But unfortunately, the rabbit is dead. The good news is Casper won his next match against the Blue by working together with AI, by working together with machine, which for me really is the moment that we're in. What is happening for our clients? I think this was key to our conversation throughout the week. This is a massive disrupt or defend moment. Foundational change that's probably going to hit every category, lots of new tools and technologies, there will be massive amounts of new behaviors and expectations on the consumer side. This is your opportunity to disrupt. You can lead for others in your category by being first into the AI space. You also need an awareness that there is an element of defending. Think about creative culture. Over the last few years, we've seen that sort of lots of -- we have lots of examples where creators are starting their own brands. We've already seen that with sort of talent at the global fame level doing the same. And those brands or businesses are eating into more established global businesses. AI to an extent, superpowers that behavior, right? It'd be democratizes to the extent being a global organization think about a major creator being maybe 3, 4 years away from being able to produce a full Hollywood level quality film for a fraction of the cost or a AAA video game for a fraction of the cost, right? Disrupt or defending is really the framing of what we see happening next. So we mentioned days earlier, managing and mitigating risk. This is what we're seeing. And again, to Scott's point, where we're sharing this in full transparency. These are the conversations we're having with clients more big pitch. We're helping our clients get past a lot of these initial days over, copyright discussions, what are the ethics of this? How do we ensure that there is a brand safety point of view, what does it mean from a technology and infra-security perspective, who are the partners that we should be working with? What does it mean for our operational model. These are big meaty questions that especially global enterprise clients are sort of getting into and clear enough were part of those discussions. The exciting thing is in day 1. This is going to be about consumer experience. This is going to be about the original intent of digital advertising. I think there, especially, there's massive, massive opportunity. So if we're in day 1, I think part of our role and responsibility when we talk about now is how do we start on working AI now for our clients. And we use the term on look for a reason. The irony is for all of the conversations we had about AI in Cannes, I'm sure for all the conversations, everybody on this call is having about AI weekly or daily. Very few clients have a budget for AI because it started hitting the radar for many of them in March or April. So a big question that we see with our clients is they need to innovate, but they still need to run the day to day. We have saw that for them. We call it synthetic media. So we're going into our clients, existing clients and now we're looking at where there are core spending. And by using synthetic media, we're helping them earn more time and money and that is time and money they can reinvest for some of the structural foundational change that needs to happen. So an example of this is, for instance, Photoshop. Synthetic media is really about a spectrum, right? Sometimes we go full virtual production when it comes to image generation where we're setting up and we're shooting and we're crafting a composition in something like Unreal Engine. But then after we have the composition, we use AI to scale out and create all the variations of that on composition. That's a typical sort of way that we're seeing this come to bear. And what that means is we're replacing out of pockets, right, in many cases, a lot of travel when it comes to content production or sets with computing power and of course, also some manual labor with computing power. What does that mean for our clients? It means we can get their work done faster, key in e-commerce, making sure your product is to market as quickly as you need it to be, great for social where you can be more immediate annual responses. We've been able to move to cheaper, right? We can do the work faster and because of that, it's often also cheaper. It also creates the opportunity to create a lot more output. And that is interesting because, again, it goes back to almost the original hypothesis of digital advertising, right? If you have more stuff, is it better? And I won't spend a bit of time on that after the next case deal. We'll look at a quick case for HP, where you can see that spectrum in play. We're still doing an actual shoot in the studio, shorter shoot than we would normally do. But then preproduction, post production is using our AI-ready pipelines. [Presentation]

So a great example of that spectrum. How do we replace out of pockets, how do you replace manual labor with AI to make sure we're unlocking time and money for our clients. This is a big question. And actually, I think important as we start thinking about what the sort of future upside of AI is more better. And again, this is an original hypothesis digital advertising, right? If we have more, we can be more personalized. We can target more, we can be more relevant, more contextual. And we've always sort of hypothesized together as an industry that drives your conversion. So what can we do if content becomes uncapped, right? We can create endless amounts of content or only incremental additional costs. One thing we can start doing is really doubling down on regional relevance, right? We know the importance of that. This example is an interesting one. So the pictures we see here, the town we see here is fully synthetic, but this is fully generative. We don't recommend necessarily for our clients to do so because we think there are some ethical discussions to be had. But it's a good example. So in this case, normally a client might do a shoot with a few models, maybe 2 or 3. And it's relatively generic. If you think about Asia Pacific, if you have one generic talent for that market, it also means lots of markets aren't happening, right? China, I didn't happen Japan now in Korea hasn't happened. So in this model, you could do a single setup press a button and replace all of the talent with generated talent. What we do for our clients is we actually take pictures of real people, right? So we define who their preferred talent for each market is, we take pictures in that market. These people on travelling to expensive shoots, doesn't even have to be in the studio, we need about 80 to 120 different pictures, then we can train a model, the model and a nation just have that person available for all of your generated output at the press of a button. Important to change talent contracts for generative output, but massive upswing and the ability to create regional relevance. Another element here is this idea of audience infinity, right? If we're creating imagery and content and messaging and content that speaks to what somebody has an affinity with is going to allow us to be more in the moment and more contextual and more relevant. Often, those in affinities and segmentations are relatively top level, right, sports and music. With this, why not have instead of 10, 100, by another 1,000, why not have an image for every person that you're connecting and that you're talking to. Again, uncapped allows for that level of relevance in ways that just literary were not possible before. And then, of course, real time. I think one of the more exciting areas where we're seeing a lot of movement is in equal this idea that you can be conversational, that you can ask for changes in the flight, that you see the imagery in the clothing as it would break on your body type. There's massive amounts of opportunity here. And again, uncapped, right, we can now create real-time responses in ways that honestly just weren't possible before. So how are we actually using this? We can talk because this gets posted on the website as well. We can talk in closed door settings about the actual clients. We've anonymized this. But for a big sports brand, for instance, we're using synthetic to scale out e-commerce, right? We're doing the regional relevance here in ways I just weren't possible before. We're also optimizing what I will call the loan funnel of e-commerce. We've had this conversation with quite a few of our clients now, 50, 60, sometimes 70% of e-commerce assets aren't optimized, right? So we have a client that sells the same product on Walmart and on target with the same imagery, we know those are completely different demographics. Start thinking about the Netflix model, right, where they have 10, 20, 100 versions of the same thumbnail. And as you interface with the thumbnail, they'll start building a profile, what type of imagery you like. We can start doing that in e-commerce and push up the engagement and conversion without a massive explosion in cost. So lots of upside in synthetic e-commerce. Same for social. Again, this is now something that is solved. There's this massive issue called creative fatigue, slightly inside your baseball. But what it tells us is there's a moment in time when your ad start losing their ability to impact because they've been seeing too many times, and you need to switch them out. All of the ad platforms have massive amounts of data, and they can sort of predict when that's happening, so we can create those ads without eventual incremental costs, making sure your media spend keeps delivering, which, of course, isn't just great for clients, but it's also really great for the big ad platforms. Synthetic CRM, we're actually sales forces, marketing GPT AI partner who will be showing up together at Dreamforce to discuss that in more detail. But personalization in copywriting, personalization in e-mail headlines, just this idea that we can be iterative in how we talk to consumers and create a level of stickiness and really a focus on lifetime value, that wasn't possible before. And again, you're not exploding the cost by doing this, but you are driving much higher ROI increase engagement, better conversion. So synthetic CRM. And then the last one, we are constantly working on these and implementing them with our clients is what I will call synthetic production. This really is at scale. We saw a great example for HP earlier, just going through the spend at a global scale and see where can we start implementing this type of synthetic production solution, again, to free up time and we need to start investing in other areas of the business. And often those are AI-related, although I think in general, marketing spend is quite consistent. So there's other opportunities to invest as well. Okay. That's AI now with time check, that's great, right? And in every conversation we had, there is, I would say, almost why aren't we doing this at reaction which is, I think, a great sort of motivator to just start going. I think it's really important for organizations to start doing AI-related work because they need to get a level of comfort with it, both from an operational perspective, a legal perspective, just to make sure they don't get left behind. Soon this should be table stakes. This is why I'm always a bit careful when I talk about this type of moment. If you look at DCO, insider baseball bat that's called dynamic creative optimization, which is really automation. It should also be table stakes. It's been in market for a good 8 to 10 years, but we can still see it have limited impact at the enterprise level. So while there's a lot that can be done, there is some complexity to getting the full upside of AI in more complex organizations, which really brings us to AI first. And I would say some of the more interesting conversations we had in time. So how do you become AI first. And this is important to frame. So let's look at the digital landscape. We're saying this is the next transformation in digital. And I would go as far as to say that there are 3 big beats. There's AI. There's iPhone opening up sort of age of social and then there is the start of the commercial Internet. Of course, there are lots of smaller ones in between. Social is a good example. Programmatic is a good example. The great news is that with each of these ships, companies like ourselves ended up with more work on the other side, right? So any time you hear small talk about this, when you talk about this, you'll understand some of our bullishness, but it's interesting to think about what this means for marketers, right? I think the iPhone had a massive impact from a societal perspective, right, completely changed the way we interface with our friends, with others, with the world, with ourselves and did that in a relatively short time frame. I think AI is going to have similar amounts of cycle impact, but the time line will be a bit longer. We'll see if I turn out to be riding that. From a marketing perspective, I think it's bigger than iPhone. And to me, it is commercialization to Internet. Although we had these conversations with one of our technology partners in Cannes, and they said it is as big as man harnessing fire. So you might have to put a fire emoji just for free. Why is this important to talk about? If you look at the general length, the medium length of companies were in the top 500, fortune 500 pre-internet, I think it's somewhere around 86 years. Post Internet, it goes down to about 18 years, post AI sub 10%, I think is realistic. So massive, massive moment, and this is what is important for our clients, right? Everybody plays in this now space, which is generalist generative AI. Everybody has accessing the same things. There's no specific understanding of your brand, and there's no awareness of your needs. So what needs to happen next. We call this predictive and prescriptive brand AI that only you have access to, it's trained on your brand, it's framing your data sets, it's bill for your needs. Part of that sits in AI models. A lot of it sits in pipeline. And predictive and prescriptive is an important note here because it plays into some of the tenants of AI. And that's really part of what we're doing with our clients, how can you mirror some of those tenants to have the maximum upside of it. predictive and prescriptive is about the circularity of AI, right? If we have well-structured data sets, and we're closely feeding that data set with insights that we're getting from what is now a much broader way to talk to your consumers, right? We have more ways to talk to them in more real-time way to talk to them, so you're reflecting more information, we should be able to get predictive on that data set, right? We should be able to start thinking about where do we invest more of our money, which channel is more important. It becomes more real time. You can already see that reflected in our latest MediaMonks product, which is AI First, which really leans into this, right, 100 different models fighting against each other to give you its best prediction of where your money should go. If you can start predicting, you can start being prescriptive. This is the circularity, right? Does that mean it starts setting up the initial version of the media plan. We have a little conversation in can about the disruption media will go through. If it starts being predictive, can we be prescriptive around the type of creativity or the type of ads or type or [indiscernible] that's going to resonate, right? So circularity. Some other elements that are important. We talk a lot about the idea of the compound effect, right? The speed of change and how AI builds on itself, which is really why we're pushing for our clients to be fast and first. And then I mentioned adversarial just now, the idea that it doesn't have to be one AI figuring something up, right? You can have 10, you can have 100 fighting it out for the best results. which is really interesting, for instance, about one AI work stream that generates a you'll probably need a brand safety in a Q&A AI checking those ads in real time. So predictive and prescriptive. I thought this was really interesting. This is a single person's view. So you have to take it with a grain of salt, of course. But somebody at Google had a memo leak. And I thought the memo was really interesting because it opens up a big question for our clients, right? The take here was the barrier to entry has dropped, a single person with some free time and a beefy lateral can do things now that used to only be possible to do in the size and strength of a big technology companies like Google. And the messaging was do we still have a moat, right? What is our moat. And that opens the question in an AI world, what is a moat for our clients, right? Everyone is accessing the same things. What do you have that only you have that can making you show that? And that's your data. So thinking about building an AI first marketing organization starts with the VEON data. And already, we have something called design to learn. If you look at our industry, I would say, the most underleveraged part of our industry is the amount of media money that gets spent and how little we learned from it, right? Massive amounts of money goes out the door. Everybody is focused on the campaign results, but there's very little learning that gets generated outside of that campaign. And honestly, still to this day, a lot of that data isn't real time, all that data has looked in ports and dashboards that don't really have meaningful input on your next campaign. So we already have an internal methodology which we call design to learn. Really, in the age of eye that becomes design to machine, right? The idea that this isn't just about having data or a data lake. It's about having constant streams of fresh data, right, as you're pushing your media out in the world, what can we learn from a brand performance perspective, right? What's converting, what's creating lifetime value? Where does our growth come from and really making this a constant optimization process. Because if you think about the age of AI, actually, this was no proper in Cannes. The last decade or so, I would say we've lived in a marketing advertising landscape where it was about optimizing, right? There was a big idea. All of the eggs were in that basket. So you had to optimize that single big idea across all of the channels and all of your funnel. And that type of optimization was incremental, right? It was, let's say, AB testing, let's change the core of a button. There's a different call to action. If you think about AI, what it opens up is not optimization but maximizing, right? The idea that you can do more. So why wouldn't you have that adversarial model, right? Instead of having one creative idea, we know for a fact that if you have 6 creative ideas in market, your cost per action will drop by about 40% to 60%. But there is a cost prohibitive component to that, right? We're getting to the point that, that's no longer an issue. So instead of optimizing a single idea, why don't maximize and do 6 why to do 10, right? This is really the shift. And my honest assessment, and this was the conversation we had in Cannes, again, Cannes is the capital to the big idea. I don't see that model beating this model at any scale. Because if you may start putting millions of models to work, it really becomes a smart as pipeline wins conversation, right? So this is a very simplified overview of our pipeline, very simplified overview. But the key components here are there is an analytical piece to it, right? This is about data. You have first-party audience data, we're talking about first party trade data, this idea that we're creating more ideas, more output, more personalization, more targeting, more information and learning, and we're feeding that back into our own data set. And then, of course, we're training it on our brand history, our product, anything that we need. That data set will be the most viable thing you have in the age of AI, right? Only you have it, if it's ready for machine learning, you can keep building on it, and it will be a massive lever and unlock growth. And then the generative AI piece allows you to respond to that insights in real time, try new things, new campaigns, new content, new experiences is really about activating the insights your analytical data set now provides. So a quick time check. 80% of all conversations that we have around AI sort of center on 3 things. One is efficiency, cost savings. The other one is job anxiety. And then the third one is the question, will it kill us all? I think it's important to also look at what we have called always imagined, right? We're in a creative industry. We're excited by the opportunities that we have because our channels open up, our technology opens up, we can now do things that literally weren't possible 6 months ago. And I cannot stress that enough. It is magical, right? And I don't think it gets enough attention. This is a very small example, but I wanted to show it because it happened last week in Cannes. We have a party on the Wednesday, we always have 4Ms in the name, massive music medium organize it together. So instead of doing pro night, we organize point night. And we had an AI photo booth I have never seen a line this long for a photo booth, because it was doing something that nobody has seen before. It was creating perfect, beautifully creative pictures of groups of people dressed and done up as if they were going to 50, right? Massive line. Lots of people coming up to us going, is it a product? Can we get it somewhere. And this is just because this wasn't possible 6 months ago. People haven't seen it before. And that to me is some of the excitement, right? We can get to some of that magic that has to be part of digital advertising, especially. I'm going to show another piece of work called Kiki, which again is something that literally just was not possible 6 months ago. This is a meta-human that takes in real-time speech, right, listens and can translate that speech to 3 different sign languages. Why is this interesting? Why is it important? #1, you could scale it up and scale it out, right, everybody could have their own version of it. But it's also important when you think about breaking use. It's not difficult to plan for sign language if the Olympics are happening. But it is difficult if there's breaking us or some type of warning. So this opens up that information distribution to everyone. We'll do a bit of voice over here. I won't show the whole video, but we'll make sure it's on the site as well. And again, this sort of responsiveness, real-time personal/personable is a real shift change in what was possible in visual initial advertising until very, very recently discussions about extending. So you get the point if we go on some other ideas, right, some of the prototypes and test projects that we're doing with clients when it comes to consumer experience. using AI to do life for the design, right? Not everybody that's creative is also a design. The ability to do that via voice and conversation now and actually pick that up, unlocks massive amounts of opportunity for consumer engagement. Conversational marketing. I think this is really interesting. If you talk about the hype cycle that we mentioned earlier, probably about 6 years ago, a massive hype cycle hit our industry and was about conversational marketing. It was about bots. It was really this idea that we could sort of move into what I would call agent assistance. The technology wasn't quite there yet, right? A lot of us have one of these machines at home. We use it for some things, but the conversational piece is still a bit light. We can see that working now. We've actually been in this space for a while now. This is a ChatGPT 2 implementation, including a deep face with the weekend for Spotify, beautiful piece of work that picked up a lot of awards last year. But this actually is a new engagement that we had launched about 2 weeks ago for the Champions League final. So if anybody is watching and is Man City fan, congratulations. This is Del Piero, famous X player, had an amazing career with Juventus. He was your ambassador, your second screen at scale, right? We had a conversational engagement all on top, where trained on his tone of voice in his career, you will be able to chat him in real time about what was happening in the match because we pulled in all of the match data as well. Massive amounts of people having these conversations at the same time, super sticky, right? So we're seeing this type of marketing really explode at the moment, lots of interest in the beauty industry, especially. We seem to be first when it comes to these types of engagements often. Okay. Breathe, so this is a message to our clients as well, right? Everyone is late, and that's just down to the speed of change, right? I think everybody got hit with this somewhere between the last 3 and last 9 months and is really trying to reframe what their plans are. The main message is don't be too late, right? You have to get started. I think that's where we are a key partner, right, unlocking, unblocking just being very practical on where to start. So what are we doing with our clients. We'll take on pieces of content production, right? This is an easy place to start creating some space that you can reinvest. It could be big photoshoots where we take out of pockets out of the production, could be just lots of e-commerce content that needs to be created relatively easy to start and just get going, right? It also gives our clients and these organizations a real sense of what's possible quickly. What should you invest on in or reinvest that in? What does your brand or what it feel like once you experience it through AI, what is the consumer experience when you look through that lens. And then we're doing a range, some of it is pure ideation and some of it is all the way with go-to-market pilots. Another element, our end-to-end model, including media and full integration data, where we're not going for the optimization of the single idea, but the maximization of both AI and walk a bakeoff, there is no way a human only model or even a human plus automation model beats that structure, right? And we're seeing a lot of interest there. I think there's a general consensus that the media part of our industry is probably a bit hesitant to embrace this for understandable reasons. But this is really where you start on walking not just a bit of an efficiency play, but a real return on investment, which is key. And then it's about starting to lay your AI foundation, right? The idea that to be ready to be machine learning ready, you really need a data structure that allows you to build your own data set. And again, that data set will be the most powerful part of your opportunity for growth, right? Because you have that nobody else -- no one is playing in this generalist generative AI space, right, you can actually get ahead of the rest of your industry or category is doing. So that's our discussion with clients to Scott point, this is the discussion we had every single day. Really interesting conversations, lots of practical. Let's get started with now lots of more strategic discussions about how we get to next. The conversation that we also had in Cannes with a few analysts, especially is about the S4 discussion. And we talked a bit about this in our last earning call. But why S4? #1, we've been in this space for a while. This image was not pointed. This is me before I had to worry about generative AI. But we've always known this is a way to do a couple of hours from output, right? It means we've had teams on this for a while now. And it means we already have AI-augmented workflows in both our talent communities and our end-to-end teams. What does that mean? A talent community like our team uses these tools at an extremely high level of fidelity, means they can do more the same amount of people. Films, we just saw the HP example, right? A lot of post and preproduction already hits these pipelines that are continuously being optimized. These playbooks are part of the largest training exercise our organization has ever done and is still doing. We're fully committed to getting everybody in our team to be the best trained casters or tech person or AI person in our industry. We've also been able to solve key challenges, I think, sooner. And again, this is very much what we're getting back from our clients. We're fast and first in many of these things, and this allows our clients to be fast and first as well. So a good example here is amongst GPT when we were reading about companies in our industry or outside of the industry, blocking access to ChatGPT because the privacy constraints or worries, we already had our own ChatGPT up and running. Everybody has access to it. There are actually some really interesting implementations in Slack where we have it as a brainstorming tool in Slack, for instance, which is working really well. We also have what we call an AI workforce, which we're building out. Turing.Monk is a great example, which is really an analyst on top of data, right? You can ask a question and it gives you insights that will normally maybe take you a bit more to figure out because you now have this sort of analytical AI doing the heavy lifting. We're partnering with all of the leaders in AI for the enterprise level. We have great access to startups as well, but very committed to the main players. We are in their feedback loops, to my earlier point with Salesforce, where their marketing GPT, launch partners from an AI perspective, partnerships with NVIDIA, Google, et cetera. We're a little more agnostic. There's been a lot of sort of press push around solutions that really focus on a single tech stack. Our position tends to be a bit more agnostic very bullish on what all these companies are doing, but we're pulling it together in ways that we think are best for our clients. And then we end where we began a Single P&L, why is this important? If you look at generative AI, the easiest way to think about it is, hey, there are some cost savings because we can be a couple of hours from out there. That's fine. And I call it base reality, right? That's clear. Let's do that. But if you want the upside, if you want to be able to link fog, if you want to be able to defend, it needs to be data, content and media as a single pipeline. Of course, that gets infused and powered by technology stack. Our company is structured this way. We're able to deliver this to our clients in ways that are really difficult to do in a fragmented landscape over a bunch of P&Ls and a bunch of labels and a bunch of brands deciding what AI strategies. So this is part of our foundation. We've been really clear on why that is. I think in AI, that just becomes even more important. And with that, we're done. And I think we have about half an hour for some questions. Thank you. If anybody has questions in general, feel free to ping me your score. But we'll have Scott be a Q&A MC for a bit.

Yes. Thanks, Wes. That was great. Really appreciate it. We've got quite a lot of questions. I've had a bunch of e-mails, but there's a few on the chat as well. So I'll read a few out, going to put you on the spot here. Some of them are quite general and some of them are pretty specific. So one of the questions is around obviously, technology is a matter part of this. So do the tech-led providers, so they're more the sort of consulting companies, systems, integrate or companies, are they better placed to guide clients through the AI transition?

So I have 2 points of view on that. One is we have a very strong tech services offering, which is -- it's a bit more practical and it's really about getting the pipeline in place. I do worry -- a lot of messaging is this initial transformation 2.0. I think you have to be a bit careful with where you invest because of the magic button problem. It's going to be extremely easy to build something for 6 months for 9 years. Digital transformation work often to 2 years, maybe even 5. The space moves so quickly that I would be careful. And you can actually see that in some of the initial moment, right? People training an LLM on massive amounts of their own data, which is expensive, knowing that actually using existing LLM with much more trading set for a fraction of the course has 98% of the same output, right? So I think it's -- I think it's technology enabled and you need to build pipelines, but it isn't necessarily -- let's start a whole digital transformation wave, which I think the big consultants are going to focus on first.

I guess following up on that, some questions around whether we intend to sort of deliver and generate AI platforms ourselves so that Marcell from Publicis has given as a kind of comparison there. So do we intend to sort of go down that route ourselves or partner? And then also on the partnerships, you touched that a little bit. Can you talk a bit more about the partnerships we have and there's a comment that the sales force on sounds a little bit more go-to-market focus than some of the other ones?

Correct. So the answer is -- and full transparency here. I think a lot of -- and I went through this, right? Remember that I, at this moment in November or October last year, like it's happening, what's next? The instinct is to build a bunch of stuff. The reality is with all the start-ups already in the space, the magic button problem is real. And I talked to several clients. All of them will have a very smart AI leader in their organization very quickly, right? A few of our clients already have. They're not going to be tricked by a ChatGPT, white label ChatGPT with your agency logo on it. They just won't and they shouldn't be. So it's a pipeline discussion, and it's tapping into the best of what's there. That's what we do. I don't think it is the smartest use of resources or time to build a lot of bespoke work knowing both the open source space and how much activity is outside of that. Flip side of that is, yes, we have pipelines, right? Pipelines that allow us to connect data and congestion in media in ways that are our AI first, but a lot that's powered by open sourcing walls in the landscape. The second part of that question…

Around the partnership with market.

So I'll start with Salesforce on 100%, it's go-to-market, right? What I think differentiates us for a lot of our clients and also for our partners is we're already doing the work, right? We have the cases we're able to show upside against these cases. This isn't, let's start doing some AI work, right? So for our partners, that's exciting because it means they can be a market faster as well. So with Salesforce, it's very much a, let's go to market together and start getting our clients the upside of using the technology in the best possible way. If you look at the partnership broadly we're a little, which I guess you want to expect us to say with [indiscernible], but we're a little less PR-worthy than the massive network. So a lot of that messaging is on very deep partnerships. And I'm personally interested to understand a bit more about some of the commercial models. For us, we have the same relationships. We just are a little more agnostic on how we implement the technology in the best possible way for our clients.

Great. I'm getting a lot of the sort of similar questions around the, I guess, the impact on the business model, if you like, something we've discussed quite a lot, and it's early days. So in some respects, it's just hard to say. But a lot of questions around if this is making things faster and better and cheaper and easier for clients -- then I guess 2 things. One, weren't they just simply in-house it and do it themselves? Or 2, where they just simply focus on this as an opportunity to cut costs, which shrinks the addressable market for agencies.

So commercial model, and yes, we have had a little conversation around that, and we're spending a lot of time with procurement teams on the client side to understand how this moves, right? I do think there's going to be changes where it can just be time and material models, but there's opportunities for other commercial models as well. So we are testing quite a few of them, very close relationships with German teams that are also interested in figuring out this space. I don't think it necessarily decreases addressable market. But there is a theoretical component to that, right? I think there is a consistency in what percentage of GTP density spend in marketing. We shouldn't forget the amount of opportunity that's now available and extended consumer trends. And we shouldn't forget that if marketing makes people money, they tend to spend more marketing on, right? It's sort of a bit of a compound effect. So I do think there's going to be winners and losers in our space. But I think that money just distributes players that are best suited to help clients as this happens.

Yes. Quite a few questions as expected, given the audience on this call around if there are sort of efficiency opportunities for us, does that mean massive increases in margin for us. And that that's something we've discussed as well before, but probably unlikely, right, given that clients expect some level of transparency and sharing on that.

So I would say there's a few pieces to that. One is, and this wasn't planned, but I was surprised by it as we went through the week at Cannes. We are probably at our scale. So if you look at the industry, sort of we're top 15. It sounds like nobody else was being clear about some of the realities here. So we've been clear that if we're generating savings and efficiencies, part of that goes back to the client probably sooner than other people seem to be saying. There's definitely internal efficiency opportunity still from a workflow perspective. From a headcount perspective, we've also been clear to our teams, like I see that was another really interesting part of can, where there was a lot of -- I won't make names, but I was at often where there were amazing AI demos from a technology partner. And then the messaging was always be, it won't replace people here like sort of world now, right? So we've been really clear to our team, we have a commitment to making sure everybody is the most well trained. We are going to focus on being AI first in market and really being at the front lines of what that means from a commercial perspective because we believe we can do more work with our existing team, our existing structure. If that doesn't quite balance out, I think, also, had to report out 1.5 weeks ago that said our industry headcount will go down about 7.5% in the next 6 years, right? It's a reality of what's going to happen. But our goal is to come out on the other side with a lot of home questing and commercial well. And our team understands that's the -- that's all.

On the talent side, it's a good segue into that a few questions around that. Is there enough talent in the industry that understand this, know how to use it? How much training is needed? And what are we doing in that regard?

Massive training exercise on our side. What I'm -- I wouldn't say surprise because we know that we have great teams. But although the innovation starts at the grassroot level, right? It's people that understand their area of expertise really, really well and can attach it to what AI now makes possible. And sometimes that's an efficiency piece or a speed piece and sometimes it's a creative opportunity. So a lot of what we've been able to do is because our teams are very excited by the technology and there's a lot of grassroot sort of ambassadorship. I would say not that dissimilar to what happens with other technologies that a great subject matter expert that also knows the technology is miles ahead of just a credit subject matter expert, but also miles or somebody with little subject matter expertise that uses technology really well, right? The combination of both really looking at some of the work that our teams are doing is standing.

Actually, an interesting question here and something that has come up before with some of the product bases that we're seeing from the big platforms. But there's a criticism of sort of generative AI advertising that is banal and derivative. So are these tools that are better suited to SMEs, doing sort of social media or search or whatever than high-quality creative for large enterprise clients.

I think it's easier to start at the SME level. Also because of risk. I don't want to speculate. I'm sure there's a million conversations on it, but if I'm a big platform, I feel a little less excited about launching generative AI and enterprise-level business, right, because one mistake has massive consequences where the SMB or SME level, it's a lot easier to, I guess, make some mistakes. I think it's too easy to say it's a derivative or now it's a discussion, right? It's something derivative because it can only create based on what it already knows. There's a lot of work out in the world from an advertising and marketing perspective, that is pretty straightforward promotion work, promotional work, activation work. I think what I saw in Cannes, I think this is sort of the rabbit is dead. Every time we think about AI, we look at the very best humans can do, and then we go, it can't replace that, right? It's not going to write shake per play, although a few years from now, maybe, right? But can you start writing PDP page copy? Yes right? So I think people struggle with the spectrum, right? It's -- they're like it can do that absolute peak, so I can't do anything and it's just not the reality of the technology. So my take is it's a great tool, you be -- you'd be crazy to use it, right? Because if you don't, somebody else is using it. And I can tell you for the fact that we won't win that in the nickel term.

I've got a few quite specific legal questions, so I'll preface those by saying you're not a lawyer.

Is not legal advice.

Anyone should get their own legal advice exactly. But one of the questions come up a couple of times. So if we do get sort of clients down that pipeline, you described it becoming AI first, and we're working very closely with them, they have their data in an AI model, and we're helping them with that. Where do you think the IP will sit on that? Who's going to own the IP? Is it going to be a client? Is it us? Is it the tech platform? How is that going to work?

On the output?

Yes.

So I don't -- globally, it's going to differ. But if you look at the U.S., for instance, the current copyright law means anything that's created purely by a machine or a computer can be co-provided which I don't see as something that holds, but case law is quite slow. So we'll be 2, 3 years out before I think that's completely formalized. So there will be discussions around if you're pushing stuff out in the world, how important is it that what you're pushing out is copyright or IP defensible because that might mean you need to make some changes in how you go through the pipeline. Those changes would be marginal, though, right? And just to be clear, this is a pipeline in people story, right? Smartest pipeline wins because smartest pipeline gives you the upside of uncapped, if you have uncut content, how do you make sure that actually drives your business in meaningful ways in e-commerce and CRM, et cetera. And there's people in that pipeline that are subject matter experts, right? They're making sure that, that pipeline is running at the very highest level. There's reinforcement components to it with other important sort of AI tenant. But yes, the copyright and IP conversation on output is going to change -- is going to be a different region to region and depending on how important you find it that an ad output on a Facebook or has that defensibility we'll have to change some of your process.

Okay. A couple of questions on sort of time lines and timing. And again, these are, I guess, our views, and we're quite bullish on what the opportunity is here. But apparently, some of the other sort of people in the industry are really cautioning around the fact that this isn't going to happen at scale anytime soon. What's your view on that from -- I know you've locked yourself in a room last week with clients, you're spending all your time on calls and clients going through this right now. What's your view on that given the conversations you're having?

So I don't think it happens at the same speed everywhere, right? There's parts of our industry that will be extremely slow and there will be parts that will be faster. The same will go for sort of industry categories, but there isn't a major client that hasn't decided that this is one of their top 3 areas of focus. It's in every conversation, there is a follow-up that sits at the C or exact level because everybody has decided that this is key. If you would ask me why you're hearing some of that trepidation, if you haven't figured it out yet, and there's massive disruptions to your existing business model, you're going to want it to move slow right? And we're probably more bullish because we actually see the upside. So we're part of that going to move on. That's the reality of at least my takeaway from Cannes.

Yes. I think that links into my next question, again, I have this a few times. But given that the sort of the partnerships and tools are all out there and we'll all be able to use them, how do we create the moat for ourselves at S4 and what's going to allow us to differentiate ourselves from our competitors in this area.

I think in the current landscape, I would say there are 3 things. One is the fast and first. Speed, speed is key, right? If you are able to be the key AI partner to your clients and being was reality being real and what's possible right now, not defensive. That to me is the biggest push that we're putting in place now because being an AI partner to clients actually opens up the aperture to the whole business, right? We literally -- half of sessions where we go through all spend of our clients. And we do what can we consolidate, what can we optimize, what can we do to make this more efficient, right? So for us, fast and first is key, and we're getting that reflected back from technology partners. Second piece is talent and team, commitment to training, no fear, being very aggressive in our messaging also internally about that we're all in on this. And what that could potentially mean. I think there was a lot of -- I find some of the messaging kind of be patronizing when it came to that test. And then our organizational model, the Single P&L because we talked about it so much, I think sometimes people start seeing it as a tagline versus an active choice that makes things possible that it's just really difficult to execute elsewhere. But we're in a room every single day of the week with an integrated team, right? We're not in 5 different agencies with different P&Ls, deciding who -- or fighting who gets to decide what's next, right? This is an organized top-down mandate that include integrated. So those 3 things, I would say, are the current moat. I think there are really interesting conversations that we're having as well around data sets at a conversation with a big credit festival term, right, training one of our AI workers on their credit dataset, right? What is good advertising, what has already been done. This idea that you can use interesting data sets and build your own AI workforce and top of those data sets. And some of those data sets are the ones we have, someone that says that we're sort of building into our offering. But I think just having meaningful understanding of how to use data in a way that impacts your business in ways that sort of make you more distinctive.

I had a few questions around the media side of things. I think you touched on it a fair bit in your presentation, but really, there was quite a lot of conversation, I think, last week and can around what kind of disruption could happen in the traditional media planning and buying approach that agencies have right now. So do you have a view on that?

Well, sure, Martin has the most point of view on this, is why trust a 25-year-old media planner when you can go with the predictive and prescriptive model. The reality is the media part of our industry at large has sort of skirted even automation at massive scale, still a massive amount of manual labor, which is not more refocused on, right? We've been automation from labor. We've been very, very focused on sort of productizing that. So I think this just changes the footprint that you will need to service clients at a global scale. Scott, you can talk more about that complexity, but that's tough because a lot of revenue margin sits in our services. And if you have to change that, I think that's going to be tough.

Yes. I think it's another opportunity for us, and we have a pretty different approach on the media side. We come from a programmatic performance-based approach. We use media into technology already. So I think yes, I think as things change in AI becomes more commonplace and disrupt some of the much more manual approaches that are prevalent in the industry. I think it's a great opportunity for us. And that actually speaks to one of the final questions I've got here on the chat is really around headcount. You referenced that Forrester report around sort of a potential reduction of 7.5% over the next 5 years. I think there's a Goldman Sachs report out there about the impact it could have on the wider job market. I think these things are all quite speculative at the moment. But are we seeing anything yet from a headcount perspective? Do we have a strategy around what the implications could mean for our business?

So for our business, we're clear, efficiency targets against every talent community and NTNT. So again, talent community is a skill set like our art and visualization team. And then NTNT is like a pipeline like film production [indiscernible]. Efficiency targets and then that efficiency target really translates to more revenue that we should be able to handle with that team or potential headcount reductions if we don't manage to sell into that additional level of efficiency. And again, we've been non patronizing about this, which is key. I think you have to be clear when something as fundamental as this happens, you have to be clear to your team what's happening and what the plan is. So that could go either way, right? If we have initial signs are pretty decent. But if we bring in more work against how we refine our business to be AI first, we can do more with our existing team based on those, we call it ratios or AI team actually, or the headcount goes down, it's the reality of what's going to happen. I think the 7.5% that Forrester had, they talked to a bunch of people. There's the component to it that is difficult to predict. Every time something as impactful as this happens, dogs go away and other jobs come back. So I at our size, I'm quite bullish that we're well positioned, right? In general, 9,000 people, I think is a great size for what's about to happen. And it's difficult to predict what the new jobs will be 2 years from.

Yes. I had a couple of questions around, again, partners and technologies. So are we ourselves going to have to invest heavily in technology buying in lots of NVIDIA chips? Or is that something that the sort of the tech companies are more focused on?

Yes. It will be stiff competition to buy NVIDIA chips. That's for sure. Realistically, the way we've always operated is -- and it's actually -- it's interesting if you go back to the original positioning that we had when we launched S4. It's the best partner to sort of build on top of the technology landscape, right? And we had 7 or 12 companies that we really focus on and then making sure we use that technology to combine -- to connect our clients to their consumers in the best possible way. That's not something that I see changing. I think there's investments in team and talent. There's investment in some of the pipeline components. There will be investment maybe in specific model training, but the concept of being massively invested or overinvested in GPUs is not in the plan.

Great. Well, I think that pretty much covers the questions we've had so far. If anyone else does have questions, please feel free to e-mail out myself, [email protected] or was directly as well. We'll be happy to answer them. This is obviously going to be a topic that continues to engage a lot of you, and we will continue to provide content and opportunities to discuss it. And it's something that's moving very quickly. I mean I think the irony is we did cover this in our FY '22 results presentation. We had a session that Wes did introduce in the topic. And no one asked us any questions or had any interest in it, and that was only 3 or 4 months ago, and now it seems to be the thing that everyone is focused on. Hopefully, you've got a good sense today of our positioning, what we are focused on, why we see this as such an exciting opportunity. And again, we'd be really happy to engage with you further on it. So feel free to contact us. And thank you very much for your time, Wes, really appreciate it. [indiscernible] Thanks a lot, everybody.

Have a good one.