Welcome, everybody, to TP Capital Markets Day. I'm Mike Lytle, the CEO for TP in the U.S. I'm still going to call us Teleperformance a few times on accident, so just I'm self-declaring I'm going to make that mistake. We have a full agenda today, and just a couple of housekeeping rules. If you'd like to ask a question during the session, there are microphones on each of the tables. Outside, you may have seen there's a few stations set up to demonstrate the technology that you'll hear about today, and a real live demonstration, so you can see how the technology actually works with our clients and end customers. The TP team will also be outside, so ask us some questions, talk about the things that are on your mind, and we'll be happy to engage. I'd love to introduce Daniel and Thomas, who will walk through the strategy overall for TP to kick off our day. Daniel. OK. Oh, video first. The world never stops moving. Markets evolve, technology accelerates, and AI? It's unlocking possibilities everywhere. With change comes opportunity to move faster, to see further, to lead smarter. Because clarity doesn't just happen. It's created with intention, with orchestration. It takes more than algorithms, more than instinct. It takes expertise in people to lead with empathy, in process to scale with consistency, in technology to act with precision, and in domain expertise, because a travel solution won't fly in retail, or healthcare, or finance. When you bring AI and EI together, when machine learning meets human understanding, you don't just keep up. You lead with confidence. You create clarity. You build control. You unlock value. This is Orchestrated Intelligence. This is TP, turning today's complexity into tomorrow's advantage. Welcome to Capital Markets Day. Yeah, it's work. Good morning, everybody. Thank you for being here. Thank you for the people. I think there are 170 people who are following us remotely. Thank you for everybody. We are going to try to be efficient and clear. Today, oh, first, my name is Daniel Julien, for the one who never heard about me. I started this company a few years ago, and I'm still the CEO of the company, by chance. Today, my partner in crime and my partner in management, Thomas Mackenbrock, and myself, we are going to introduce the whole session, telling you where we are, where we want to go, and how we want to go there. There is something very important, because you are going to see the day is going to be very focused on the AI integration. There is something very important for you to keep in mind, which is that the motto, the main line of Teleperformance, is AI augmenting our human experts. Why? Just because all human decisions are made from two factors: emotions and rationals, and reason, emotional and rational. All the practice, decades of practice, all the academic work shows that in a human decision, typically, at least 70% of the decision has an emotional component, and the rest is supported by the rational. Our role, of course, is to have human experts who are empathetic to the problem that meet the customer, but we want to empower them with all force of the AI. This is what we are going to try to show today. Where are we today? Then Thomas is going to tell you where we go tomorrow. OK. Our job is very simple. We are a business service company, and we are here to create competitive advantage for our clients. How? Through digital business services, meaning taking full advantage of the digital. You know we are a global leader. In the customer experience management market, we are the number one. We serve 170 countries. Our revenues are over EUR 10 billion. That is a scale. We have scale, and we know what we are talking about. Second, we are trusted by our clients. In the core service, we have more than 1,500 clients with an incredible stickiness, an average of 13 years, meaning that many among our largest clients are clients for 20 years, 25 years. I mean, one generation. Very interesting, specifically when those companies can be among the top companies of Silicon Valley, for example. Two-thirds of the top 100 companies, of the top 100 brands, are clients of Teleperformance. We have scale. We have a strong relationship with our clients. Yes, we are a people service company. Being a people service company, we want our people to work in the best possible condition, intellectual condition, and physical condition. That is why 97% of our people work in subsidiaries of Teleperformance that are Great Place to Work certified. Also, more than 70% of our promotions are internal promotions, meaning that we absolutely want our partners, our colleagues, to blossom to their full potential. The number of people who started as customer experts at TP and who now are CEOs of a company or heads of a large vertical is something pretty amazing. Finally, we are a public company, and we deliver value to the shareholders. By the way, we return quite a lot to the shareholders. In 2024, our EBITDA margin was 15%. Yeah. We We increased the dividend per share between 2022 and 2024 by 75%. Much more importantly today is the fact that we created, in 2024, in the environment that you know, EUR 1 billion free cash flow. We expect to continue. Before I pass to Thomas, I would like to show you that. You You know Everest Group. Everest Group is the business analyst that is probably the most renowned business analyst company. This is where they place us in the customer experience management. You see for the Americas, and you see for the EMEA. This is external data. I'm not going to mention the industry awards, but I can tell you that I was very happy two days ago to discover that we got a golden award in India for integrating AI solutions in our services. I can tell you that the company that I created 47 years ago has changed multiple times. We are like this animal who leaves their skin, changes skin. We are continuously morphing. In fact, the pre-2000 have been the years of the geographic conquest. This allowed us to start the geostrategy, the offshoring process. The 2000 to 2010 was the fact to put together fragmented channels in one single seamless experience, the omnichannel. 2010 to 2020, we started to integrate much more the efficiency of the digital with the robotic process automation, with the optical character recognition. That was a big transformation. After the 2000 and now with the emergence of the transformers, of the large language model, of this fantastic possibility that is offered today, we are going to the next step, which is the cyborg customer expert, made of the human empathy and all the intelligence that we can get from the digital and Gen AI. Of course, we are going to speak about Adjantic today. Having said that, I think I'm finished, and I need Thomas to come if he accepts to stop to speak with his friend. Thanks. Good morning, everybody. It's a pleasure to be with you all today. We spend the next two hours to talk about the future of TP. We will talk about why we are excited about the future of TP, what our plan is to deliver this exciting future, and what ingredients and sources and capabilities we already have in place today to do so. There will be plenty of information, but I would like to leave you with these four messages. We will guide through all of these messages now in the next following presentation to explain what we already do today, what are real use cases we see in our businesses, and how do we want to act and expand upon them going forward. First is our strategy. We have a clear strategy in place to expand our business and to accelerate our growth momentum going forward. As you know, the guidance for this year was 2%-4%, and we see clear pathways to accelerate this momentum going forward. Secondly, we're very proud to announce the platform TP.ai FAB. Fab, as you might think, stands for not for fabulous. It also stands for fabulous, but it stands for foundational AI backbone. Essentially, it is the platform that is built upon the hundreds of AI projects we have done the last two years, our deep know-how working with our clients across multiple verticals, our own proprietary solution, as well as the partner solution we want to bring in our ecosystem. We will explain to you how we combine this Adjantic AI layer with human orchestration, because we believe this addresses the real pain points and the use cases that our clients ask us. Third key message, and you will see many colleagues today on stage, is that we believe we have the right team in place. They're hungry to win. They're hungry to transform the company. When I travel around the world, I'm always amazed by the quality of people TP has in every geography. It's really an all-star team. You will see some new faces today and some old faces who will explain the journey we're heading to. Lastly, the financial numbers. TP, as you all know, is an asset-light business model that is highly cash generative. We expect this to continue in the future. We estimate around $3 billion net free cash flow in the following years. This includes our internal AI efforts. A significant portion of that will be returned to the shareholders going forward. How we're going to do that, we will guide you through as well. Let's double-click on these points. First, our strategy. I always say our business is 90% about execution, but you need to know what you want to do and what you don't want to do. You need to know where you can focus on, because you have to have a right to win in the different areas. It's very easy. First, we see plenty of opportunities in transforming and growing our cooperation. Yes, we need to change what we do, but we can build on the capabilities of people, process excellence, technology, and expertise in our core business, and thereby broaden our existing clients, extend our share of wallet with them, and thereby drive growth. You will see multiple examples of how this is happening already today and how we want to expand on this in the future. Secondly, and we talked about this in the last investor calls, I'm very excited about our capability to expand our value chain. What does that mean? Historically, we have been very strong on front-office and mid-office services. We see today in our business, as you know in our numbers for 2024, we see double-digit growth in our back-office-related tasks. You will hear from the colleagues Mamta, Miranda, and Himadri how we expand our value chain from the front office to the back office in our core business services, as well as for our verticalized solution. Expanding this play, having a more vertical view, gives us more growth momentum going forward. Thirdly, AI value chain. With the capabilities that we have today, we see us well-positioned working with our existing clients to drive new opportunities in AI data services, as well as technology and consulting. We launched, as you might have seen in February, TP AI. We are very happy to have Anish being with us today as our new Chief AI Officer. Here we will demonstrate how we see already today in our business the capability to capture new businesses, in particular in data services, as well as in tech as a service. All of these elements combine—it is not a surprise, because we operate in a large market. Yes, it is changing, but it is a huge market—allow us to increase our TAM and really focus and execute on those areas we see the growth momentum. The ability to act quickly with our capabilities that we have in place will determine success. The second key message was TP.ai FAB. You can say every BPO announced a new platform. What is different? What is unique? Our approach is not—we are not a software company. We are a B2B services company. We are listening to the needs and problems, if you will, or opportunities that our clients present us. When we listen to them, one key topic we hear often and often again is, how do you orchestrate? How do you intelligently, seamlessly, safely combine the human experts with the AI? Because very rarely, every tool and every news you see, every demo you see on the media is a plug and play, in particular in enterprise environments. How do you build intelligently that orchestration? I really do believe TP, with its global footprint, its global scale, and people, process, and technology, is uniquely positioned to drive that orchestration. We will show you how this is already happening today and how do we expand this in the future. Second key topic and message we receive from our clients is safety. In order to win and to operate working for the largest companies in the world, you need to have a secure, safe, open cloud infrastructure. We have, if you have in the breakout time, our CIO, Dev Mudarya, here to explain to you how we are building this infrastructure. We invested three-digit million amounts already today in our safety. We will continue to invest in that area. This is an integral part of our TP.ai FAB. Thirdly, specificity. What does that mean? If you read about and if you hear about AI, it is very, very rarely that a general AI solution is able to drive real outcomes for clients. You need to pick on the value chain, on the process that you drive for a client, a specific use case. You need to apply for that use case to make it practical. It is not that you use a general tool that will solve all the answers. It is about vertical-specific blueprints that drive outcomes for specific needs, whether this is claims management, subrogation. You will see some examples later. All of these three elements we put together, many things are already in place today. This is now the orchestration platform that we show. We will continue to invest in it and build blueprint layer for blueprint layer in the coming months and years. All of this is built essentially on TP's expertise. As you think about it, today, TP manages already in an existing business billions of interactions every year. We have 40 years' expertise with processes. We have thousands and thousands of experts on the tech side, but also on the industry side. This all feeds our TP.ai FAB, where we will go further to the market. We have in our next session a detailed look to guide you through how this is live today, where we further invest. This will be an integral part of our strategy in the coming years. Third message, do we have the right team? When I talk to many investors, they say, "Yes, you're so successful over the last 40 years." Do you have the right, not just the right resources and capabilities and clients, but the right team on board to do that? This is really, for those of you who have visited the TP side, who have interacted with the management, really part of the DNA. It is operational excellence to drive it. That will be a key component going forward. As Daniel said, we are very fortunate to have great talents in-house, but also bring on new talents. In today's presentation, we have four new faces that you might not have seen before. In the next session, we have Anish Mukha. Anish has had a very successful career at Genpact and Amazon, then joined us in 2022 as the CEO of India. India is today our largest market and really on the forefront on large-scale AI transformation. You just heard the latest award they got. Anish agreed to take on this new role as our Chief AI Officer to really bring this life for the group. Very happy to have him. The other person that is new you might have seen in the past and who's presenting together with Anish is Akash Pugalia. Akash is a former Accenture, then Cognizant. He joined us and built the trust and safety business. As you know, trust and safety was a new business for us, which is now over $1 billion in revenues. He left, unfortunately. I do not know why, but we brought him back at the end of last year and actually to drive TP AI data services. If you hear all about the company, Scala, et cetera, this is a huge opportunity for us. It's a market that's growing close to 30% a year. Akash and the team are laser-sharp focused on driving this. We're very happy also to announce an acquisition, a technology platform in that area today. Third new phase you will have later is, where is he? JC. JC is also a BPO veteran for over 20 years in the industry. He looks older than he is. He worked at IBM, was then the CEO of our Latin American business. As you know, it's a really very successful business across Latin America. He took on the role as CEO of Specialized Services to really not just support the growth, but even to ensure a tighter integration with our existing business process services business. The fourth new face you will see today, unfortunately, just by video, is Sherry Turpin. Sherry is the CEO of ZP, based out of Austin, who came through the acquisition we announced last November and closed in February and very successfully led that business. It is a super exciting business over the last years. We are very happy to have her on board. These people are not, of course, the whole team, but they are representative of the strength and management team that we have. We will continue to invest in the people, because having the best people on board will be critical to drive the execution. As I said, our business is not about finding something new, but executing on the strategy as well. I do believe all of them are very hungry to win in the coming years. Last message, as I said, how do we drive value? We are just the agents of our principles, which is you. We want to achieve higher top-line growth. We want to increase our margins. We want to deliver cash flow to you. All of that is part of our strategy. We're seeing essentially a doubling of our growth rate, post-transformation and expansion of our EBIT margin, as well as a continuous delivery of a significant cash flow in the years that will be used to invest in the business, to further deleverage the company, as well as return it to our shareholders. That's in a nutshell. You will see all of this in detail with use cases to bring it to life. Before we do that, I just want to tell you three stories, because our reason of existence is essentially our clients, how we have this not as a future strategy, but as a reality today already on the ground. The first example, how do we grow the core is a global electronics company. It's been a long-standing client. We have also Danny, where is he in the room? He's working quite closely with them. It's a great client example where we are able to drive by implementing every month, I would say every quarter, new technology to the client, the customer satisfaction up, actually the cost per interaction down with the client. The actual cost per interaction down. We increased our revenue with the client by offering new services, new solutions to the client, and broadening our scope and share of wallet. It's a global business. We continue to invest with them. You see here our real-time translation software, our customer copilots. It's a super case where we basically orchestrate our existing human delivery with AI and thereby expanded our revenues. The second example is a major U.S. bank. As I said, I talked about that we have really grown last year significantly in our back-office services. This is a great example. TP is a reference for the BFSI industry. We have more than 65,000 colleagues around the world supporting banks, insurance companies. We have more than 350 clients in that industry. Really, really deep know-how. We have Mamta Rodrigues, who worked in that space, presenting in more detail what we do with them. We are really proud to work with the largest banks in the world. Often we start with a front-office service, a customer service. This was here also the case. Over the years, we expanded line of business of line of business, now managing the entire end-to-end customer lifecycle, entering into fraud prevention. You can see how we really, if you look at on the client discussion when we do our client reviews over the years, added line of business of line of business. We have now 12 business lines, quadrupled the revenue. Really being the best strategic partner for these clients, delivering operational excellence, delivering innovation is the path for us to grow further. This is an element where we have, as I said, grown double-digit last year in back-office services. We continue to invest in that area. We are very, very proud of the work that we have done for this client and many other clients in that space. The last example is, and I know that many of you also asked for this, is data services. Why do we feel we have a right to win in that space? What do we want further? Obviously, any AI is only as good as the data that it uses. You see almost every week in the press news about hallucination, wrong data. This is an opportunity where AI does not replace work. It is even the wrong answer, where it creates new job opportunities, because you need human-reinforced learning to make the AI better, to check accuracy, safety. It is a whole new industry that is just about to start, really just about to start. If you look at reports, it is considered to be single-digit billion, but growing rapidly. If you see a future where AI will be ubiquitous, having humans checking the data, the results, and creating this reinforced learning between both parts will be absolutely critical. We are very fortunate to work with, as you know, some of the largest companies in the world. Many of them are obviously also active in that space. We do have a successful history in the trust and safety business, as you know. That is content moderation, data labeling to some extent. These clients approached us and said, "Can you help us also in this area?" This is interesting. You see some companies just focus on that one. When it comes to how do you have access to global talent, to specialized know-how, do you have the process know-how to do this, do you have the technology and the platform in place? We are actually very well positioned to do so. We want to further invest in this area to scale it further. This is a client that is one of the most important clients of TP, a long, long-standing client. They said, "Can you help us?" Within a month, we built this team. You see here we have 650. I think the number is now even higher, machine learning graders that are specialized talent to help to provide better outcomes for the AI. As I said, we are setting up dedicated teams to further accelerate this area. The acquisition of Agents Only, which is a crowdsourcing platform, focuses exactly on that space, how to have better access to global client, not as a full-time employee, but as a gig worker to drive AI data services. With this, I leave you now in the good hands before the summary of my colleagues. You will hear first from Anish and Akash how does TPI allow us to capture more opportunities in the AI value chain. Secondly, you will hear about Miranda, our Global Chief Client Officer, Mamta, our Global Head of BFSI, and Himadri, our global solution experts, how we drive growing the core and extending the core business process services business, so the first two pillars of the strategy. Then we have JC, Sherry, and Jeff Cordell, who is our CEO of Health Advocate, explaining how we do the same thing, growing and expanding the first two pillars of our strategy for our specialized services before we come to a wrap-up. Have fun, enjoy, ask tough questions, because that's why we're here for you today. With that, I hand over to Mike. Fantastic, clear vision on where we are today and where we're headed in the future. Anish will talk us through the how does TP bring all this together into an AI platform that solves complex, interesting problems for our clients. Anish. Can you hear me? I'm going to move this. Okay. AI has moved much further than the orchestration stage at TP. It is at the foundation of how we reimagine the operating model, how we deliver outcomes for our clients, and how do we drive growth. TP.ai FAB is a new fabric of agentic AI systems, outcome-driven copilots with customers at the center, driven by intelligent automation. It is embedded in everything that TP does now, from customer care to vertical-specific solutions like underwriting to finance operations and to internal functions like talent management. I'm Anish. I am head of AI and transformation at TP, been with TP for about three years, as Thomas said. Prior to TP, I was leading the customer experience or the shopper experience for Amazon across 12 of the largest marketplaces. I am very excited today to be part of TP's AI forays and contribute to the transformation journey of our clients. Unsurprisingly, today I am going to talk about our AI business and our AI Fab, which is a version of our enterprise-grade modular AI operating system. I am conscious in the next maybe 15 or 20 minutes, my conversation may get a bit more technical. I will give you a few business examples and the context as I go through the pages. TP and our ecosystem, our industry, our clients, our peer group is at the inflection point. Customer expectations for us are rising. Cost pressures are also real. 41%, 41% of our CX customers see GenAI as an incremental as well as transformational element of this strategy by 2026. As we look at it today, AI is good enough to rewire operations only if you know how to do it at scale. TP.ai FAB is our answer. There are four distinct trends that I want to talk about today that inform and lead TP's strategy. First of all, AI is at the top of every client's agenda, every client that we currently service, every client that we currently pursue. The client could vary from tens of billions of revenue in each marketplace that they operate in, each country that they operate in, and they could be in many countries, to a client which is less than a billion in revenue. Every such enterprise is prioritizing AI adoption. GenAI, which was once confined to the boardrooms or to innovation labs, is now moved from experimentation stage to the implementation stage at a very, very fast pace. Hundreds of our existing clients and potential clients value three things. First of all, AI has to be scalable. It has to be secure. It has to be based in real-world use cases. For instance, a healthcare client of TP enabled several service systems which are now beginning to handle claims and appointments with north of 90% accuracy and creating a huge amount of capacity within TP for higher value clear. Why it is important for us to know is because of that released capacity, we are going upstream and downstream and creating far more value for our clients and, in turn, creating far more value for ourselves. The future-ready workforce is currently a challenge. The challenge is determined because of the increased demand, which is determined by the hype that AI has created, a sort of a frenzy going on, also sort of expectations which are not mired in reality. At the same time, there are some valid opportunities that our clients want to pursue. Underlying this lack of capacity are two major reasons. One is the lack of real-world experience, the curated cases that are relevant to our clients, and lack of industry domain. That is what our differentiation is. We have decades of experience across major verticals where we control the outcomes, we deliver the outcomes, and we can bring it to the client's benefit. That is our competitive edge. We bring deep domain and transformation mindset. Coupled with the horizontal solutions that you see all around us, it is an unbeatable combination at our scale. Important to note, the mushrooming of AI companies that you see all around us, they are right now even still more focused on the horizontal solutions. I'll give you one more example of that later in the slide. Secondly, clients want end-to-end scalable solution, not a point solution, ownership of customer outcomes. They have moved beyond asking a question that, "What can AI do for them?" to the next level of granular question is, "How can TP embed those in their enterprise operations and do it sustainably and do it with scale?" Customer expectations have changed. Thomas earlier alluded to omnichannel. They expect seamless, personalized, and omnichannel experiences. Only when it is applied thoughtfully with agentic AI solutions and human experts working dynamically together and self-learning. In conclusion, TP isn't just riding the wave. There is a big wave going on, and we would love to ride it. It is not just about riding the wave. We are reengineering from inside out. We have the deep domain capability of more than 40 years of deep domain expertise and transformation roadmap that we can bring to our client's benefit. That's what we are doing now. It's all integrated in our TP.ai FAB, which is now ready. Let me bring this to reality. The ask of those clients existing as well as the potential clients is causing TP to unfold the future of CX with both domain services and agentic AI capabilities. Fab's TP orchestration platform, which is modular and AI-powered, weaves in agentic AI experts into vertical-specific solutions. There are four key tenets of Fab that I would want you to register. First is fairly intuitive. It's structured in three modular layers. At the top is the blueprint layer. In the middle is the AI orchestration layer. At the bottom is the foundation layer. The nature of these layers allows us the flexibility, speed, and use case orientation. This structure is designed to scale across clients, even though you might build an AI-curated case which is contextual only to a client. A large part of that is portable. It is portable across clients. It is also portable across different industries, different verticals, because it is a combination of different horizontal services. Eventually, you can do it in a region. You can do it in the second region. It is portable across geographies as well. Each layer is there for a reason. It adds unique value, whether it is foundation layer or the blueprint layer that I am going to talk about in the next charts. In the blueprint layer, this is where Fab for us becomes a product. These are outcome-driven use cases for our clients and TP's most compelling transformation differentiator. These are also the processes that our clients very intimately relate to. TP creates the AI vision which is embedded in our vertical knowledge. We redesign the CX, and we improve fulfillment of end-to-end processes. I'll give you a few examples on what we mean by the vertical knowledge here. If you look at financial services, from account servicing to customer acquisition to anti-money laundering, risk services in financial services, claims, customer onboarding, policy issuance, and insurance for horizontal services as well, for closing, accurate closing and timely closing of books, filing of 10-K and 10-Qs, automation of manual journal entries in finance and accounting processes, and reducing the defect, improving the shopper's experience for some of the world's largest online retailers. Those are the use cases that we are talking about. You can imagine the blueprint as the pre-orchestrated package which combines AI, expert, and logic together, which is tailored to a real use case. You can also visualize them as pre-configurable small blocks of small language models or SLMs, which can either be sequentially deployed or in parallel in combination with human expert and with agentic AI. They include orchestration logic, data flows, and domain-specific tools. That is how TP is moving from capabilities to solutions. Simply said, if I were to explain it a little bit less of technical terms for our clients, it is an end-to-end process redesigned with envisioned AI tools and human experts which are dynamically working together and self-learning, self-serving. At the bottom of the page, there is an example where our horizontal solutions and vertical capabilities come together to create state-of-the-art banking and financial services process that Mamta and Himadri will talk about in the next session. AI orchestration layer is our heart of Fab. Every task in Fab starts with AI. That's the default. TP at this layer combines AI agents and human experts and lets the system decide. The system decides the following: who should handle the task, what logic should apply, and when should we escalate. This Fab logic also incorporates outcome orchestration, task allocation, real-time learning for the machine, knowledge, and the feedback loops. Even if the task in the question goes to an expert, AI doesn't disappear from the scene. It sits by the side, learns, provides live prompts, sits on the top of the knowledge base, suggests the next best action, which the agent can then communicate to the customer. In this example on where the interaction went actually to the expert, we are talking about AI making the expert more efficient and more effective. At the same time, the machine is learning. AI is learning. The expert is making the machine learn. Therefore, AI will become better over a period of time. This orchestration layer is not rule-based. It is dynamic. It is based on the complexity of the use case. It is based on the customer intent. Underlying this dynamic process is our probabilistic model that supports the decision tree. The best part, as I said, is it gets smarter over a period of time. Finally, the strength of this orchestration layer lies in the modularity of it. There are proprietary tools. There are TP tools that can be selected or deselected depending on what the customer wants us to do, the amount of ownership that we have of the process. Due to the demonstration of improvement of customer outcome that the client wants to achieve, they very easily allow us to go in an accelerated way to upstream, downstream. What that causes is even greater impact on customer outcome and a greater impact on the value share that TP derives from that engagement. This is how TP will build, scale AI that works. Finally, the foundation layer. It is of utmost importance to have a scalable, secure, and globally accepted ecosystem of hyperscalers. The foundation layer is where the trust and the scale begins, not just for our clients, but for TP as well. Fab is built on enterprise-grade architecture, cloud-first, and LLM orchestration and model-agnostic foundation. Why is model-agnostic important? There are new clever models, LLM models that are coming up, which are also more efficient. If we are model-agnostic, it gives us a unique ability to be future adaptive and bring those into our fold. The core AI skills of reasoning, summarization, classification, and automation are built into this layer. You might imagine and question that, why is this important? Because this is something our business clients may be oblivious to. This is where the most importantly secure data flow resides. Observability happens. Prompt orchestration takes place. LLM integration happens. Foundation gives Fab its ability to scale securely. Now, let me bring the Fab chart again with all the three layers defined, with all its critical parts displayed. This is Fab under the hood for you. Let me try to make it real with the help of a Lego analogy. Imagine that the task for us is to build a Lego car. We have a team of perhaps smart robots who can do that. Imagine each Lego brick to represent a small AI capability. One could be of language. One could be of vision. One could be of decision-making. One could be of sequence. A car is envisioned on how will it function, what parts will come at what time, and what different components needed to be added. That design phase of how the car should function using these Lego bricks happens at the blueprint layer. In the past, humans had to pack and place each brick separately, carefully planning. In the AI orchestration layer, there is a dynamic handoff happening between those robots that I alluded to and the human experts. Now, between both of them, they'll pick the right bricks. Bricks here are the AI tools and APIs. They'll decide the right order, which is the sequencing or the workflow which is embedded in our framework. They'll make the adjustments, which is learn and adapt. They'll deliver a working car. You might imagine that this is completely done by robots. It is actually not. Robots, even today, in any of the use cases that you'll see around us and for a long while, will not have the capability to do end-to-end. If I were to extend this, robots may not know how to plant a steering wheel, how to attach the wheel so that the car could move. That human supervision layer will continue to become more and more important. The result is a fast, flexible intelligence system that does not just reside in the predefined task. Foundation layer is a little difficult to explain in this analogy, but I will try anyway, because that is where it scales. If you were to extend this analogy a little bit more crudely, you would imagine the safety policy, the purpose of the car, and even the highway infrastructure would become part of this analogy because it ultimately allows our analogy to scale and the car to function. Key points I would want you to take away from this structure: our Fab structure is built on the bedrock of the best hyperscalers that you can find in the world today: Microsoft, Amazon, Google. It is LLM agnostic. It drives scalability. It drives security. Intelligence layer of our AI orchestration layer is built with our proprietary tools. It is built in partnership with our AI solutions. Orchestration is owned by TP. It is modular. It does not need to be bought all at once. You can buy part of it. You can buy an additional part of it later. It gives our clients a unique ability to work with us and start small and eventually grow. Deep vertical domain expertise that is responsible for customer outcome design with embedding AI and the blueprint layer. That is our main differentiator. With this, we firmly believe as a unit that TP will be the industry leader. We will be modular in our solutions. We'll work with our partners building an open ecosystem. Fab's—I'm sorry. Fab's intelligence layer actually is not a clean state rebuild, because you might imagine suddenly we have woken up and we're starting to build something new. It's built on what's already proven. It's built on our decades of experience across the major verticals. On the left, you see mention of a few tools: TP Interact, TP Protect, Power Steering, etc. Those have been in existence for years. And just in 2024, we have done more than 200 implementations. However, in today's rapidly changing AI landscape, partnerships are critical. They are a strategic imperative for us. They enable faster innovation. They allow us access to specialized services and the flexibility to scale without being locked into a single technology stack. I have a few names here. Sonos helps us enhance communication clarity globally through real-time voice translation. TP is the exclusive reseller of speech tech for many of our largest client companies across the world. Pablo Alfares brings agentic AI multilingual voice expertise that complements our already huge scale on voice-heavy interactions. We have Malte here somewhere. He is a CEO and co-founder. He would love to spend some more time later in the session with you talking through what he brings on the table and what the partnership is doing for our clients. Our collaboration with Emma accelerates Copilot deployment across enterprise operations. We have Surajit Chatterjee here. He would love to talk about his case studies when we get to that in a minute. We are extremely proud and excited about what this partnership ecosystem can do for us. What you see on the slide are just a few. We have already earmarked EUR 100 million for investing in those partnerships. Finally, not stealing Akash's thunder, he's going to talk about that Agents Only. We are very excited and proud to present Agents Only, which is a flexible and gig-enabled workforce model that can be intelligently matched to clients' demand using AI. We are only unfolding the monumental opportunity that presents on this. Together, these partners that are already in our fold and many more that are going to come in the next few months, these partners extend our capabilities far beyond what a single platform can deliver for us. They allow us to remain modular, adaptive, and fast-moving when we focus our internal investments to what differentiates TP the most in the blueprint layer and in the orchestration layer. This allows TP also to become more resilient, more scalable, and better positioned to win in the AI-powered customer experience economy. Finally, Fab capabilities across these three layers enable us to go faster and deeper into new markets. That is something that Thomas spoke about in his presentation earlier. As these three layers and AI in general reshape how the industry is going to change, reshape, and how we are going to strategically expand into our high-growth adjacencies: digital marketing, consulting, technology, data services that together represent north of $600 billion total addressable market. Those are the markets that are present to us as new, new markets. Those are strategically significant opportunities for us. What sets TP apart is that we are not entering these markets as newcomers. We are entering into this market with Fab as the foundation and our real execution expertise and transformation playbooks at scale. We are bringing that to our clients' benefit. In doing so, we move up the value chain, owning not just the execution but the end-to-end customer outcome, therefore expanding the value share that we get from them. I'll give you a couple of points. For example, in digital marketing, we are using customer consumer intelligence enabled by AI to drive engagement, build brand equity, and power growth for us. In consulting, we are using future-ready transformation models which are grounded in AI feasibility. We propose what's possible for our clients. In many of these instances, our clients trust us to actually execute these transformation roadmaps. Whenever we do, we transform those client operations, obviously impact their customer experience, outcome, reduce cost, bring scale to them. In turn, we do far more business with them expanding our value share. In technology, we deliver GenAI services, GenAI-infused IT services, and cyber-resilient infrastructure which is built for scale and speed. Finally, in data services, we enable everything from AI model training, descriptive, diagnostic, predictive, and prescriptive analytics that soon Akash will talk about. Thank you. Am I audible? Yes. Hi, everyone. Good morning. My name is Akash. I've been with TP for about five years, in the industry for about 25 years. Very fortunate to have built the trust and safety practice here at TP, which is a little over EUR 1 billion now. Very excited to talk about the new journey that we are taking in the data services space. Data services, when we think about AI, there are three things that are required in AI: energy, chips, and data. Data is actually the foundational fuel of AI. If you think about any large language model, if the large language model does not have curated, high-quality, diverse data, the outcomes that you will get will not be appropriate for what you're looking. From a TP standpoint, we are very excited to share this new strategy with you. From a data services standpoint, TP has the end-to-end data services available for our clients. What does that mean? I think when we talk about data, a lot of us understand that data is nothing but data annotation and labeling. While that is one of the major parts of data, we also have data analytics, which is huge, as well as data engineering/data management. We serve end-to-end all our clients. What clients need here is a provider which can provide high-quality, modular, scalable, secure data for their enterprise. All the companies beyond technology companies will soon look at service providers like us who have the understanding and experience in delivering data services. What's happening in the industry? If you think about the last one year or one and a half years since the time ChatGPT has come into play, a lot of us, our searches have changed. We no longer now generally go to Google as the first place to search. We do ChatGPT. If you look at the last one year, there have been 48 billion searches on ChatGPT alone. It's growing at 80% plus. While we say that we don't Google it, Google also has their ChatGPT version, which is Gemini, which also has been enabled within searches. That's happening. Why is that important? Again, everything that comes out of the search is data. We are consuming data. That is where the entire piece comes together. The other thing is that the AI ecosystem is expanding. We are not only talking about industries, which is tech, but across different industries. When we look at asking questions from ChatGPT, we are asking various questions, be it an academic question, be it a travel question, and so on. It is going through different industry segments. It is going through different languages and modalities. Data is the foundational fuel. This is actually driving not only just our large language models, but think about autonomous vehicles. Think about robotics. All of that requires high-quality curated data and secure data. That is why TP is well-positioned to deliver because of our great strength in terms of our vertical expertise across the different domains, our great expertise across getting the different languages, and the context awareness that is required in data services. Now, when I talk about TP.AI data services, it's an end-to-end platform that is available for our clients. What I mean by that is that it does not only span horizontally across all the industry segments, but it also looks at the entire vertical AI chain, AI spectrum. We do not only look at consulting for AI implementation on data for our clients, but we go through the full end-to-end cycle where we build, operationalize, as well as govern it. We also look at the different modalities in terms of text, images, audio, video, etc. That is why clients prefer us, because they get scalable, on-demand, curated, high-quality data. They get it with the highly specialized skill resources that we are able to bring in an absolutely agile and adaptable way. That is where everything comes together. When we look at the different areas which are growing, if you think about large language models, if you think about computer vision, if you think about physical AI, which has been the new thing that has come in, that is where data is required. Physical AI, just so you guys know, is nothing but where you have drones, you have robotics, you have autonomous vehicles coming together and gathering information. All of that information is data. That is where this is happening. How we help clients and why clients value us. Obviously, we are able to provide data as a service, not just curated data, but high-quality, scalable data for them. We are able to provide human-in-the-loop scenario. I think there's another thing that we always think about is that this is a one-time exercise. It is not. It is an ongoing exercise that will continue because the model needs fine-tuning, and the model needs to adapt to the changes in the industry that's happening. Human-in-the-loop is extremely important. The other piece that is important that we bring together to clients is our trust and safety expertise, which means bias mitigation, which means hallucination checks, which means AI not generating its own set of facts which are not correct. That also is something that we are able to bring together. The last thing that we are able to bring together is the understanding of all the laws and the regulatory environment and the regulatory framework that's coming across the different countries. Again, this helps us because we are across 100 different countries and our understanding of the country and the regulation. The other piece that they love about TP is the flexible, highly specialized skill resources at scale that we are able to provide. Now, let me spend 30 seconds on this. When I talk about highly specialized skill resources, what are these? These are mathematicians, statisticians, doctors, lawyers, opticians, pharmacists, radiologists, and so on and so forth. When we talk to ChatGPT, we ask any question. ChatGPT is just an example. There are many more large language models. We ask any question that may be there. We can ask a medical question. We can ask a mathematics question. It has to provide an answer. The point being, when you look at the answer, those answers need to be accurately looked at by someone as a human who's an expert in that space. We are able to provide that at global reach, at scale, and at a cost that is really attractive to our clients. That is why TP is at the right place for our clients. Now, this is not new. I want to emphasize on that. This has been going on for the last year or so where we are growing this practice significantly. Thomas touched upon this earlier. For one of our largest global tech clients, we started with data annotation and labeling, and it has grown tenfold. We have over 800 to about 900 people doing this work. We've moved from doing one-time data labeling exercises to actually changing the brief over a period of time where we are looking at LLM. We are looking at LLM summarization. We are looking at prompt quality and so on and so forth. We are going up the value chain as well in terms of managing this work. This shows that we do not only understand what the client wants. We are able to provide it at speed and agility. Before I move on to my last slide, I also want to mention the way this is getting delivered across the AI spectrum. There are three ways that the AI companies are reaching out. One, through managed services providers like us. The second way is looking at crowdsourcing providers. The third is pure-play platforms. What we are doing is we have a three-pronged strategy. Obviously, we are really good at providing managed services at scale at a cost where the client wants. We are also partnering with platforms where we can be their managed services provider and help them in their AI journey. We are also moving into the crowdsourcing space where we are able to find these specialized skill resources who do not work full time, but they are able to provide their expertise through a gig space. This is why I'm really proud to announce the AgentsOnly acquisition here today. AgentsOnly is an AI-enabled crowdsourcing platform that connects global brands with curated on-demand workforce, highly skilled workforce. We are able to provide it with speed and the quality that is required. We are able to provide it at a scale that our clients expect. We are able to provide it at the flexibility and the security that they need. In summary, while AI is really changing and disrupting industries, TP is well-positioned because of our global scale, our human-in-the-loop network, our highly specialized skill resources, platforms like TP.ai FAB and AgentsOnly. We are not just looking at AI as a partner here. We are actually creating the infrastructure layer for AI. Very excited for you to experience data services with us and handing it over back to Anish. I think the first question that you'll wonder about is, why now? And why are we supposed to win in this market? That's underpinned by three core messages that I want to talk about. First, TP comes from a core practitioner DNA. Unlike a lot of advisory firms that you see around, we build and deliver where AI meets the real world. Our solutions are grounded, agile, and also impact-driven. We have done 1,500 installations, so to say, across 700 clients already. Secondly, we have scale-ready talent, which is north of 3,000 specialists. Those specialists have solution architects, software engineers, data experts who are already working through all our existing clients, as well as a huge pipeline of opportunity that we see. We have over 10,000 supervisors who are getting trained on AI and transformation as we speak. That will increase our reach to fourfold to where we stand today. Finally, we have operationalized TP.ai FAB. That's our backbone that integrates data models and copilots across functions. It's a scalable way to embed AI into every transformation play that we make. These aren't just future ambitions for us. It's happening. It has happened in our context already. We have built a social media hub for a beverage player, resulting in 900,000 new followers, 850 million new impressions. For a global airline, 15% productivity. For a U.K. logistics player, where we built digital engagement platforms and so on and so forth. Akash has already spoken about the digital case. This is what makes TP credible. That we do not just advise on transformation. We deliver it, and we power it by AI. I'm not going to elaborate now on these three to give you a little bit more context on what actually it means beyond just that specific case and why is that case important. First, on the technology side, TP partnered with a U.K. logistics carrier to deliver a full-scale AI-led transformation that resulted in a 15% increase in agent productivity and $2 million productivity gains. We used TP.ai FAB as the backbone, deploying GenAI-powered interaction analytics for 100% of the customer queries with real-time agent assist and next-best action models. The solution also included Genesys CCaaS integration, AI chatbots for improved self-service, and accent transformation technologies. This is a full suite of Fab that I earlier spoke about at display from blueprinting layer to the foundational layer. Therefore, why is this important? Why is the single case important? Because it demonstrates TP's ability to combine deep domain expertise with advanced AI and tech, and we deploy it with unmatched velocity. Earlier in the day, I was speaking to one of the investors, and we were speaking about the need of being agile in today's world because solutions will come. The differentiator lies in how can you quickly scale it up faster than anybody else, and how can you determine the early bets that will work? At the same time, detect early what will not work so you're not spending a lot of energies chasing the dreams that will not happen. We have demonstrated these applications in complex, really complex service environments. The second case is about our transformation play for a leading global airline for consolidation of its contact center operations. That delivered north of $10 million of annualized savings. We leveraged our deep domain expertise on the airline domain knowledge with CX design and AI capabilities, conducting demand and supply analysis, deploying AI simulation models across, completely redesigning the target operating model, resulting in a 50% reduction on the onshore centers that they have, and almost 60% of the workforce moved into a remote model. This case underscores our ability to lead strategic AI-informed consulting assignments that can optimize cost, elevate experience, and drive sustainable value for our clients and for TP. In this final case on digital marketing, we partnered with a global consumer products company, one of the largest in the world, to design, build, and operate a centralized social media hub. It transformed how the brand interacts with their customers across the platforms, an omnichannel communication engine that enabled real-time AI-informed engagement across social, chat, and digital touchpoints. This drove exponential growth in digital reach. Their impressions increased from 1 million to 2.5 million, resulting in a disproportionate increase in revenue on the TP side as well. This program demonstrates our ability to deliver marketing transformation, blending consumer intelligence, experience design, and engagement to drive growth, brand equity, and loyalty. My three closing comments before I hand back to Mike. TP.ai FAB is an industrial engine that's designed to scale. It brings the best of TP's domain knowledge and cutting-edge horizontal solutions. Secondly, we will accelerate the deployment and be the industry leaders with best-of-the-breed tech with the help of hyperscalers. Finally, this allows us to expand our total addressable market dramatically. It allows us to get into margin accretive accelerated growth. This will become a self-sustaining flywheel for TP. Thank you. There is a lot of rich content there. Just a couple of points that strike me as I think about what we heard. One, all of our clients are talking about AI. It is at the top of their agenda. Fortunately, they are all talking to TP about how we can help them through that journey. TP Fab collected all of our best practices, knowledge, our deep domain expertise to allow us to address those needs in a systemic, repeatable way that does not require re-engineering each time. As we generate value for our clients, they offer us the opportunity to solve new problems with them, opening new lines of business, growing new areas of work. As Akash mentioned, this opens completely new markets in things like annotation and labeling and actually helping build AI systems for some of the world's largest companies. Truly an incredible opportunity that we face into and one that gives us all a lot of energy. Up next, we'll take a look at our BPS, our core business process services, and how we're evolving that core with these new tools capabilities. Last, we'll cover the vertical expertise and how we're looking at the entire value chain of a particular industry, where we can play, where we can help solve those problems. I will also introduce Miranda, our Global Chief Client Officer, to the stage. Thank you, Miranda. Check, check. Good morning. Thanks for being here today. It's nice to see some familiar faces, as Mike said. I'm Miranda Collard. I am the Global Chief Client Officer. I, too, have a very distinguished career. I've been at TP for 31 years now, starting as a TP frontline expert. Ever since, I've been passionately serving both of our clients and our employees. Happy to be here. I have the distinct pleasure of covering for everybody today a couple of pillars that Thomas mentioned earlier, which is our global business processes, the core growth with AI, as well as the verticalization. First and foremost, I have great news. The market's growing again. This is very, very good for TP. We have significant opportunities across the board. I'll start first with some of the key market trends that we're seeing. First and foremost is the AI and the humans. Really, what is the top of the agenda for all of our client partners and prospects is the organizations are certainly embracing the hybrid model, which really I like to define as an enablement in the terms of we need solutions to bring forward that help the frontline with a greater level of emotional intelligence to really meet the customers where they need to be met and that our partners get the outcomes that they're looking for. Second is the partner consolidation. Any near onshore delivery opportunities are certainly accelerating as they do in these times. These are not new objectives inside of our business. They're just being met differently with the wave of AI and all the rich architecture you heard from my colleagues, Anish and Akash. Last but not least, AI is driving demand for that integrated customer experience platforms. We need to meet with the technology and the solutioning that comes forward and allows a roadmap for that enablement that I explained before. Accelerating the growth. First, the first pillar Thomas shared is to grow the core enabled by AI. How we're doing that? We're going to share a couple of examples as we move forward of what that looks like from our rich and deep expertise. First, our strategy, of course, is to use that tech. It's our responsibility within the partner network to bring forward all of this incredible capability and solutioning that you heard today. You can see that it's deep and rich. For people like me inside the organization, it's exciting because this is where the growth comes from. There's an example of this, of course, in the banking financial services and insurance group that you'll hear from Mamta here in a bit. Embedding that AI across all of our core services, we enable with the TP.ai FAB, we accelerate the transition of in-house and capitalize on that adoption. This is why you'll hear also that the verticalization is so critical for us. We have a multitude of verticals. By the way, I could talk to you about them all day long. We're highlighting BFSI here. Just to give you a small example of another vertical with the retail and e-commerce, since we're all consumers here, we can all relate to personalize that online journey across the retail and the e-commerce space. It reduces the churn that all of us have experienced as consumers and in the business. It really optimizes those business outcomes. You think about the retail space and the retention of the customer, bringing the customer back, having the repeat purchases, and all of those is extremely important for our client base. We will get into some of the vertical plans in the extended space. This is one of the most exciting things that we get to do in the solutioning group here at TP. We can now expand inside of those verticalization that, and you will hear some examples of that. We build those industry-specific end-to-end AI-enabled offerings based off the very deep expertise that we have in our leadership teams. In BFSI, and I think everybody probably has something to say about BFSI in this space, but if you think about the amount of workflows, the amount of domain experience that you really have to have a knowledge of to transform, this is a key critical component for us. Having a full understanding of the end-to-end, the front-end customer journey, the back-end processing, and all of the thousands of workflows in between is a significant opportunity for TP. From the retail e-commerce space, we can co-develop now those solutions. If you think about demand forecasting, when are those interactions going to come in? When can we automate? How can we capitalize on it? Because as you think about the mundane automation of repetitive workflows, what's going to happen and needs to continue to happen is when the customer needs to speak to you, they need to speak to you. Demand forecasting is critical. Then, of course, like inventory optimization, automated returns processing, all of those things that allow us to take friction out so that now with our emotional intelligence training and program, you can really show up in those ways that you need to and the customer needs to hear from you. We are going to, I know that everybody's asked to speak to our clients over the years. We have invited three of our esteemed partners here to join us in a panel conversation that will be moderated by the wonderful Mamta Rodrigues. I'd li ke to invite you up, my dear friends. I'll allow them to introduce themselves. Okay. Hi, everyone. Thanks, Miranda, for that amazing introduction. I'm Mamta Rodrigues, and I run banking, financial services, and insurance. I have the easiest or the hardest job today because you guys are going to have to help me. Keeping these four quiet about this space is going to be interesting. I'm also going to play time cop a bit. We've got 20 minutes. We're going to start with introductions. We're just going to start here. We'll go that way. Paul? Sounds great. Good morning. Happy to be here today. I'm Paul Vincent. I am a Senior Advisor with Boston Consulting Group. I have been doing that job for about 45 days, so I know it all. I'm on panels, so I'm excited. I also have 25 years of banking experience before that, both at Capital One an Information-Based Strategy Company in the financial services industry and USAA, the leader in customer experience, in roles that have spanned operations, channels, product strategy, transformation, and eventually being president of the bank at USAA. Good morning. I'm Lance Grinner. Also very, very happy to be here. Thank you very much. I've been in the industry well over 30 years. Most recently, I have been with Mastercard as the Executive Vice President of Global Customer Care, responsible for service and support in 210 countries and 74 different languages. Hi. I'm Lisa Stoner. I'm currently the Vice President of Support Operations for Pinterest. Previously, I was a buyer of services at TP when I was the Vice President of Product and Support Operations at Meta, as the Vice President of Global Support at Meta, and as the Head of Global Support for Uber. Okay. Let's jump in. The first question, Paul, I'm going to start with you. The environment for AI has changed radically. With your multitude of years of experience, you've seen a lot of the evolution of not only AI, but a lot of technology that's come underway. As you look at AI and disruption in operations, what do you see the change in customer expectations in the omnichannel and multiple channels? What do you see as the change now? What do you foresee coming forward? I don't know if I like starting with the multiple decades of experience. That'll age you pretty quick. It's interesting. We use the word unprecedented so often right now. I think it's losing its actual meaning. We are in a significant disruption from a cycle perspective. It starts with the consumer. If you think through the average consumer for financial services, you look at digital commerce, which has doubled in the last five years. You look at embedded finance, your ability to pay while you're in that transaction with debit or credit or alternative forms of payment. That's up 5x in that same period of time. We are just getting to things like agentic commerce, where you'll have an agent that is actually shopping for you. If you watch PayPal or Visa right now, they're introducing agents for the first time. You'll give them your parameters, what you want to look for, the value you want to pay for it. It will bring back the best options for you. You have a wave of change coming from the consumer, which ultimately means we have to change as well as financial services companies. We are looking at AI and how it plays a role in the entire value chain, ranging from targeted marketing, underwriting, servicing, collections, you name it. There are roles that will play to make that experience more efficient, more effective. By the way, it is not a one-size-fits-all. You have a digitally native group that has now grown up with the phone, the younger generation. You also have those moments that matter, those highly emotive experiences that Daniel talked about, where they still want to talk to a person, want to go through that, want to know that there is someone they trust on the other side that is there to help them. We're dealing with needing to handle the broad needs of different groups in a vastly changing environment. You know, if I could add, I think not only that, but adding on, you've got the speed of change that's occurring today. You have post-pandemic customers' expectations have significantly changed. You have the need of customers or companies to anticipate where the customers are. The need of the companies to provide services has changed. That change is now becoming so the speed at which it changes is the need to use AI is ever more important, the ability to have more data at your fingertips, the ability to anticipate customers' needs, and physically what that change that is necessary to compete in today's market. Anything you want to add, Lisa or Miranda? Are you guys? No. Good. Okay. What I heard from you guys, that's a first for Miranda. I was told to be sure. I was just kidding. No. What I heard from you guys, though, which is paramount to also what we heard from Akash earlier, is the power of data. And we're just getting started. While in the space of payments, one can say we've had data and the data lake for a while. The monumental pivot now with AI and what I heard from both of you and the leapfrog effect that we're going to see, especially when you double down on that customer leading the way, the decisioning coming from the customer is phenomenal. Second question, and I'm going to start with Lisa on this one. Lisa, you've worked with TP on a multitude of opportunities that you've been at in your years. Where do you see, where have you seen partners like TP, not only TP, working and partnering with you, especially as it relates to integration and ability to be that leader in operations and with AI? Yeah, it's interesting. I actually started as a competitor to TP. I worked in this space my whole entire life, like many people here. My first time buying from TP was when I went to Uber, and it was pre-IPO during a time of just incredible scale. Right? We were in the infancy of the AI revolution then, too. It just wasn't sort of as top of mind as it is now. At the time, TP was just a leader in just enormous change for us. Think about pre-IPO at Uber, think about post-IPO, think about the COVID years, the need to vastly expand in Eats business. Right? TP was just incredibly well-positioned based upon the global footprint, nimbleness, the ability, the huge range of services, the ability to provide things like local languages in India, all the way to highly advanced enterprise support. Fast forward to Meta, and I had the privilege of working there during the year of efficiency. Think about, again, a time of a tech-enabled company going through vast change. Again, Teleperformance was a leader. Today, I'm at Pinterest. Again, significant change. Just to contextualize Pinterest for people, it's a company where AI is a core competency. It's at the heart of our curation signal. As pinners, people using the platform are, as an example, searching for outfits or looking for things to buy in the e-commerce world, the curation signal is key for Pinterest. AI is already making advertisers more effective and more performant. That's a huge change in the ad space. It's making our employees more efficient. To put a number against it, our CEO announced in the last earnings call that 25% of code was written by AI. That's an enormous change. We're a platform that has inclusive AI, and there's an enormous demand for that. Think about being able to search by body type or skin color or hair texture. We found that search is enabled at 75% faster with inclusive search. It's real and can be quantified. TP is an important partner for us, ranging anywhere from the trust and safety space to working with pinners to working, again, with a wide range of SMB to enterprise advertisers. TP is an important partner and a critical part of my personal and the company strategy. Thanks, Lisa. Anything you want to add? I'll just double down on something Lisa said. AI is a core competency. And you've heard a good bit of that today. Yeah, I think when you look at AI right now, you've got to make sure it's in the tone at the top, in the strategy. You've got early wins and use cases. You heard a lot of that today. But really working means it's embedded throughout your organization. And so those that embed this and get your frontline practitioners, those that really know the work, are fluent in AI, they're going to be the winners. You know, and I have to say, everything that you've heard this morning about TP, the strategy is in play in reality. Because over the last several years, when you look at how, as a company, we've had to change because our products have become complex, the expectations of our customers have changed. The ability to partner with a company such as TP, being able to go along the way and being there for us with the scale, with the information, has been vital. I just think the range of capabilities as well. Right? Both human and technical capabilities. Our needs are changing. I'll just give a very specific example in the support space. Right? Previously, you had, I think RPA was mentioned. You previously had various different, like IVR was a million years ago, or declarative chatbots. Now we're in the age of GenAI chatbots. Right? They can take on the scripted work. We are using that as an opportunity to invest in higher value work. Our need is for people with greater skill, more language capability, more personalized interactions, again, indexing toward enterprise and the monetization level of the platform. TP is well-positioned in that space to be helpful. Yeah, that's probably my favorite thing to talk about with clients, not just all of us here, but just generally speaking. Probably just between this panel, we've all evaluated tens of thousands of interactions across the space that you're in, previous roles, obviously I have across all of the group. One thing that always remains relevant for me in those interactions is that, and because I've been a TP expert, I've sat at that front line. I've taken the call. I've made the call. I've done all those things, is that we are in such a ripe opportunity to have that enablement factor for the front line. It's really unseen friction that behind the scenes, just for the training, just for the nesting, just in the preparation to handle that interaction, it's monumental now what we can do for that enablement. Like Lisa said, what happens in so doing is now you have a rich need for a high-value interaction and experience happening that requires a level of emotional intelligence that we have yet to see. It has been the constant evaluation of what we've seen now for decades as automation has come in, chat has come in, or other omnichannels, is that this has been something that has been transforming over the decades and now is just on its next wave of transformation. What happens is that this creates a rich experience for consumers and also for the employee, for the next person that's coming in and showing up. We can enable them in a way that we've never been able to before to be successful and have exactly what they need at that experience level. I think that's what you guys also highlighted. It was just phenomenal to what Miranda just wrapped up with. They were talking about this at breakfast so passionately, is you're investing in the talent. You're investing in the career. The number of times I'm in front of clients in a good way, and we ask the question first, how does this help the expert? How do we make an impact to the economy that we're contributing to? How do we make an impact to the customer experience at the end of the day? Because people matter. With the last question, I'm going to start with you, Lance. We touched upon it a little bit, but let's bring it back. There is a lot. There are a lot of partners. There is a lot of disruption in the industry in a good way. How do you see partners like TP working with you as you go down the journey of leveraging AI in your operations? How do you see that partnership working generally as we move forward? What you heard a lot about the strategy as a company is something, as a partner, I look for. It is something that is really, really important because not only is it just about putting somebody or a head in a seat, it really is about partnering and understanding my business, providing me the data that I do not know today, allowing me the functionality to learn more about my customer, and being closer to TP with a company that has that strategy that says it is not just about answering the phone. It really is about getting a deeper understanding of your business, what your customers want and need, and to be able to solve that need. That today is the secret sauce for companies. It really is, how do I partner? How do I use the data? And how do I use information and technology to better arm everyone involved? I'll just add, and you said this, it was not just the need to get to know me as a customer of Teleperformance, but it was my customer at the end. Coming from an organization with a tremendous military affinity, that went as far as hiring veterans. It went as far as hiring their spouses, training and onboarding military acumen as you brought new hires into the association. Just those were the things I was doing and I was looking for in the people that would partner with my organization. I think one of the things that stood out. I think the other thing, I'll say that I was a support specialist. Right? I was an expert in the area of support, but a generalist being on the outsourcing side. When I moved to the client side, I really needed to find partners who were experts in the industries where I worked. Right? At first, it was rideshare, which was category creation. Moving on to both Meta and Pinterest, our platforms that are monetized through advertising. TP's got deep vertical expertise. I think that's a differentiator, especially when you're thinking about the global scale of the kind of tech companies where I've made my most recent career. Great. That wraps up our panel. Any lasting thoughts or comments? Otherwise, we're going to hand it over to Himadri. Who's next? No, thanks for having us. Thank you. Thank you. That was great. Wonderful. Thank you, Miranda. You're welcome. Good morning, everyone. Am I audible? Thank you. My name is Himadri Sarkar. I'm very grateful and fortunate to be playing the role of a Chief Solutions Officer in an era where truly AI is getting power enabled by artificial intelligence, as Daniel mentioned in his opening speech today. The way we are doing this is those three strategic objectives that Thomas included in his section. What I'm going to do over the next few minutes is double down on two of those three strategic objectives. Now, if you think about accelerating the core with AI solutions and then aligning those AI solutions to vertical value streams, from a solutioning perspective, I feel that's extremely important for three key reasons. The first, we need to protect our existing portfolio of customers. We need to grow the count of new customers. Additionally, we also need to improve the revenue per customer. For this discussion, what I've done is I've shortlisted four themes, two for each of the strategic objectives that I'm going to talk about. What we have realized over the past few years working with clients is that early enough in the process, when we are building the opportunity roadmap and creating a very clear, articulate business case, we would need explicit buy-in from our clients. They need to be a part of the process. The way we do this is through a collaborative design thinking process, which I'm going to talk about in greater detail as I move forward. That's just the first part of the story. Just creating a business case and an opportunity roadmap is not really going to cut it unless there is also adoption of those AI solutions by our clients. The way we do that is we embed and assemble those AI solutions in our client's tech ecosystem. I'll give you a few examples on that too. If you were thinking, that's it, that's not it. Since after that, unless now we have taken those AI solutions and anchored it to our client's industry and vertical value chains, it's going to be very generic. It is not going to be domain specific. That's quite frankly just not important for our clients. It's important for us too because that improves our speed to market. This entire AI and AI transformation wave, as you would agree, that's never one and done. It needs to go through a continuous iterative life cycle, again, driving symbiotic outcomes for both our clients and us. Let's get started with the first one. As I said, I'll double click. How do we do collaborative design thinking? We start off with the customer journey mapping. While we are doing the customer journey mapping, we also do a very detailed due diligence on the trifecta of people, process, and technology. In order to prioritize those opportunities, we would need to obviously add some quants. Those quants really come in from those metrics or KPIs that really matter to our clients. We supplement it with best-in-class benchmarks so that we put our clients always ahead in the game. When we have seen this opportunity roadmap evolving over the years, we have also realized that analytics plays a very, very key role in terms of unfurling these opportunities across various segments of the operating framework. I take a few examples of products and solutions that we have worked on. Let's take workforce management. For workforce management, we have an AI-enabled solution that works across the entire spectrum of forecasting, staffing, scheduling, and real-time adherence. For training, we have a solution, AI Coach. You would see a demo later on by Justine, where it's a simulated training environment for TP experts where AI impersonates the customer. Let's talk about quality assurance, the perennial problem in our industry. If you think about the way quality assurance was being historically done, there were three key challenges. The first, very low sampling, usually 3%-5%. Even that low to 3%-5% sampling, just because it has to be done by human auditors, there used to be a lag of, let's say, easily five to seven days. When there is a lag, it keeps widening the gap in terms of the error propensity. Even after doing all of this, the kind of insights that you get are so shallow because it just talks about where the experts went wrong. We have actually solved for that with a GenAI interaction analytics platform, where now we are doing 100% sampling real time. It is just not about people insights, but a well-rounded perspective on people, process, technology, and policies. Now, let's also talk about what we have done as an advent of analytics as far as agent-to-service solutions are concerned. I think one of the common voices or the common things that you would probably resonate across the discussions that have happened thus far is that it's not AI coming in first. It's really emotional intelligence that is getting powered by AI. As we all agree, the interactions are getting more and more complex across all industries: financial services, healthcare, retail. If the interactions are getting more complex, we will probably have to have sentiment analysis as a part of the way we drive interaction management in general, be it voice, be it digital channels. In order for sentiment analysis to be done right and next best actions to be driven right when the agents are taking those interactions, we have done something which, quite humbly, I'll still say is remarkable. We are able to do voice-to-text transcription, text-to-text transcription. While we do transcription, we are also doing redaction, anonymization, intent identification, classification, and summarization, all at a record speed of 5,000 milliseconds. That is giving our TP experts to truly concentrate on driving resolution with the highest levels of empathy without really worrying or getting bothered about the mechanics of the interaction. That is what we are trying to enable with artificial intelligence. Now, whatever solutions and products I gave you examples for is that middle layer, the AI orchestration layer. This is a diagram that Anish introduced in his segment. This AI orchestration layer, as you see, is well nested within the vertical solution blueprinting layer, which includes the customer journeys as well as the interaction channels or the engagement layer. You have the foundation layer where you have all the backend apps. You must be wondering that there is one layer that this guy is still not talking about, which is the fact that when all of these layers are orchestrated in the right fashion, it drives a very, very differentiated and rich level of analytics, which goes far beyond just CX and productivity metrics. Now you're able to understand customer lifetime value proactively. Now you're able to understand what is the propensity of customers churning and just not one number by different persona and customer segments. All of that comes to life. What I'm going to do as I deep dive into the vertical plays would be to double click on two parts of this slide: the vertical blueprinting layer and the way we are driving this as not in a one-and-done approach, but a continuous iterative life cycle, which is really the layer which says operational impact guided by analytics. Now, whenever I show this slide to somebody, they say, "Hey, come on, I've seen this. How is it different?" I'll tell you how it's different. Most of the times, a key factor that aligns customer journeys with internal processes is ignored, which is that if customer journeys aren't linear, shouldn't processes be non-linear too? We have addressed this. When you see this map, you will appreciate the way we have aligned all the AI solutions to a blend of customer journeys, as well as taken away the silos that usually exist into front, middle, and back office by creating a more one-office integrated solution that puts everything together. By virtue of doing this, we have been able to create this kind of a view for all industries. We have been able to also align our homegrown proprietary solutions as well as our partner-enabled solutions, such as Emma, Parloa, Sanas, onto this framework. That gives us the added energy, agility, and acceleration to go to the market and probably pick up any use case which our clients might be struggling with as of now and drive the right level of efficiencies and effectiveness. Like I said, this is never one and done. It has to be a continuous iterative life cycle that aligns all of these AI interventions to not one metric, but to a hierarchy of metrics so that we keep measuring, we keep course correcting, and we also save a lot of dollars on implementation costs just because it's no more a one-to-one relationship. It's a one-to-many relationship, or in some cases, a many-to-many relationship. What our clients are appreciating the most about us is they're saying, "Hey, you know what? If earlier we were running 10 projects to solve for something, now we probably have to do just two projects thanks to the way you have helped us orchestrate the analytics in terms of driving KPI measurement." I'll say one more thing, which is why this one-and-done approach is needed. What is so dynamic about it? In my view, there are three things that are very dynamic. The first part is our client's products and service landscape is constantly evolving. They are not settling to status quo. That's one. Second, our clients, consumers, or end users, their needs, behaviors, preferences are evolving. A lot of those consumers might be in this room, us. Last but not the least, AI in itself is on a dynamic path. In my conclusion, this is what I would say with a halo of optimism. I feel the way this approach is all bringing to life with TP.ai FAB, we will be able to transform confusion to clarity, but more importantly, enable our clients with a very clear call of action to driving tangible outcomes, which is going to be symbiotic for both our clients and for us for those three things that I mentioned initially when I started off, which helps us protect our existing portfolio of accounts, helps us grow new customers, as well as improve the revenue for each customer. Thank you so much. She's going to double click on the answer. Hi, everybody. Can you hear me? Oh yeah, perfect. I've already introduced myself that I run banking, financial services, and insurance. I'm going to take us 30 seconds and also share with you my background and why I chose TP as my home for the last five years. Prior to here, I have 30 years in total of experience in banking, financial services, and insurance in companies such as American Express, Mastercard, Visa, and Synchrony Financial. One of you asked me a very good question out at breakfast, which was, "What did you do in those 25 years prior to TP?" I ran operations, product, marketing, servicing, and lastly, ran digital transformation at the bank that I just came from. I have the best job. What keeps me excited about my job is that I spend every day thinking about my clients. I spend every day thinking about the old me in what we do today. With that, let me get started and share with you the key highlights of what I'm going to cover today. Miranda alluded to this earlier. Thomas spoke to it earlier. We are growing again. This is an exciting industry. When you look at my vertical, which is what I'm going to peel the onion on, there's even 50% of growth that's existing in the captives, an opportunity to start outsourcing that today that hasn't happened. We'll get into that. Two, AI is creating opportunities. We heard it from our clients up here as well, creating opportunities for our talent, creating opportunities for ourselves, making educated decisions. When we, all of us, you're all customers in BFSI because you all have a card and wallet, you all have a bank account. We all have that. AI is creating those opportunities to create special moments and memories in those engagements that create growth, that create expansion in those lines of business, and ultimately create stickiness not only with our clients, but the consumer's last mile as well. We are uniquely positioned to win in BFSI, one, because 65,000 experts in this space globally wake up every morning thinking about the customer, thinking about what keeps them up at night, and solutioning to what you saw with some of the examples from Himadri on how we solution for the customer, starting with the persona, building down to the customer journey, and then proactively many times sharing with our clients as well. That also results in trust, trust that you hopefully saw on stage, but trust for us to be the leading partner when they have a question or a conundrum, we're the first that they strive for. None of this is possible without the foundation layer that TP.ai FAB built, playing and has been playing for us and is part of our DNA for years. Now we're rapidly scaling even more so on that as we verticalize and invest even more from what you saw from Anish and Akash's presentation. I'll jump in. Growth I already covered on the right. What you see on your left, actually, is the industry trends. There's a multitude of trends, and you know the space just as much as I do for BFSI, but I'm going to cover three things. AI is top of mind and top of the agenda for my customers. An example of that is in the insurance space. In insurance, there is 76% of U.S. insurers, it's a phenomenal number, that are leveraging AI first in their strategies. They're building out their integration layers and the next generation of transformation with AI first. They're starting to lead the way as well, whereas in the past, many times there were other verticals, and BFSI is leading, especially in insurance on that front. Outsourcing is delivering efficiency and efficiency gains. That's been there. That's table stakes. The reality of the pivot also is that there's loosening in regulation, and that backdrop is easing the approach and the ability to grow and expand operational efficiencies. It also touches upon the fact of the 50% that's not been outsourced today and the growth that can come in that in extended lines of business as we look at adjacent plays and adjacent vertical plays that were touched upon earlier. Lastly, there's a growing demand in integrated solutions. I'm going to give you a case example of that in a bit. An example of it here is rising fraud. Fraud has been prevalent in the space of payments in my space for years, $10 billion in fraud. How we address that fraud and how we also address the solutioning for it is something that we have invested in. You'll hear about some of that in specialized services. You heard about it with Anish as well prior to that. Banks are coming and asking us, "Help us solve for the fraud tools." You'll also hear as an example from specialized services, which is prevalent in my world, in what we're doing in collections and how we're leveraging AI and AI tools to make smarter decisions when it comes to collections and also those that take EI into account, all underlying with TP.ai FAB as the foundation. As I keep peeling the onion, this is me. This is my vertical, $1.4 billion of revenue in 2024, 14% of the company's revenue. What you see is the microcosm of the key hubs for BFSI. We are 65,000 experts strong, and we do not take that lightly. That is what excites me, 65,000 leaders globally, the global approach with the local servicing, the global aspect that reminds ourselves of the regulatory environments by market, but also reminds us about the global investments that we can then bleed into and roll into the hubs that service our customer better. With 350 clients globally, half of which are the top banks in the world, of which another 60% are the top 25 companies in the world, I'm humbled to service them day to day. As Himadri alluded to, we're a one-office model. We not only think about a solution for front office and forget mid and back office, we think about the customer. That customer journey many times resides along all of that. On top of that, you layer on Akash's world and Anish's world of data and AI and how we can service our customer kinder, smarter because we know more about them. As you heard about on the panel as well, where our customer is in charge, they are in the steering wheel of making those decisions. As I think about stepping back and saying, "So how do you do that?" You have talked about your strategy. You have seen double-digit growth over the years. You say you come from the industry. You are excited every day. Here is a snapshot of how my team and I, our 65,000 member team, think about breaking down those verticals within BFSI. As you go even further, there is a modular approach. Anish spoke about Lego pieces. I also love Legos. I think the Lego example is a very fair and a good, accurate example of how we approach our clients. Sometimes that approach is proactive. Sometimes that's through an RFP. Whether you're at the customer acquisition cycle of the customer, or you're at customer retention, or you're assessing risk, there's a gamut of capabilities that come with that, all holistic from a one-office approach that then create the differentiators that truly decipher us from our competition. An integrated one-office delivery model, which the team has talked about, sub-industry microservices that really create the nuances and the uniqueness of the vertical and the servicing that comes with that, strategic AI partnerships, some of which that you've already heard about and you're going to see and touch and feel in a little while at break and the ecosystem that's built, embedded analytics and agentic AI that is servicing, empowering, and is the DNA. It's my growth. It's what I have. It's the stickiness that I'm having with the customers, and it's the expansion and the wins that I'm having with new lines of business that are resulting in that cycle back to that modular approach. Scaled agile and transformation. I said I was running it, so it's been around for a while. The importance of that in the banking world and being able to pivot fast, have quick sprints, learn from your mistakes, and pivot even more has been critical. We are solutioning and supporting our clients on that front. Last but not least is the adherence to the regulatory environment and the audit infrastructure that is essential in my vertical. Here are a couple of examples of execution. Thomas spoke about a large US bank, one of the largest, whose headquarters are right around the corner from here. I am going to speak about another one, another very large US bank that we have had the privilege of having a relationship with for over 10 years. This was us from a core and building out the core platform with AI and us going proactively to the bank and saying, "We have been working with you. We have investments that we want to co-pilot with you. We have a proposal of what we can do, and our North Star are some of the impact metrics that you see. We exceeded our North Star because they believed in us. It goes back to the trust and the relationships and the partnerships. We exceeded the impact by showing six points in net promoter score, 15% reduction in handled and interaction time, and an improvement of efficiencies to 20%. That is exciting. This was proactive engagement. That resulted in the growth that you see year over year of 12% in revenue, not only in the existing line of business that we had, but expansion into the expanded vertical plays because now we had trust. We woke up that morning thinking about them. As a customer, we also, and that's an expectation of everybody on the team, is you need to be a customer of the clients you're servicing. In 500,000 employees in the company, there's definitely a customer in every single client that we have. That's a secret sauce to leverage the customer interface and the use case to say, "I actually was in this interaction as a customer, and I have some ideas that I'd like to proactively share with you, not as a problem statement, but as a solution." That's what we did here. Another example of that is extending vertical plays. This was a startup U.K. bank when we first started working with them. They believed and partnered with us, and we believed in them as well to partner with them to start what resulted in them being the largest U.K. bank in Europe and growing. Their expansion plays has and already is expansion into the US and a multitude of other countries. We've been their partner from the onset to grow with them and with them. What you've also seen here is we've grown from care to an example of a conversation, a simple conversation with the Chief Operating Officer because as you grow as a bank, there's unfortunately the unfortunate likes of fraud and fin crime that grow as well. We partnered with them to solution on what we could do to service out of one of our centers of excellence, one of the key for fin crime and fraud in India, and we solutioned that as an expanded vertical play and showcased to them, and you see the results. On both slides, there's also quotes from our clients, which I didn't cover, but it's at your leisure, and you can cover that when you please. With that, I'm going to hand it back to Mike. I was thinking about our client conversation, and each of our clients had different problems, different opportunities, different things they were trying to solve for. Something that I heard as consistent throughout was TP as a partner, TP evolving, TP finding a way to create a solution for each problem over time. We heard Miranda talk about the return to growth of BPS and how our solutions evolving with the market, integrating technology. Himadri bringing the solutioning together, which seems wide, diverse, complex, but building a process flow that allows us to do it in an efficient, effective way means that we can bring that solution to each and every relationship. He also made the point that it evolves over time. It's not a one and done, which means we continue to solve and change different things with our clients and earn, again, those opportunities for new growth, new ways of working. All of that is built on the 40 years of TP experience. Now, as we head out to break, I'll remind you again that we have some great demos in our demo rooms over here. For those who are joining us remotely, we'll be sending a landing page so you can experience those demos as well and see how technology is changing the way work is done. Thank you, and I'll invite you to the demos. Thank you, Mike. Great, great. OK, welcome back. I hope you all enjoyed the break and even more so the demos, getting to see how we bring technology to help solve clients' needs. As we shift gears a little bit, we'll talk about our specialized services offering. JC will walk us through each of those businesses, how they solve niche problems in high-value, high-importance situations. Without further due, JC. Thank you, Mike. First, let me present myself. My name is JC. I've been 20 years in the outsourcing business, 10 of those in IT, BPO, and 10 of those in Teleperformance. For the last 10 years, I've led the LATAM Teleperformance family. Always seeing the specialized services as one of the most beautiful spots of Teleperformance. Ten days ago, I got a call from Thomas and Daniel. Do you want to lead the specialized services for Teleperformance? Let me tell you, I didn't take 30 seconds to say absolutely yes. I hope today, in the next three, four slides, I'll show you why. First, we fit perfectly in the industry footprint that you were seeing this morning. We are an extension. We are very specialized companies that match TP's footprint and blueprint in every industry that you saw this morning. Let's talk a little bit more about that and about our AI transformation. Let's start with Alliance One, a collections company fully embedded into BFSI, health care. We have been doing collections for years. We have a deep expertise that now we are leveraging with AI. There's lots of synergies, lots of things that, with the complement of the core of Teleperformance, we will continue to build an even greater collection company. Let's go to TLS, TLS Passports and Visa Company. This is a full integration with AI and with India and our TP.ai FAB. Imagine from the back to the front, all the automation that we are going through. In this post-COVID world in which we are open to travel, open to do business, open to get everywhere on the planet, the visa is an amazing moment to live in. PSG recruitment. Just think about how many people Teleperformance recruits. Think about having the strength of supporting the core and what of the clients? Think on the client footprint that we have in the core of Teleperformance. All those clients are going to hire us. This is a huge opportunity to continue developing in a very specialized niche of our business. Health Advocate. I will not jump into Jeff, where's Jeff, into Jeff's thunder, but we have a beautiful employee advocacy company. Think about that. Think about having a specialized company in the health care vertical. The amount of opportunities and transformations we are living as we speak. This is going fast, and we are going to continue specializing ourselves. Lastly, the interpretation companies. ZP, our latest acquisition, Sherry will later share a video, but our deaf and hard-to-hear and sign language business is beautiful. There's a lot of potential to leverage our current Teleperformance footprint with this. I'll share a little bit later about this. Then LanguageLine, mission critical contact. Mission critical interpretation. Imagine the amount of situations that globalization and contactability, the new use cases for this industry, the compliance and regulation that will come to give a prosperous future to LanguageLine. I'll speak a little bit later about this. Then the addressable market, $100 billion. Think of that. We have a huge pool to develop our business. We are, through all this leverage that we do with the core of Teleperformance, plus the specialization of these businesses, going to continue to go even further in the growth and in the development of this specialized business. Just some growth cases for you to think about. CP, expanding to new markets, the major markets of the world. The U.S. has 10 million, and the global population that needs us to support them is 70 million. There is a huge possibility with the TP Global Network to expand this business by two, by three, by four. It is using the global footprint and capillarity of Teleperformance. By the way, a lot to still be done in the domestic U.S. market. PSG. Dave will be later presenting our latest automation, but we build ANNA. ANNA is our AI platform that has already screened 50,000 people. Imagine that. We are doing part of Teleperformance recruitment, and many clients using already ANNA to screen candidates. This is developing fast. I would love if you take some time and you go and see the presentation and see it live. It will be great for you to build a better impression. Expanding to adjusted markets in our Health Advocate business. Imagine payers and providers. They are all Teleperformance clients. We have huge implementations. I remember in my LATAM days, these clients are huge. You see, Thomas, I'm aged, but I have good memory. We have beautiful, beautiful health care clients around the world, and we are going to leverage and boost. It will shine. Health Advocate will be maybe the most beautiful star in the whole Teleperformance universe. If this allows me, I'll deep dive into LanguageLine. We have gone from voice to video interpretation to AI. Think on the use cases. Think of critical situations. Go to an emergency in a hospital. I'll share some examples later. Understand the person that's having an emergency, how he's feeling, how he's not using his native language, and how he needs to be understood to really be able for the doctor to understand what type of treatment does he need. Imagine a 911 call. Now, with globality, everything is interconnected. There are many new use cases. Compliance and regulation will take us to really do deep understanding in all the mission critical contacts that the world is enabling. LanguageLine will continue to grow because there are new use cases. There is new legislation. Of course, AI interpretation will continue to boost the business. Yes, 35 years of experience, our enormous footprint around the world, and the knowledge we have in the business will leverage mission critical contacts. Have no doubt LanguageLine will continue to be a booster of Teleperformance in the years to come. Next, I'll share a video about what do we mean by mission critical. My name is Jennifer. I'm a Mandarin interpreter for LanguageLine. I've been an interpreter for the past five years. There was one situation where a pregnant woman, actually, she's about to deliver, and she's there in the delivery room. I get on the call. It's a little chaotic already. Everything happens really fast as the doctor goes, "Push, push, push." I go in Mandarin, "Toy, toy, toy." The doctor has to count, "One, two, three." I just go, "Yeah, son." I was able to be there for the mom and the baby as a new life comes to earth. It is very important to me personally that I do my best on each call to help. As an interpreter, I help people bridge the language barrier gap. I am there, whether it is a happy time or a difficult time, but I am there to help them when they need me. OK, now we will talk about ZP, our latest acquisition, and how this expands our portfolio. For this, we have Sherry, who has prepared a beautiful video. Let's go for it. Hi, everyone. Thank you, Juan Carlos. Hi, I am Sherry Turpin. I'm the CEO of ZP Better Together. While I wish I could be with you in the room today, I am very honored to be able to walk you through the most recent acquisition of Teleperformance. We were acquired in February of 2025, and we have had an amazing journey from when I started in 2015. I'm, we actually, everyone at ZP is very excited for our future. I will tell you a little bit about ZP. Our mission is really clear, and it's one that has inspired me since I started in 2015. We are here to break down communication barriers. We are here to give the deaf community in the United States, which is where we currently are servicing, equal access that the hearing world takes for granted. It was astonishing to me how much the deaf community in the U.S. did not have, per the ADA law, just equal access and technology where they could live their life at work, at home, and on the go. Our mission is very clear. We're supporting American Sign Language, the community that's in their native language. We are really breaking down all barriers and providing everything they need to live their life to their fullest. We're very proud that 78% of our workforce is female. A lot of people don't know we not only service the deaf, but we hire the deaf. 55% of our employees are deaf and hard of hearing. We have over 4,000 American Sign Language interpreters. They're just amazing. A lot of people don't know that we're very proud. 2020 through 2023, we've won the 100% score, which is the highest score for the Disability Equality Index. We have three cultures here. We have a hearing culture, a deaf culture, and an interpreter culture. What really brought us together and made us be who we are today was really how we care for each culture individually. Together, we are the powerhouse that has become the provider of choice in the United States. A lot of people are unaware that ZP was the first provider in the United States to really launch the program that we service today, which is Video Relay Service. A lot of people, I will be the first to raise my hand. I had never met the deaf when I started in 2015. I wasn't even aware that there was a program. I think before I get into where we want to go in our future, I think it would be really wise for me to just pause, show you a video that we've put together. It took me a while, and I've had so many of my hearing colleagues and hearing friends go, "Oh, I get it. I now know what you're doing." I'll pause. I'll let you watch the video, and then I'll be right back. Every day, we make phone calls to do the things we need to get done and connect with the ones we love. What about those who are deaf or hard of hearing using American Sign Language? ZP's Video Relay Service breaks communication barriers for deaf and hard of hearing callers. Utilizing sign language, an interpreter facilitates seamless interaction in a video call, relaying the conversation to the hearing person on the phone. Hey, John. This is Amanda. How's it going? How did the job interview go? Yes, the interview went very well. They actually offered me the position right on the spot. Oh, I'm so, so happy for you. Congratulations. Whether it's for business to discuss a project, managing daily tasks, or simply connecting with friends and family, benefit from ZP's 20-plus years of expertise in Video Relay Service, fully funded by the FCC at no cost to you. Enjoy seamless communication at home or on the go. Break down barriers now. Hi, I'm back. What'd you think? A lot of people didn't realize it's a three-way conversation. A lot of people didn't realize that the interpreter and the deaf are in video together. A lot of people did not realize that the interpreter is literally interpreting in an audio call on a headset to the hearing party. We are 24 by 7. A lot of people do not realize that we are their 911 if they choose us as their provider. It is a wonderful business where we are doing wonderful work. To say that I am extremely proud that Teleperformance has acquired us and gives us what I am now about to cover, which is the future. The opportunity for the future is so much brighter than the last 10 years that we have been building in the United States. That is because internationally, there are over 70 million deaf and hard of hearing that need our service. What is great is we have paired our future with where Teleperformance already is. Their strong presence in the countries of Australia, Canada, Germany, and France is just the immediate jump-off point that we are interested in. We will be leveraging the relationships with Teleperformance. We will be partnering in the same way to bring communication access to the deaf in those countries through that government for Video Relay Service. We have built and know how to perfect in the United States. All of those skill sets will be easily implemented in all of these countries. I'm excited to get to work. I'm excited to show everyone, not only Teleperformance, but the markets and the deaf community, that we are doing exactly what I told them we would do. We would build in the United States. We would perfect in the United States. Then we would find a partner that could help us grow globally. We are doing just that. Thank you for your time. If you have any questions, please do not hesitate to reach out. We are excited about our future. I hope you are as excited as we are. I hope you understand exactly how important and critical this service is for the deaf. Thank you for your time, everyone. I will now turn it over to the CEO of Health Advocate, Jeff Cordell. Quick mic check. Hello. All right. My name is Jeff Cordell, and I am an unapologetic geek. I am excited to share with you a little bit of my background. I have been involved in some of the most disruptive communications events that have changed the way that we live and work today. I hold over 30 patents in the area of cloud before cloud was cool. I also have patents in the area of digital communications and artificial intelligence. I'm excited to share more today. I joined the Teleperformance family with LanguageLine back in 2014 and then continued to help the specialized services divisions grow before I took on the role as the CEO of Health Advocate. It was an acquisition back in 2021, and I joined the organization with the unbelievable responsibility and honor to be able to lead this organization. It was 2,500 years ago that Confucius said, "A healthy man wants 1,000 things, and a sick man only wants one." Imagine how much has changed in the world of health care since that statement was made. The advancements are staggering, and it's exciting what's happening. Along with that is the increase in health costs. Employers in the U.S. who are part of the health care and private insurance in the U.S. are seeing a rise between 8% and 9% in the 2025 year alone. I want to introduce a new topic. It's called population health. It's how do you approach addressing that increasing ramp? It involves a number of things: preventative and proactive engagement to take care of your own health and that of your family. We're trying to bend that cost curve. That is the basis of population health, and it's where Health Advocate plays. A lot of new terms. Those are staggering numbers. Let me bring it back to you who are sitting here in this room. Three things that matter most to you. One is your own money. You're paying for portions of your health care insurance in the United States, and we need to make sure that you get full value of that. Number two, it's your health. I challenge probably every person who has a health ring or a watch that they're tracking. And with access to data, it is easier, and it is smarter than ever. The third one that's probably the most important of all is your family. There's no expense you would spare to be able to take care of your children, your wife, your partner, even your parents and in-laws. This is the area where Health Advocate can provide services to that entire range of capabilities. Here's Health Advocate at a glance. Again, hired and acquired and joined the Teleperformance team, the TP team, in 2021. You can see by the numbers. Let me give a really salient example of the type of experts we have. Maybe you're new to town, and you need to try to find a primary care physician. You want to find somebody that's in your insurance plan, near your home, and accepting new appointments. That's a challenge that our health care experts would love. Maybe you had the worst news of your life, and you or a loved one was diagnosed with cancer. Where do I start? What do I do? How do I find an oncologist? How do I go through the process? These are the things that we do. We have nurses, clinicians, and doctors that are available to you, our members. Today, our clients are employers, and they buy the services just like they work with insurance organizations. Employees become our members. We have a large group of people that we provide a range of services. It does not just end there. I am going to speak a little bit about mental health. If you want to talk about an unbelievably difficult and pervasive problem, not just since COVID, but boy, did it put an exclamation point on the problem that we need to address in mental health. We offer what has been called traditional EAP, Employee Assistance Programs. You might be familiar with that. More than anything, there are mental health solutions that we are investing heavily in in this space. Finally, there is a range of services that we provide that are for biometric screenings. That is early detection to try to identify somebody who has hypertension, who has a high A1C, and those things with some preventative work. Maybe a really inexpensive statin can avoid the emergency room or worse. These are the areas where Health Advocate performs services. We're not done. Again, I'm the geek. We are applying technology at a staggering pace in artificial intelligence. If you think that AI started in 2023, you're wrong. It's been decades. It's been used for a very long time. I've got a couple of salient examples and case studies of how we're using it today. Our clients are also informing us about new products and services. Here's one. How many in this room are dealing with parents that are aging and they need help? We have a product that we launched called Generations. It's been wildly successful. It's how do you, the caretaker, the worker, take care of mom or dad who might not be where you are? When do we consider alternatives to be able to help them through that period of time? It's a pervasive and growing problem today. Finally, I've got an example to talk about how not just mental health, but there's a correlation between mind and body. These are the new products and services that we are launching within Health Advocate. First case study, a U.S. cyber tech company. Think 5,000 employees. When we add the rest of the lives involved, it's about a two-to-one. So about 11,000 actual lives when we consider the families that we support. The point that I would love to take away from here is we created proactive digital engagement. This is before Agentic became cool. The ability for us to have an outreach campaign on a new term that I want to introduce you to. It's called Gaps in Care. We have ridiculous amounts of data about those members of ours. Not just census information about the ages of your children and the programs that you're enrolled in, but claims data as well. We're able to evaluate that. Maybe you're about to turn 45 years old, and it's time for a colonoscopy. If he doesn't do it, we can head off a whole series of these things. We can find him, and in the channel that is of his choice, we can text him, and we will connect him to somebody who would be able to provide a colonoscopy. That is an example of closing a gaps in care, and we do it by the millions. There's a whole series of other things that we did with this client with proactive digital engagement. The results were astounding. Population health, remember that was the term that I introduced earlier. Literally, the person who wrote the textbook on population health, Dr. Ray Fabius, had a third-party independent group called Health Next. He followed this organization with us. They have a 1,000-point scale to measure how healthy of a population this employer is. The results were astounding. For Dr. Fabius, we drove a 115-point improvement on a 1,000-point scale. He thought it was not possible in a single 12-month period using proactive digital engagement. Again, analytics to find those gaps in care and digital engagement to close those gaps in care. It was incredible. It was exciting. There is more to come. We have case studies that are available on this if you're interested. I've got another one. This was a challenge. A large rail freight company, 50,000 employees. Again, think 2.1, so about 110,000 or 120,000 lives that we're supporting. It is hard to reach this group of people, primarily male. Tough guy. How do you get them to reach out for mental health? We built a strategy with this company right in hand-in-glove with them. The strategy was we are going to provide proactive engagement, those on-site. We incented it. We identified a number of chronic disease conditions. At the same time, we provided information about available mental health through our services. The combination of those was astounding. You can see the immediate correlation between somebody's mental health, and in this case, hypertension was a chronic disease condition, very addressable. The combination of those programs was able to remove people from hypertensive down back into normal range. It is an exciting time. I have this, it came across my desk just the other day. I'm too excited not to share this real case study if you want to understand how artificial intelligence is improving our human engagement. Here's how it works. Our EAP, our mental health, if you use our services, you can contact us either through web or chat or through our member engagement portal or the traditional phone. About 60% of the people elect to go see somebody face-to-face. If you're in the city, all around New York, you'd have access. There are 120 million people in the United States who are in mental health deserts. There is not access to those services. We've paired them with video mental health services. Not only is it available same day, next day, it's able to reach that population. Here's where AI comes to play. I just want everybody to pay attention. It is a wonderful opportunity. We've already recorded all of these calls for quality and monitoring purposes. That was the olden days. Now we use LLMs, and we do speech-to-text. It's gotten so much better with AI. Now we have a transcript of the entire 45-minute conversation. With really great AI engineers, you then create a summary that are the notes. Why did we do that? At the end of a 45-minute therapy conversation, typically, you have to handwrite notes and transfer them to the case study. Now AI has made that so it's automatically done. It's scribed the notes, and the results are astounding. 63% improvement in the reduction of post-call note-taking, mundane tasks. 83% of those texts are adopted. Metrics that are fantastic. That means that it's basically done, and I might do some post-editing, a little. The most amazing thing about the product and the service that we offered is the member responded. It felt like the therapist was connecting with me in a new and different way. They were not busy taking notes or typing. They were more engaged. It is easier. It is quicker. The quality of the services that we provide is much improved. It is so exciting. I cannot help myself. Again, data is the lifeblood of everything that we talk about. Dr. Alyssa Scott is on my team as my Chief Analytics Officer. She has a team of data scientists and AI experts. Now we are not just making in our system great gaps in care closure. We are now exposing it to our clients. Here is an example of a dashboard of somebody who uses our EAP mental health services. We can now describe the outcomes, the improvement if somebody engages in our programs that they got better. If you're an employer, that means that there's a number of things that are downstream called presenteeism, absenteeism, and ultimately attrition. Now we're ringing the bell of communities of why are we losing people when we can address services to keep that tenured intellectual property inside of the business. Dr. Ray Fabius talked about population health. It doesn't just improve the bending the cost curve. He has a direct correlation that the valuation of a company is tied to their employee population health. Lots to read about at Dr. Ray Fabius's name if you want to read some more. I'm going to close on this. Three things to remember. One is our clients today are employers. We are super excited about the fact that numbers of Teleperformance TP clients are able to use our services. We are starting to see with Juan Carlos' help, JC's help, that we are able to bring our services to those larger enterprise organizations. That is one. The ROI is easy to calculate when you are able to keep people that are called high-cost claimants. It is easier to calculate if you have been able to head off heart attacks and strokes. Easy to calculate. The second one is we are a high-tech human touch company. I hope you can see by the couple of examples of what we are doing. Our nurses and doctors are ready and available to help you. We are using that same technology in every single one of our group. We call it Project Summer and Summerization. At the end of every call, we have to type case notes. We're shrinking it for every interaction, inbound and outbound. We're using that same AI power and magic that I described in that mental health scenario. Here's the most exciting part of all of this. It turns out that the solutions we've created, which we call member engagement, primarily employees, is completely transferable to other worlds, to the pharmaceutical world. Maybe you've launched a new drug. Maybe you have results, and you have to reach out to your patients. Did that person that used one of the early detections for colon cancer and they're positive, did they follow up? Did they share it with their primary care physician? Did they get a colonoscopy? Do we have an oncologist available? This is where we are our best. We're moving into those markets. The one last statement that I would say is the payer insurance markets, in order for them to have a HEDIS five-star score, there is a number of patient and member engagement things they need to do. We are excited about expanding our services to help them as well. That is it. I believe I handed off correctly to the man that needs a colonoscopy, maybe. I'm not sure. A couple more years to get there. It's important to remember, as Jeff talks about all of the technology enablers within specialized services, those are all built in TP.ai FAB as one technology solution that underpins both of these large services, both specialized and core. You also heard about a number of these services that have been integrated in our service delivery and TP core, whether it's the Health Advocate product that helps our employees take advantage of the benefits we provide, whether it's PSG with a recruiting solution that allows us to produce a better core product in our BPS division. Additionally, by having these specialized businesses in a joint go-to-market, we're able to solve more problems for our clients, extend the relationship, and deepen the business that we do. From here, I'd like to welcome Anish back on stage to introduce our next session. Anish. Okay. First of all, I'm going to invite Professor Martial Hebert from CMU to come on the stage. And then Akash Pugalia, please. I'll sit here. Can you use this? First of all, Professor Marshall maybe does not need introduction for many of you. He is the dean of the school, which is ranked as world's number one on AI research. Thank you, Professor, for being with us. Thank you, Akash, for joining us. My first question to Professor, to you, is that you would have seen in the last few interactions that we have done today and before that companies like TP are investing a lot on AI and agentic solutions. There is a lot that is happening in the innovation labs that you run. What we are deeply curious about is knowing what is actually cooking in those and how does it impact companies like us in the future. Sure. First of all, let me thank you for the invitation and point out how happy we are to be part of the TP family and working with you. We started a human-computer interaction department, a machine learning department, language technology department before any of those things were popular. From all the presentations this morning, those are all very relevant to what you do. I am looking forward to this partnership. To answer your question, let me point out a couple of things and relate it a little bit to the history of AI. Along the history of AI, we came across a number of solutions, a number of approaches that at the time were labeled as being the solution. Okay? This is it. We have solved it, right? We just need to perfect it a little bit. Of course, what happens is that the next approach, the next revolution, the next solution comes along. This is going to happen again. We all heard a lot the word LLMs, transformers. This is assumed to be the solution. This is going to change and very quickly. Let me give you one example. One of our faculty members in machine learning, Albert Gu, developed an architecture called MEMBA, which is a possible replacement to transformer with much higher performance and, more importantly, much higher efficiency, both in terms of computing and in terms of data. This is just one example of what I mean here. That transformation will take place very quickly. The second aspect that I'll point out, and again, going back in the history of AI, this time I'm going to go all the way to the beginning. 1955 was the first AI program, the Logic Theorist, which was done at CMU by our founders, Alan Newell and Herb Simon. Why do I point this out? This program, the reason why it was significant, it was the first time that somebody looked at computing not just to crunch numbers or organize data, but actually to do reasoning and work with symbols. In fact, for many decades, AI was entirely about reasoning and symbols. We moved to data-driven techniques. We moved all the way to the other extreme, where now there's no reasoning. There's no symbols. There's only data and statistics. The exciting thing is that now we're moving back a little bit, actually a lot. We're now reintegrating those concepts of reasoning and knowledge and symbols into AI. Why is that exciting? It's because now there is an opportunity to not just look at a black box and look at its output, but actually have things like people have coined the term chain of reasoning, for example, if you look at the most recent OpenAI developments, for example. That gives the opportunity to have much deeper output from the AI system. Something that may not be completely evident, that also affects how we think about AI safety. One aspect of AI safety has to do with explainability, which is totally inexistent, let's face it, on black box system. Now having that kind of reasoning output allows us to go to that explainability. One issue with those new approaches, bringing in knowledge, et cetera, is that they may be limited in terms of computation because they now require additional computation. In your world, for example, where you might have issues on guaranteed delivery time, for example, that becomes a problem. There is another thread and another wave of development having to do with anytime computation, et cetera. Finally, that development allows a lot more integration of domain knowledge. There is a vast amount of domain knowledge in any domain, any of the verticals that you indicated. That is not necessary to learn from data. That can be integrated directly, except we did not have the formal method to link those two things, one symbol, reasoning, et cetera, on the other side, data, statistics, et cetera. That is the exciting development that I see in the future. Thank you. With that, I'm very proud to announce we are joining hands with you to build the next level of research and bring some of the advanced research that you do for the benefit of our clients. Thank you for joining us in that partnership. Akash, I'll go to you. What you heard from the professor on how the new advancements are happening, what specific role do you believe that data will play? In turn, TP can enable that to take the advancement on the research forward. Thank you, Anish. It'll be similar to what I mentioned earlier today. Data is the foundational layer in AI. If you think about models, how much ever sophisticated the model is, the outcome will not be relevant if the data is not accurate. If the data quality is not good. What we need to look at is how the fine-tuning of the model is happening, what sort of data quality that's there. That's where TP comes into play. Now, TP is at the juncture where we are implementing how humans are training the AI over a period of time, as well as AI is training humans at the moment, right? That is something that we are embedding in our operations, that cyclical process. It's the feedback loop mechanism that is helping our clients over how we can provide inputs to the large language models. That is where the orchestration layer is happening, where humans' expertise as well as machine intelligence are coming together. That is where TP is already playing. TP is also embedded with a lot of our clients in their AI development life cycle. Not only model training, model fine-tuning, but also looking at prompt quality and so on. We are embedded throughout the life cycle and able to answer multiple questions that they're looking at through our highly specialized skill resources. Yeah, I mean, it's throughout the value chain. Thank you. Professor, to build it further, Akash spoke about teaming up of AI and the human expert. I understand you have done a lot of research on that personally, and your university has too. How do you see what makes them effective as they work together? Are there more new frontiers to be established? Yeah. Let me give a little bit of background regarding CMU in particular. The School of Computer Science, which was the first of its kind, has seven departments. Interestingly, one of the departments is human-computer interaction, which is where most of our interaction is, by the way. Here is the interesting thing about this department. Even though it is under the School of Computer Science, most of the professors in that department are not what you would think of as core computer scientists or AI scientists. They are designers. They are social scientists. They are learning scientists, not in the sense of machine learning scientists, but in learning science education. Coming together creates that new field, if you will, of computing interaction. Now, to answer your question as to where that's going, one of the transformations over the past few years that I think we have not completely appreciated yet is the degree to which the interaction between the AI and the human is completely transformed. Okay? It used to be, and we did a lot of work for decades on AI-human teaming. Honestly, what we called human-AI teaming was you have this AI program here. You have the person here. They exchange information, or the one gives input to the other and gets the output back, and so forth. The level of interaction that we're able to do now is unprecedented in terms of the behavior modification on both sides, just like people interacting, really, except that now I'm not interacting with you as a person having a person's personality, but I'm interacting with an AI agent having an agent personality. I try to avoid, by the way, using anthropomorphic, sorry, adjective to AI, but I'll make an exception for that one. There is an opportunity to do the following, to go much further than just having AI-human interaction in systems and doing better in the engineering, but to create a whole new discipline. To answer your question, that's what we're trying to do, to create a whole new discipline at the intersection of AI, psychology, social sciences, et cetera, to formal models of those interactions and be able to use those now to make progress in designing those systems. Thank you. Allow me to push you a little further on that, Professor. Companies like us are the practitioners of the dynamic interaction that is happening between the AI and the human experts. In your belief, what is it that companies like us need to do right or avoid the risk of it going in the wrong direction? Part of the direction is what I mentioned, which is the formal modeling of the human interaction and couple that with the AI models. In other words, not seeing those as two separate pieces that come together, but see that as an integrated piece from the beginning. Yeah. Akash, as you see, these interactions are resulting in a lot of data. Agents are generating data that's being used to train them even further. Human experts are generating data that's being used to train the AI even further. What role do you think that we can play or companies like us can play to make the best use of data to train AI in the future? Yeah, I think there's one fact that I don't think a lot of you would know. I was reading it yesterday. In the next three years, the amount of data that will get generated will be more than all the data that has gotten generated in human history till now. You can imagine the amount of data that will come in. That means that the AI cannot just manage all the data on its own. The data needs to be, again, the data needs to be fine-tuned. It needs to be high quality. It needs to be curated. That is where TP comes into play. Think about agentic AI. We are already embedding ourselves in the vertical solutions, in the domain-specific solutions that are required to look at that data, to define what the capabilities and outcomes should be. That is where we are playing. I think another thing that the professor touched upon that we are already doing is the cultural integration of AI with humans. That is something that, while as consumers, we are using it, but from a B2B standpoint, how we are ensuring that our agents are ready to use AI in helping our clients, that is the other big wave that we are driving. Yes, I think there is enough and more that we are doing for our clients in this. The last point that I'll say is that, again, data is at the infrastructure layer. We at TP are already playing in that infrastructure layer at scale. Thank you. Let me just turn the needle in a different direction. There are a lot of investors in this conversation. They are obviously evaluating companies like TP on the investments that we are making, both on AI and transformation. For their benefit, would you like to elaborate on what are the indicators or the trends they should observe to really ascertain that the investments are going in the right direction and it's a sustainable competitive advantage? Yeah. I'll concentrate on one aspect, which is what I call the AI measurement. In fact, this is related to something that was mentioned earlier. I think Thomas mentioned this, the checking AI, basically. This is one of the hardest things and one of the key things to differentiate different approaches and different implementations of AI, meaning to what extent we are able to make measurements on that system that are informative of performance. If you look at classical engineered systems, any of the systems that we use every day, the reason why we can trust them and the reason why companies can make progress in developing them is that there's approximately 200 years of engineering science behind it, meaning things that range from formal mathematical method to statistical method, all the way to experimental best practices, right? We don't have any of that, or we don't have most of that in AI. Even something as simple as being able to define the right metrics and the right measurement protocol to evaluate AI systems is not something that we have at least formally. That is something that we need to look at. That is something that is a key evaluator, a key evaluation feature for any AI development and AI system. Now, just to make it clear how this can go on, even in things that can feel very obvious, let me give you an example from my world, from computer vision, right? One of the things that one needs to do in computer vision systems, for example, for self-driving cars and things like this, is to detect objects and to label regions in images, right? I want to say that this pixel is on a person, this pixel is on a car. An obvious measure of how well that works would be to say, what% of pixels do I label correctly? That sounds obvious. In fact, if you think about it, what application, what scenario actually requires 100% of the pixels in an image to be labeled correctly? The answer is none. You never need to have every single pixel labeled correctly. This is an example where, by using this measure, we're using the wrong measure for the task, clearly, but we're also using one that is much harder than what's actually needed for the task. You see, even in this case, when one would think, well, it's obvious, that's how I'm going to evaluate my computer vision component and maybe how I'm going to decide to make investment, because the one that is better in that metric is probably better to pursue, right? That's actually the wrong measure. That's true across the board, actually. This is a key, I think, a central point to concentrate on is given what we're trying to achieve, given the context, the application, the verticals, as you call them, what are the right measures internal to the AI system? I'll say one last thing on this in terms of you asking earlier about current developments, right? It's maybe very difficult to define those measures explicitly the way described in the number of pixels and things like this. There's a field that emerges now with its own conference. It's called experimental machine learning, right? Which for people like me, older people who were raised old style in science and technology, is an atrocious thing. Why would you have something that you created now need experiments, right? In fact, it makes sense in the sense of using protocols to be able to measure those machine learning systems and to have measures that can actually be used to evaluate them. All right. Akash, anything further that you would want to add on that? Nothing. I think the one point that I'll just add on to what Professor said is the feature on responsible AI and ethical AI is also an area where there is no consistency. Responsible AI is about bias mitigation, about accurate data, about hallucination. Ethical AI is about governance and ensuring regulatory compliance. That is something that is not standardized. I think as the field is developing and growing, it'll become more and more focused on what we can do as service providers to help clients in that situation. Okay. Thank you. Thank you both. I'm just going to add one comment. Our intent of doing this partnership with the world's number one institute is simply very, very easily understood. We are going double diving into our overall TP.ai FAB. This partnership allows us to do a lot of research on possible industrialized solutions that we can take to market much faster than any of the competition. Thank you, Professor. Thank you, Akash, for joining us today. Thank you so much. Thank you, guys. Super insightful discussion. I think it's interesting, as I reflect, we can all probably imagine how computer-to-human interactions have changed, even in our time, where most of our interactions were done at a keyboard to being done at a smartphone or some kind of tablet to interacting via voice with technology systems. We can imagine in the future that technology becomes more and more integrated in how we live our daily lives. Being at the forefront of those new technologies, understanding how we build them into our service offerings, how people change around that technology will be super critical as we go forward. I'd like to invite Olivier, our Group CFO, to the stage to take us through the future outlook. Thank you, Mike. Good morning to all. I'm going to be much more down to earth to explain what our outlook. You have heard our colleagues all today, all this morning, and the plans that were presented to you for the next three years. We translate this plan in figures that I'm going to share with you. I wanted to reiterate the financial ambitions that the group has set for the strategic plan, sorry, over the next three years, ending at the end of 2028. First of all, growth. I would say three things of it. First of all, we are setting a target for three years. For those who remember these years before, they were asking us to deliver a guidance for more than one year. We are delivering a three-year guidance, 4%-6%. Third point. Second point, we are setting this target at 4%-6%, mid-single digit by 2028. I just wanted to hammer that. This is a significant performance. It means that either in 2026, either in 2027, you will see an increase of like-for-like versus 2025 guidance. I just wanted to be clear here. We are going to continue to grow in 2026 and in 2027 before achieving, hopefully, the high bar of this guidance for 2028. Second point, operating margin. The objective here is to continue, and I'm using this word not by chance, to continue improving the margin to reach an operating margin on revenue of around 15.5% by 2028. Of course, and you understood that from all the days that we had today, we will make significant effort over these three years to build, develop, and enrich this AI platform that was presented to you. These financial organic efforts, either OpEx and CapEx, will be largely offset by a series of cost reduction initiatives that can be grouped around three priority actions. First one, the completion of the implementation phase of the synergy plan following the merger with Majorel. The implementation of the plan in France that has just been finalized, started finally to start, will enhance the competitiveness of our operation with effect from full year by 2026. Second one, furthermore, an advanced, and you understood JC is there to help us, an advanced integration, an internal synergy plan for specialized service activities with core service one will be implemented starting as early as the second half of 2025. It aims to pull several costs with shared service, but also systematically rely on internal resources for all offshore interpretation activity as an example of further integration for LanguageLine Solutions. Finally, you heard a lot about generative AI to sell, to develop for business, but generative AI will greatly simplify the group organization by making it more fluid and more integrated across its various functions, thus allowing the amplification and delaying of organization. Clearly, you understand there is a clear set of plan of savings that are going to largely offset the investment, either OpEx and CapEx, that are designed to develop the business today. Let's move to the balance sheet structure. The group aims for net debt total to recurring EBITDA ratio for around 1.2. TP has the strongest balance sheet in the sector. We wanted to rev it further. The group intends to strengthen the situation by dedicating a significant portion of our cash flow to this debt reduction. Finally, and I wanted to be clear on that, the group is going to deliver $3 billion cash flow over this three-year period, just looking at this figure versus the market cap, in free cash flow over the period 2026-2028, offering for substantial potential return to shareholder while maintaining a strong financial discipline and investing in the future. This is the case. So this group is going to generate $3 billion to deleverage and to return cash to shareholder. How we are going to make it? What is the plan in capital allocation over the next three years? 20% of this free cash flow, roughly around $600 million, will be allocated to additional investment to accelerate the group AI transformation. You understood that we are working with a lot of people, either partnership, small bolt-on acquisition, speak with a lot of people to enhance this AI transformation across the group to be the leader. This money will be spent over three years. This is a basket that we need to adjust precisely. This is something that we need to do on our capital allocation to continue to drive growth. As you understood, 30% will be used to further strengthen our balance sheet. We will finish at the end of this plan with a debt to EBITDA ratio of 1.2, largely better than all the competition. This is something that matters for the Fortune 500 companies to work with. Lastly, 50% will be returned to shareholder. We are speaking of $1.5 billion. It's not small money. There are still discussions about how we are going to make it between dividend policy and share buyback, but at least we will maintain the dividend policy. There will be share buyback, and this $1.5 billion will be returned to shareholder over these next three years. Of course, the group wished to retain certain flexibility for any strategical and financial value creative acquisition. What is clear today is not our as a group intention to carry out such an operation in the short term. That is what I wanted to let you know. Now I hand over to Thomas that is going to conclude, to make the wrap-up. Thank you, Olivier. Maybe to sum it up, and before we go into the highlights, to really try to give you the big picture from our perspective. If you think back what we've heard the last three hours, we're trying to paint a picture from the future. What is our view as TP about the future? Not in 15 years. Who knows? Maybe by then we live in the singularity if Marshall will develop the AI. What is the future the next three to five years? Where do we see the business developing? Is our view that AI will replace humans and basically take the human interaction completely out of the picture and act autonomously? Or is our belief, yes, AI will be ubiquitous, but it will be, as I said before, like electrical current. It will supercharge the work we do, enabled by humans working together in a new form together with AI. We are a B2B services company that works in complex enterprise environments that have to fulfill security data, responsible use of AI, being specific to the use cases we see in our clients, and drive outcomes for them. Our core belief from everybody on stage today, and when you see also recent articles in the news, is not that AI will take humans out of the picture, but it will change the way we operate. It will make us more efficient. It will make us more effective. It will create different relationships between human empowered by AI and agentic AI. It is about a future we see for our business that enables the interaction with both. The history of TP over the last 40 years was a history of change. This is what we're seeing the next three years: changing our core with our strategy, expanding our value chain, and capturing the new opportunities you heard from Anish and Akash. We do this always with the same four ingredients: managing people at scale, which is a key component in today's world. It's not an easy task to do this around 100 countries worldwide. Being process excellent, being obsessed with process consistency, adherence, and excellence performance. Having the technology know-how. You heard about the 3,000 people we have, driving this change, partnering with the best companies we see in our space. You saw Emma, Parloa, Sanas, CMU today. These are the world-class partners we are happy and we are privileged to have on our side. Having domain expertise. You heard from our colleague, Mamta, who spent many years in the banking industry and now joins us with her 65 colleagues to drive that domain expertise. As a company that is capital-light, we do not have large assets. We do not have large factories. We are agile and can adapt to these opportunities quite rapidly, using our capabilities that we have and drive this. You saw this in TP.ai FAB, the foundation of what we are trying to do: intelligent orchestration based on secure infrastructure of human and AI for outcomes, for industry-specific outcomes that matter. There, we do believe we are the best partner for our clients, for our business partners that we have, for our employees, and at the end, also for our shareholders, because this allows us in an efficient manner to deliver significant cash flow, yes, that we will responsibly invest in our business to grow it. We see growth opportunities in all three strategic avenues, but also to return it to you, because that is our commitment as management to put the money that is not necessary to develop and transform the business back to our shareholders. With that, I will hand over for the closing remarks for our legendary founder. Over to you. Thank you so much. I hope that you have seen that there is a lot of positive energy in this group and a lot of talent. That is the first point. The second point, I'm going just to say two things about AI. First, data. Garbage in, garbage out. There is something very important today: there is a bottleneck in new human data because everything has already been integrated by the LLM. Clearly, there is a need for reinforcement by the humans of the data that can be generated by the AI. Second, yes, we live more and more in a digital world, and our kids even more than us. You know what? The digital world is as dangerous as the real world. It is even more dangerous because there are less cops. Yes, we do a job that I'm super proud to do, which is business integrity and trust and safety. Business integrity is against a deepfake, and trust and safety is maybe to protect your kids from the porn or whatever. By the way, I would like to invite you to read what the new American Pope said two days ago. Not only do we use AI, but we work on AI. We enrich AI. We control AI. We transform AI. It is a full partnership. We live in, with, beside, inside. We are totally integrated. What I want to say also is that the beauty of the world is its diversity. First, companies have very different strategies depending on the product, the service, their positioning. Second, the reason to interact for an individual with a company is multiple. Some are mundane. Some are extremely important, emotionally speaking. You can have a lot of dissatisfaction that can be created. I remember when we started to do customer service in the 1980s, most of you were not born. There was something that we knew already, which is a satisfied client is going to impact three persons when a dissatisfied client is going to impact 10 persons. Our key role is not to be just a facilitator, but a loyalty builder. Finally, yes, all populations are not the same. There are different ages. Of course, if you are born after 2000, maybe you spend your time on—I'm not going to mention the name of our clients—but you also have an aging population. By the way, Jeff was saying, "Yeah, you even have to take care of your parents." Can you imagine? You know this aging population prefers to have human contact. Think about what it means statistically. It is not one size fits all. It is super important to have that in mind. This goes also with the education level. I can tell you we can make a lot of correlation, a lot of segmentation correlation, and find different ways to communicate to different public for different kinds of topics. That is why. Because we live in, within, beside AI, and at the same time, because the world is diverse and because the world changes every day, that we have a great future. Thank you. Now we'll open up to questions in the room. There should be a microphone on your table. Excuse me. Hello. Yeah. Hi. Good afternoon. Thank you for taking my questions. This is Suhasini from Goldman Sachs. I have three, please. One is on the top-line growth: 4-6% like-for-like growth in 2028. I appreciate the color you gave on your expectations for 2026 and 2027. Given AI potentially has deflationary effects on the pricing, how do you see the pricing versus volume component in your target when you set out that for 2028? Do you have any expectations on growth by division? Specialized services versus core services and DIBS? Maybe I'm going to try to answer, and then as he's much smarter than me, he is going to finish the answer. Really, you are right. AI has deflationist impact, but AI generates also new needs. It is super complex. You can imagine it is super complex to make a model. We build the future depending on what we see already today. What we see today is that the kind of service and the kind of solution we provide to our clients are much more engineered, like Himadri was presenting. At the end of the day, the revenue that we generate with our largest clients tends not to be deflationist. There is more volume. There is more geostrategy. There is more digital integration. At the end of the day, we see a kind of stability. What is important is that, yes, it's important that we open the new line of business, like Akash was presenting in data annotation, like a very traditional business that nobody mentioned, and that is massive for Teleperformance, which are the B2B sales, because we are passing from an ecosystem in which a company used to own their technology to an ecosystem where the company goes to cloud. Cloud generates much more interaction than when you own a technology. The numbers that we give are a mix between our business development plans, what we see with our largest client, and the fact that, yes, we consider that we are not going to be able to increase the price per interaction because we are in this very, very transformative environment. Thomas. I have a mic. To add to what Daniel said, it's, I think, too simplistic to just look at volume and price per interaction. You have to look at the line of businesses we do provide. That's when we try to build a model for the future and to resource for the respective areas where we see growth we want to invest. As you see in our numbers, we have roughly 6% or so of our revenue is sales, for instance. It's a line which is not a cost factor with companies, but a revenue-generating business. We see a lot of growth momentum in B2B sales. The same thing is true for our data services, which is reported within trust and safety. It's a line where we see a lot of potential. There we will see it's not a question of interaction, but opportunities to provide that service. Another example is our back-office services line where we extend in what Miranda and Himadri presented to see opportunity, how we can grow further, high single digit in this area of back-office services. It's really line by type of business. Same thing, Health Advocate, you saw the presentation, recruiting services with PSG, LanguageLine at TP, where we see where do we see growth and where we want to double down on that growth. Yes, technology will make in a status quo the price per interaction or the cost per interaction less. That's been the case for many years. It's about adding additional layer of services solution on top of that. We have tried to be careful. Thank you. Any color on growth by division, core services versus specialized services? I can repeat what we said in our Q1 numbers. We are very happy with the growth momentum we see in our core BPS business. We are a little bit cautious for the growth perspective for our specialized services as we have this special environment right now for LanguageLine. In both areas, we are super excited. As Juan Carlos said, we have beautiful clients in both divisions, and we want to serve with them. I would say one thing. We see our core service business growing stronger than what one could have expected or what could be the market fears expressed in 2023 or 2024. Now, it's true that we have a specific difficult timing directly related to what's happening in this country, and that everybody can understand. That is for the specialized service and specifically for our largest company. Now, our largest company is in great shape. The margins are not impacted. In fact, we think that the kind of stabilization or lower growth that we see right now is something that is going to be temporary. We, in fact, had this phenomenon already, not at the same scale in early 2017, and I'm not going to comment more. Thank you. Just moving to margins, 15.5% margins, it is 50 basis points expansion between 2024 and 2028. Given AI can potentially give you margin accretive revenue, so I think it was mentioned in one of the presentations, why not look for more? Of course, we could have bragged for more. What we have designed in our model is, of course, some cost associated to that, some savings. I tried to explain what kind of savings we will do. We have been putting that in a model to see how it's going to be phased, because this problem is phasing, so it might change from small from 2026 to 2027. Finally, we thought that landing at 15.5 was a good promise. We hope to be better, but we thought it was not really needed at this time to bet on the future. We are reasonably confident that it could be higher, but let's do it. Let's deliver it first. Yes. There's no free lunch. Of course, we have to invest to reap the benefits. Got it. Thank you. My last question is on the free cash flow, please. $3 billion of cumulative free cash flow by 2028. You generated $1 billion in 2024. You're planning to grow your top-line margins over the next three years. Why is your free cash flow not more than $3 billion? What am I missing here? Thank you. Conservatism again. No, it includes also the internal area. Yeah, yeah. If we invest, for instance, in the internal teams, you saw the strategy from Anish Mukha. We're investing it. We're investing as we speak in our B2B sales team. We do internal investments. This cash flow is net of those. Is there any change to CapEx to sales guidance? No, no, not dramatically. The CapEx will be between $2 billion and $2.5 billion, probably go close to $2.2 billion. It's not going to dramatically change. There might be a little jump by 10 basis points or 20 basis points a year on our side, but it's not going to be dramatically changed. Thank you. Thank you for your information today and also the excuse me. Thank you so much. My name's Soren Shaw. Thank you for today and the demonstrations. Two-part question. You made the comment earlier about one size does not fit all. We saw a demonstration in the back with one of your travel digital agents. The demo gave the example of the cancellation of a reservation, switching dates. What do you think in terms of segmenting the AI offering? Do you think there will be one base level of functionality? If you want additional functionality for that AI tool, there'll be a pricing mechanism. You'll have opportunities to collect lots of different revenue off of one offering based on functionality, a little bit like Bloomberg does, right? You can get one layer of functionality, and if you want additional data pieces, you pay extra. Second question is, with the human agents, they're very good at cross-selling. How do you envision kind of a digital agent doing a cross-sell? Thank you. Thank you for the question. The first question is a very easy answer. Absolutely, yes. There will be multilayering of the time of agentic AI the client wants to consume. As the consumption increases, the price increases. The good part is, and it is Olivier's score, but I'll say it anyway, the more they consume, it is geomacretive. Therefore, it adds far more value to us if they continue to buy more. The likelihood of them buying more from us is even higher because the customer outcomes are far more greatly impacted. You saw that in one of the cases. Before, we could demonstrate, there was supposed to be an outbound call which would have registered the customer for an alternative restaurant. This is an additional service they could buy. Absolutely, yes. We believe that a lot of this design that's going to happen is going to get ported across many other industries. Whatever you could do, for example, in insurance claim that we earlier spoke about, can potentially be transported to healthcare claims. There will be that portability that will happen, and you could take one case to another and then make margins out of it. There's another question. Core services has already started to rebound, but it's going to take you guys another three years or so to get to the mid-single digit growth rate. What's going to happen in between now and then that's causing that? Maybe there's a misunderstanding. We're seeing progressively improving. As you know, the guidance for this year is 2%-4%. If we adjust for this non-renewal, as you know, from specialized services, it would be 3%-5%. We see the growth rate progressively moving to this mid-single digit. Yeah, I don't think so that we are going to need three years. Hey, thanks, everyone. It's Nicole Manion from UBS. Just one question from me, actually. Olivier, I think you already kind of hinted at what you said on the CapEx to a% of sales outlook. Could you maybe talk a little bit more about what you see as kind of the split within those AI investments between sort of what runs through the P&L, what might be additional CapEx? Yeah, any additional kind of detail on that? Thank you. Difficult to give you a precise figure, but clearly, we do expect to have some additional CapEx, not in a big size, not in a big size. I am speaking of the CapEx, not of the investments that could be used for the free cash flow. It just makes the difference. I do not expect something dramatically higher than $100 million-$150 million max. There is something very important already. You know that we pass in our infrastructure, we pass mostly to cloud also. We significantly have less hot CapEx than before, but it means that it has been translated somehow into OpEx. You move from depreciation to club subscription cloud. Other questions, maybe from the remote listeners? Fantastic. Marcus Schmidt from ODDO asked, you said you aim for a net leverage of 1.2 by financial year 2028 from 1.9 most recently. What rating do you aim for then? Today, the group is rated BBB, as you know, Standard & Poor's. We have no precise target. I would say, let's put it this way, we have no public target. We might say that if you have 1.2 times EBITDA, recurring EBITDA debt ratio, you will be at least BBB plus. That's what I can say. Thank you. An analyst from Kepler asked, the group's pivot toward data services and consulting will expose you to new competitive landscape. Which companies do you expect to increasingly compete with TP going forward? Could you provide some examples and the key competitive advantage compared to those players? I can try to start to answer. As usual, we are going to make our deal. First, the borders become blurry. They are blurry between BPO and CX. They are blurry between IT service and BPO and CX. The low- code, no- code change many things. It means that the competitive environment is moving. Yes, there are the direct competition, but I would say that the direct competition, here maybe there is one other player that present characteristic more or less similar to TP. Otherwise, we should take advantage of this situation because we are in an ecosystem, in an environment of consolidation. Really, consolidation means that we have more opportunity to grow and to be successful. Now, I prefer not to mention any companies, but yes, we can be in competition for some line of services with large companies that present themselves like consulting companies, but that are also BPO companies. We can be also in competition with IT service companies, with BPO companies. I would say that right now, I think that there is one geography. Strangely, I'm not going to speak about the nature of companies, but I would say that there is one geography that today presents a lot of positive characteristics, which is India. Fortunately, our largest company, where we are the stronger, is in India. Clearly, I would say our most direct competition are going to be several Indian companies. I would add, if you think about the markets we described, yes, we have the trend in the core BPS business of consolidation. Yes, with our new offering, we're entering new markets. We do have two distinct advantages. One is these markets, in particular, when you think about data services, are just evolving. It is an open space that exists. You have newcomers coming in, startups. You have large established players trying to make their venue. TP comes from a position of strength, given our global network, given our existing client relationship, and given our process capabilities. It is more a question about execution that we see winning when you look at the pipeline, winning the deals we have at hand, and proving our excellence operational delivery in these elements. Because we do come, I think Anish talked about it before, from the actual floor, from the operation where the AI is at place. We do not have a DNA coming from a PowerPoint and then trying to make it happen. We come from a status that we are already in the midst of making, bringing the AI to life, talking to the interaction, managing the processes at hand, and then basically moving up the value chain. From that perspective, I do believe we have a lot of strength and tailwind with us as we are actually coming from the actual operational floor. Maybe one other point on that. I would say that our competitive advantage is the diversity and the stickiness of our client base. It is our kind of distribution power. That is the reason why many new-gen AI companies are very interested to partner with us. Because somehow, we have a fantastic advantage, which are the decades in which we have been serving our clients. We have a deep trust relationship with our clients. Mr. Schumacher asked, why don't you stop the dividends and focus on buying back your shares? The return on your shares is higher than any other options out there and higher than the AI investments you're making. First of all, the question is not raised for 2025. You understood that 2025, we made the acquisition of ZP. The question will be raised again in 2026. What we have set up as a target was to give back to the shareholder EUR 1.5 billion over three years. First point, I am sure for those who are not perfectly aware, but buying back share costs on top of the price that you pay, 8% of tax. I just wanted to remember that. This is something that does not help. Decision of how it will be exactly split between dividend and share buyback has not been made. The board will probably decide that later on. I have had time to see that such a company like us will cut the dividend, announced to the market that we will cut the dividend. I do not see that happening clearly. To serve a dividend, a reasonable dividend over a period makes sense, makes sense for most of our shareholders. Not all of them, but I know there are some of us that are just interested in share buyback. We do believe that dividend is also part of the story. Again. We can say also that the opinion on this topic are very different. Yeah, both sides of the Atlantic. Again, I just wanted to highlight a point. I have not made the math, but I believe that the free cash flow yield should be beyond 20% as we speak. I am maybe asking you, both of you, but you are much more closer than 25%. I know companies that are in difficulty, but companies that are 25% free cash flow yield, I have seen a worse situation in my life. Great. Remi Grenu from Morgan Stanley has two questions. Margin guidance of 15.5% after AI transformation, to confirm. Can you help us understand what level of margin you expect through this transition? Could it decline from 15% you're targeting this year? Look, we provide today the midterm guidance. We provide the update for 2026 and 2027 and the respective guidance for the year. As Olivier indicated below, we are investing in the business, but at the same time, we are also investing in efficiencies that should offset the investments. We will provide more details as we move forward. We are committed that every efficiency we gain is, of course, translating higher margins, including the investments we do. The company is totally committed in these three plans that I just mentioned earlier on that are starting. Some of them are already launched. Some others are already started. We are absolutely convinced that we will be able to deliver our company and to improve our engineering and direct costs on top of that. What led to the change of management within specialized services? Oh, it's very simple. Scott Klein, who has been a fantastic leader for specialized services and specifically driving the incredible growth of LanguageLine, I would like to remind everybody that when we acquired LanguageLine in 2014, it was less than half of what it is today in revenue and, of course, profitability. It was very interesting because at the time, there were many people who were telling us that we were crazy to acquire LanguageLine because that would be a business that would disappear in the following two years. That was 11 years ago. LanguageLine is in very good shape. Scott Klein, over the last several years already, was considering to retire. He made the decision to retire by the end of July. By the way, the recommendation of Scott to us was also to have JC, in fact, his real name is Juan Carlos, to be the one who will take the succession. Thank you. Carl Raynsford from Berenberg asked, guidance of MSD organic to 2028 implies that '26, '27 will be below that. Could you please clarify what you mean here? Say again, I'm not sure. When he says that the 2026, 2027 will be lower than 2025, that's what he said? I believe lower than '28. If you could clarify. No, again, I'm not going to enter a detailed guidance for 2026, 2027. The budget process has not even started as we speak. We have, of course, plus and minus. It has been mentioned both sides by Daniel and by Thomas. We are working on it. Of course, we are a reasonable company. We will adjust to deliver something that will be in line with the market. You know, I'm going to try to give a conceptual answer. We are in the range of the marginal variation. It's extraordinarily difficult three years in advance to come and to say exactly what will be the plus and the minus in 2026 and the plus and the minus in 2027. We decided to give a commitment for 2028. We do not expect to have something material impacting our margin in 2026 and 2027. I hope that this conceptual answer clarifies the things. Last question from our online attendees. Rob Denek from Juno Investment. TLS has contributed nicely in the past, but seems to have less in common with others. Corporate-focused parts of the group. Is this still an activity you wish to develop? Yeah, TLS is a fantastic company. Unfortunately, TLS did not win the renewal of a significant contract last year. It is not the same perimeter. Still, with the new perimeter, it is a fantastic company. Clearly, we are clearly and systematically open to study what makes the most sense for the shareholder. We have not taken any position or any decision today. The question is not irrelevant. Any additional questions from those in the room? Is it working? Yes. Hi, I'm Courtney Fingard, journalist and consultant in London. This may be a question for Thomas. How do you reassure the workforce and partners around the issue of jobs? Because that's always the emotive issue when we think about AI. You have articulated really well the implications. Do you get nervousness from TP employees, but employees of your partners and clients as well? How do you address that over what the implications are for people's jobs and incomes? Look, if you see TP today, we are present in 100 countries. We have demonstrated growth the last years. We talked about the example of India. India has grown now to more than 90,000 people, almost doubling the business over the last years. This is the best practice. If you are investing in businesses that are growing, you see the success on the ground. TP as a whole has grown its employment numbers over the last years. What the future brings depends on our hard work. It's too early to say what the net balance will be. Thank you. There is also something that you have to keep in mind, which is at Teleperformance, typically, you have 85% of the workforce on the front line and 15% in management, support, and administration. Typically, this front line business is a very demanding business, by the way, because it is quantitatively and qualitatively very demanding because it is emotionally demanding. All our industry has a significant churn. Typically, people would stay in the front line something like two years or something like that, which means that if any adjustment would be needed at one point or another, it would not present the painful characteristic that it presented for many other kinds of industries. OK, thank you very much, everybody, for your time, questions. Great dialogue so far today. For those still here in the room, we'll continue our discussions outside. Thank you. Thanks, everyone. Thank you. Thank you very much.