Thank you, everyone, for being here. I am happy to launch the 2026 Capgemini Capital Markets Day, along with the members of the Capgemini leadership team. I look forward to spending this afternoon with you. During the upcoming presentations and conversations, we will share our strategy and market opportunities. We will hear client and partner perspectives and talk about performance and ambitions. As usual, you'll be able to find the materials on the Investor page of our website, capgemini.com, after the end of the event. Before we proceed, I would like to remind you that the information in this event contains forward-looking statements with respect to Capgemini's financial condition, results of operations, business strategy, and plans. Actual results may differ materially from the forward-looking statements as a result of a number of risks and uncertainties. The risks described in the Universal Registration Document Capgemini filed with the AMF. You have on-screen the program of our afternoon. We will take two breaks and hold a Q&A session at the end of our program. We will not field questions during the event. Please keep them for the Q&A session, during which we will also take questions from our online audience. Let's begin. Let's Make it Real by hearing first from our clients and partners. [Non-English content] Hello, my name is Cedrik Neike. I'm excited to join you virtually today. I am Marc Vrecko, CEO of Valeo Brain and Group Executive Vice President at Valeo, a global technology provider for automotive and mobility. At McDonald's, we're undergoing a once-in-a-generation digital transformation, redefining what it means to be a modern, connected brand, all powered by our Digitizing the Arches strategy. Siemens and Capgemini have long been partners. We've been transforming industries together, like aerospace and defense, pharma, life science, water, automotive, and many, many more. In recent years, we have accelerated our joint go-to-market. One of the cornerstones of our partnership today is bringing AI to our customers in industry and infrastructures. Together, we're making AI an industrial purpose technology, because the questions our key customers are asking are currently shifting. They're shifting from what can AI do, to how can we implement AI and how can we scale it? In the industrial world, AI is facing a different environment than in the consumer world. They're disconnected systems, siloed data, brownfield environments everywhere. At the same time, we need precision. AI needs to be predictable. It needs to be explainable, and it always has to react the same way. In automotive, everything is about execution, speed, quality, safety, cost predictability. AI only matters if it improves these metrics at scale. We've moved fast from exploring AI to focusing on where it delivers real value. In engineering, AI runs inside complex systems, legacy software, hardware, and strict constraints. AI can unlock value, but it doesn't remove these constraints. It raises the bar for governance and accountability. Industrial AI demands disciplined execution. That's where value is created, combining AI with deep industry expertise and engineering rigor. Our strategy comes to life in three distinct ways. First, we're transforming our consumer platform to make every customer feel like one in a billion through one consistent, borderless, and more personalized consumer experience. Second, we're transforming the experience behind the counter for our crew by elevating productivity, speed, and operational reliability through AI-enabled innovations. Third, we're modernizing and simplifying our company systems so that our employees have a more seamless, intuitive experience that boosts productivity and unlocks new capabilities. Each of these transformations is significant on its own. Combined, they work together to unleash our scale and global footprint to accelerate growth. As part of our three-legged stool philosophy, we rely on strong partners like Capgemini to achieve our ambition. Recently, we renewed our strategic partnership through a multi-year agreement. What do we need to bring AI into the industry at scale? We need strong data foundations, the integration of information technology and operational technology, and the ability to deploy AI while we keep running the operations without any disruptions. This is something no one can do alone. To scale AI, you need an ecosystem, and Capgemini plays a critical role in this ecosystem for us. They combine deep industry expertise, engineering and manufacturing capabilities, and the ability to integrate and execute transformation at scale. Most importantly, they have our trust, the full trust from Team Siemens. What matters is the ability to connect deep engineering with data and digital capabilities, infused with a strong industry expertise, and to operate at scale with full accountability for quality, safety, and delivery. That is why trust is essential. Why we chose to partner with Capgemini. This means that Capgemini will play a critical role in helping us standardize and modernize our technology stack and support consumer, restaurant, and company platforms, all powered by AI. For instance, we're continuing to connect thousands of restaurants around the world using edge computing. This is the foundation that will power AI and our IoT-enabled smart kitchens. Using the power of data, these kitchens will enable faster service, fresher food, and ultimately, more satisfied customers. We believe Capgemini's deep QSR expertise, their focus on engineering excellence, and their AI capabilities will help accelerate our platform transformation. Value compounds over time through continuous delivery and improvement. Ultimately, succeeding with AI is about turning technology into reliable and scalable execution. We're excited for the future as we work with Capgemini to build the most intelligent, scalable technology platform in the industry. Capgemini experts understand our tech stack, from industrial software to our automation technology, to our industrial AI solutions. Together, we're making AI an industrial purpose technology to make operations more adaptive and more efficient, and to create durable competitive advantage for our customers. Thank you for being the best partner we can imagine. I'm now happy to welcome on stage Group CEO, Aiman Ezzat. Good afternoon, and welcome to Capgemini's Capital Markets Day. I want to extend my thanks to all of you for joining today. The last time we gathered for Capital Markets Day was actually in March 2021, as we were emerging from the pandemic. Back then, we set ambitious goals for Capgemini midterm performance. We targeted 7%-9% annual revenue growth at constant currency and operating margin of 14% by 2025. I'm pleased to report that over 2021 to 2025, we delivered a constant currency revenue CAGR of 7.3%, effectively achieving our growth ambition. Now, this included strong organic revenue growth of 5.4%, squarely within our targeted 5%-7% organic growth range. Importantly, we steadily increased our operating margin by 140 bps to 13.3% by 2023, and sustained it even as we navigated significant market volatility. A resilience far stronger than in prior downturns when margins fell by up to 150 basis points in 2008. This strong performance delivering our growth target and improving and holding margins reflects the robustness of our strategy and execution, even under difficult conditions. Beyond the financial metrics, Capgemini has fundamentally strengthened its business over the past few years. We have significantly expanded our presence beyond the CIO, deepening engagement across the C-suite as clients increasingly see us as business and technology transformation partners and not just an IT services company. We have deepened our industry and sector expertise to differentiate our solution in addressing the specific needs of each market. Meanwhile, we have also accelerated our ecosystem strategy. Roughly 2/3 of our revenues are now tied to engagement with our main tech partners, highlighting how thoroughly we have embedded partnerships into our value proposition. Data and AI and cloud were already embedded as the core pillars of our digital strategy well before the current generative AI wave, positioning us ahead of the curve as AI becomes pervasive. The net result is a more resilient, more diversified Capgemini, one with stronger client intimacy at the executive level, richer domain and IP-based offerings, and a partner-augmented business model that can deliver greater client value and sustained performance. We have also delivered on our ESG commitments. Our progress on sustainability and inclusion has been exemplary. We have dramatically cut our carbon footprint, already achieving key targets well ahead of schedule. By the end of 2025, we reduced our Scope 1 and 2 emissions by 94% versus 2019, and slashed business travel emissions per employee by 70%, exceeding our 2030 goal of 80% and 55%, respectively. We also met our 2025 inclusion targets. Women representation in our global workforce reached 40%. We added ethics in AI as a ninth ESG priority, underscoring our commitment to responsible technology adoption. We truly believe that our ESG leadership is fundamental to long-term value creation and trust with stakeholders, reinforcing Capgemini brand and competitiveness in a world that rightly demands responsible innovation. These past three years have seen extraordinary uncertainty, inflationary pressures, shifting client spending patterns, and now the massive emergence of generative and agentic AI. Over the last 18 months, investors have begun to question the long-term economics of the industry in light of rapid AI advances. There is a widespread perception that generative AI could compress our industry's revenue, growth, and margins. These concerns have weighed on sector valuation. IT services stock have fallen significantly from their peak in recent quarters. My message today is that the rise of generative AI and agentic AI is not a threat to Capgemini. On the contrary, it represents a structural growth opportunity and a catalyst for margin expansion. We are confident that Capgemini will remain a long-term winner in this new era, just as we have successfully navigated every technology-led transformation cycle in our industry, from the advent of ERP and global delivery to cloud computing and SaaS and the digital wave. In fact, Capgemini core mission, helping businesses Transform and Operate more effectively, becomes even more critical in the age of AI agents. Today, we'll explain why AI is fundamentally a business transformation opportunity for us and for our clients, outline the five key AI-related value pools driving this transformation, and demonstrate why Capgemini is uniquely positioned to capture these opportunities. We will focus on economic logic and value creation, how the shift expand our addressable market, how we help clients realize tangible outcomes, and how we ensure that efficiency gains translate in sustainable growth and profit for both our clients and Capgemini. Let's begin with a fundamental question. What is agentic AI, and why it is a structural shift for business value creation? In simple terms, agentic AI refers to AI systems capable of reasoning, planning, and acting autonomously to execute tasks with minimal human intervention. Think of them as digital agents that don't just analyze or recommend, but can take action across enterprise systems. This is a profound new way of working. Instead of humans using software tools, we are moving towards human delegating tasks to AI coworkers or digital colleagues who can then carry out complex operations, coordinate with other agents and people, and continuously learn and improve. For enterprises, this could drive step- change improvement in cost, speed, and innovation, unlocking new levels of growth and profitability when deployed effectively. Harnessing agentic AI potential is not as simple as plugging in a new tool. In fact, the gap between what today's AI promises and what enterprises actually achieve has widened significantly over the last year. Many organizations have experimented with AI pilots or chatbots, few have realized significant value at scale. Why? Capturing tangible business value from AI requires more than just adopting the latest model or platform. It demands a comprehensive transformation, rethinking technology architecture, data foundation, governance, operating models, and the Human-AI interface. Deploying AI agents like a workforce introduces higher trust thresholds and complexity beyond traditional automation. In other words, to unlock AI's full value, companies must do the hard work of organizational change. Those that succeed will reap huge rewards, and those that don't risk falling behind. The good news, that our clients are ready to invest in this transformation, and Capgemini's expertise is exactly what they need to do it. Enterprise spending on AI is rapidly shifting to five broad value pools, which together cover the full cycle of becoming an agentic enterprise. Let me provide some overview of how they create value for our clients and growth opportunities for Capgemini. The first value pool is modernizing enterprise technology. The journey to AI at scale begins with trusted AI-ready core systems and data. Many large companies are held back by technical debt in legacy application, incomplete data integration, and aging infrastructure. AI is both a spur and a solution for modernization. AI capabilities like code generation make previously daunting modernization project more feasible, while modernized systems, cloud APIs, clean data, agile infrastructure, are non-negotiable prerequisites for effective AI. The second one is the new agentic AI tech stack. Beyond updating existing systems, companies need to adopt a new architecture for AI agents, what we call the agentic tech stack. This stack comprises new layers of technology to allow AI agents to plan and execute tasks across enterprise workflows. Deploying this stack is complex endeavor. It requires integrating off-the-shelf platform with custom development, addressing specialized engineering challenges like multi-agent orchestration, and navigating a shortage of advanced AI engineering skills. The third value pool is the new agentic control plane. As AI agents proliferate, businesses will soon be managing a digital workforce of millions of agents by the end of the decade. This explosion of AI workers is immensely powerful, without proper governance and cost control, it will be like employing a huge workforce with unlimited roles and unlimited payroll. This is why a new AI control plane is needed in every enterprise. It's a cross-enterprise management layer, the central nervous system of the agentic workforce, to monitor, govern, secure, and steer AI agents in real time. The fourth value pool addresses agentic products and services. The next wave of innovation will come through entirely new offerings powered by advanced AI, from radically accelerated R&D cycle to AI-enhanced software and physical products. AI agents dramatically accelerate product design by generating and testing hundreds of options. Additionally, the convergence of software AI with industrial automation promises cognitive production lines and intelligent supply chain that continuously adapt and optimize. Our clients recognize that this innovation will be key to future growth and competitiveness. Finally, the fifth one is agentic enterprise operations. Agentic AI's first real showcase is on the business process themselves across the front office, middle office, and back office. This goes beyond incremental productivity tweaks. It's about redesigning how work gets done by blending human employees and AI agents into new workflows. Budgets traditionally allocated to non-IT operational costs, for example, industry-specific core processes, customer service centers, supply chain operation, finance and accounting, are now potential sources of value and efficiency gains through AI. I want to emphasize the importance of domain expertise in this value pool. We strongly believe that reinventing processes with AI requires a deep understanding of how those processes run in practice. That's where the acquisition of WNS Holdings, known for its deep industry and process expertise across verticals like banking, insurance, travel, and shipping, significantly strengthens our position as a global leader in agentic AI-powered Intelligent Operation. Together, these five key AI-driven value pools span the full scope of enterprise AI transformation, from modern IT foundations to new business models and offerings. Each area drive significant services demand, and they reinforce each other. For example, modernizing core system is often a prerequisite for implementing agentic processes or launching new AI-powered products. If I step back, the overall addressable market for technology and operation services is set to expand dramatically as AI adoption scales. We estimate that the services market, including agentic-driven transformation and operation, will increase by around EUR 400 billion to reach EUR 1.9 trillion by 2030. Growing at a CAGR of 4%-5%. Crucially, much of this growth comes from net new areas of enterprise spend, capturing operational budget and transformation project that historically were not part of the IT services pie. We are not addressing technology cost, but the full operating cost of the enterprise. In other words, AI is expanding our opportunities rather than shrinking them. By helping clients achieve the next wave of efficiency and innovation, we unlock new revenue streams for Capgemini while providing immense value to our customers. Now, how will Capgemini capture these opportunities and convert them into durable growth and margins? The answer lies in our positioning as a go-to partner for enterprise transformation and operation in the agentic era. It's a natural evolution of our strategy. In 2025, we sharpened our vision as the leading outcome-centric transformation partner under our Make it Real proposition. That vision means we are committed to delivering outcomes for clients way beyond just technology. It's embodied in our dual value creation model, Transform and Operate, that we will share with you today. In both cases, we align our success with the client's success. We bring our pre-built asset AI platforms and industry solution to accelerate results. We leverage our ecosystem partnerships to ensure clients can harness the best of hyperscale cloud and AI platforms under one roof. This pivot from selling effort to delivering tangible results position us to capture and share in the value we create, enhancing our margin profile over time. Executing this strategy requires world-class breadth and depth, and this is where Capgemini truly stands apart. Operating with AI at scale is a profoundly multidisciplinary challenge. We have one of the industry's broadest sets of capabilities, from C-suite strategy and transformation consulting to deep technology and engineering expertise, from cloud and data infrastructure services to AI and software engineering, from managed operation and business process services to digital design, functional domain expertise, and robust change management. In short, we can help a client both engineer the solution and transform the organization to use it effectively. Few can credibly make that claim. What's more, we can draw on a powerful ecosystem of partnership to access the best models and platform for our clients while ensuring we remain agnostic and trusted business and technology partners. We are co-innovating with industry partners to address emerging needs around data sovereignty, security, and compliance. Beyond agentic AI, we are also riding the wave around sovereignty and defense, which we'll address in a future setting. In summary, Capgemini enters the agentic AI era from a position of strength, resilience, and adaptability. We recognize the concerns swirling around our industry in light of AI, but we strongly believe the future will belong to those who help enterprises harness AI's full value, not just as a cost efficiency lever, but as a catalyst for end-to-end business transformation and new growth. Capgemini is that transformation partner. We are helping clients reimagine their businesses from their technology foundation to their operating models, to their product, so they can thrive in a world where humans and AI work together at scale. By doing so, we are expanding our role, capturing new value pools and ensuring that Capgemini own growth and margins remain robust for the long term. Finally, let me turn to our financial ambition, which underpins our confidence in balanced, attractive growth, and value for our shareholders. We are targeting a constant currency revenue CAGR of 5.5%-7.5% to 2028. On the profitability side, we are introducing a new all-in headline metric, adjusted operating profit, which is equal to operating profit before acquisition-related expenses. We do target an improvement between 2025 and 2028 of 130-150 bps. On the organic free cash flow, we aim to generate more than EUR 6 billion over the period. This will result in double-digit shareholder return over the cycle. The group is entering this agentic AI era as a business transformation leader with a clear vision, the right capability, a proven model, and credible financial discipline to deliver profitable growth. We are tremendously excited about the opportunities ahead, which address the full operating cost of our clients and expect further growth acceleration beyond the transition phase to the new agentic world. With that, I will conclude and say again, this new phase of transformation, driven by agentic AI, is once- in- a- generation chance to create value. It plays directly to Capgemini's strengths, and we stand ready to make the agentic AI opportunity real for our clients, and doing so to drive our own growth, profitability, and long-term value. Thank you for your attention. Thank you very much, Aiman, for your introductory address. We will now have the first of several fireside chats this afternoon, and I'm going to ask you to stay on stage as you'll be gathering the first client perspective with Barclays Bank PLC Chief Client Officer, Stephen Dainton. Stephen Dainton has over 30 years of leadership in global capital markets with expertise in equities and derivatives. As Chief Client Officer, he focuses on holistic client franchise growth and building strategic CXO-level relationships across region. Let's welcome on stage Stephen Dainton. Stephen, thank you for being with us. Thank you for having me. Please. Thank you for being with us. It's a great pleasure to have you and, of course, to be your partner. You know, the whole subject today is about AI. I have a few question, and of course, we look forward to share a bit your thoughts. Investors hear a lot about AI as technology theme, and from Barclays vantage, is AI becoming a structural modernization cycle of large enterprises? Well, firstly, thank you for having me. There is no doubt we are all in the midst of. I think you probably need the mic. Sorry. Thank you. We're all in the midst of an industrial revolution. Every region, every industry is going through a material industrial revolution that is truly exciting. As yet, we have not even seen the application layer. I just arrived and heard you speaking very lucidly about governance, and I think any industrial revolution requires a material amount of governance to ensure that you're taking the appropriate path. If I look at it from the Barclays perspective, undoubtedly over the course of the last two years, we have embraced, A, the new technology and the application of that new technology within every single piece of workflow across every single one of the divisions in the firm. I think, of course, it is a structural modernization, but not just for us, for you, and every single industry on the planet. Talking about Barclays specifically, so how does this broader AI cycle connect with your own modernization agenda across the bank? Well, I think first of all, you've got to look at your technology debt, because before you put anything on top of something, you've got to make sure that it's sitting on a robust platform. I think, with Anne Marie Darling and Craig Bright, who are our two key co-COOs in the organization, we've worked through understanding where our technology stack sits today, and looking very closely at how we embellish that and bring the new technology on top of that. Number one, that's critically important. Secondly, you have to have a plan, and understanding that plan is really important. We've laid out, it's a note of public record, the Catalyst Program, that both Craig and Anne Marie have talked lucidly about, around a number of key pillars within the organization that address workflow processes and procedures, to simplify that journey. Very much it's about improving workflow and productivity, and it is extraordinary, even in my own job, the extent to which it is enabling that change, at a velocity which probably some of us could not have even imagined. It's happening very quickly, and we haven't even started on the application layer fully. The industrial change is real, sitting on a robust technology stack that enables you to compete productively in the future. In every single one, we have five operating divisions, in every single one of those operating divisions. You talk about AI and modernization have to go together. We cannot do AI without modernizing. Where do you see AI starting to really reshape critical banking workflows, and what does it take to scale that safely in a highly regulated environment? In markets and investment banking where I grew up, I'll give you an example. I started as an FX trader, with a pencil and a blotter and a telephone, a bunch of Quotron screens and a series of brokers screaming at you. You would walk your blotter to the middle office. The middle office would confirm your trades, that's the notional and the rates, and they would walk that back to the settlements office so that the trades could settle. We had a thing called a fax machine that was truly revolutionary. Some of you might not actually know what a fax machine looks like. It was very revolutionary at the time. We moved from a fax machine to PDFs, to instant messaging. What you're having right now is an industrial revolution in every single composite of the workflow. If you take it from my perspective, being able to look at, I'm presenting at a Capgemini event, I can look at exactly the returns on Capgemini. I can understand how much we pay them, how much they pay us instantaneously. Understanding who's engaged with who, what relevance do they have within the organization instantaneously. The improvement in productivity. Previously, I was using an example, in a previous session back at work, where we would be discussing in a strategy session, a comparative position relative to one of our competitors. You can look at that today very dynamically. If the data is housed inside your IR department with certain hierarchies, you can examine that in the meeting. The productivity improvement, even where I sit, both at a strategic level and at a tactical level, is critically important. As analysts, you will be embedding AI into your workflow. An analysis of your models, an analysis of the changes that you intend to make in the model, the sensitivity of that change in your model, and you can be looking at that instantaneously. It allows you to make better judgments and adjust that sensitivity with the human-in-the-loop, your judgment, that is better informing the data set that you are seeing. You talk a bit about how AI start moving workflows and become part of the operating model. How do you see the role of partners changing? What do you expect differently from a partner as we move into this new world? Well, look, first and foremost, it's strategy, the tactical delivery of that strategy. The first thing you have to understand is an enterprise. Understanding how an enterprise moves, how it evolves, where that strategy is going. I think that is critically important on this journey where you have a trusted partner. Remember, you are sharing an enormous amount of data that is unique to you. You need a trusted partner, as Capgemini are with us, on that journey that can both understand a strategic direction. That is where do you want to go top-down. Underneath that, even in a markets environment, the pre-trade and post-trade environment is changing so dramatically. You need somebody with deep expertise in your industry. Understanding how you enable that technology, understanding what good looks like, and bringing some of the parameters of what has worked from their judgment that could work inside your enterprise. I think it's really important, enterprise understanding, ensuring that you are a deeply trusted partner. Those trusted partners are going to become more substantial in the organization, not less substantial. I think that that's a very important piece. As you understand how the partnerships can get built, over the course of the next two years, that is going to be critically important. That's going to bring us naturally to Capgemini, right? As Barclays and Capgemini have worked together for many years and the agenda is now moving from technology delivery to AI-enabled transformation, what would make Capgemini the right partner for Barclays? I think first of all, you've worked with us, you've seen us on the journey. C.S. Venkatakrishnan l aid out three years ago a very clear new strategic journey. You were part of that journey. Strategies tend to be a continuum. They tend not to be rapid pivot changes, particularly in maturer and regulated industries. First and foremost, you have to stay safe. A big part of what Capgemini has worked with us on is ensuring that we are safe. Secondly, as you look to deploy that strategy, an understanding of what that strategy is and where you're trying to get to, the destination, is critically important. Capgemini, as you know, have helped us materially on that journey. Then also within any taxonomy of a business, most firms have different divisions, different products, really understanding that it is not linear. What happens in one division might not be relevant for another division, but it may have some application. I'll give you an example. We have five operating divisions, five different CRM systems. Bringing that together seamlessly is critically important. Within the compliance department, keeping us safe, the utilization of tools across this journey and understanding of the industry, actually a deep understanding of the industry from our trusted partners, is critically important. When we think about financial services, we think all the way from asset managers to insurance companies into banks, heavily regulated industries, ensuring that that strategic pathway is enabled in a safe way is materially important for us. Capgemini have advertised that and delivered that for us along the entirety of that journey. We shouldn't underestimate the magnitude of change that all of us are going through today. It is an incredible change-up. I think the point that you made earlier, the groups that embrace, deliver, ensure that their people are trained. The single biggest advantage for us, in our opinion at Barclays, is our people actually learning will inform future outcomes. The utilization of the new technology actually will develop new realms of opportunity that we haven't even discovered yet. They actually have to use it. Ensuring that you have that embedment in the organization, people are trained to use it, and they're not using AI as a sort of embellished Google search. Some of you are smiling because that is actually what happened when you first deployed some of the newer technologies inside the building. Really learning what this new technology can do in your daily workflow gives you all a material advantage, and by that I mean us all, a material advantage for what it's going to look like in a year or two. When you wrap around that what good looks like from someone like Capgemini, that makes you a long-term trusted partner. Sometime, people think about that as like, "Okay, we do transformation, we are done." Do you think this is a one-time thing that basically you transform, you introduce agents, and then things are going to operate on their own? Or do you see a continuous path of evolution in that new world? I've never seen a transformation that's ever done. I think this is a rapid evolution and it will go through many cycles. It will go through many cycles. Those cycles will adjust along that journey because as each industry discovers the deep relevancy of this changing technology for them, they will be able to adjust their model to adapt to that. That's why I make the point about our own employees have to adopt this technology because we're going to learn from them. I think that's really the powerful thing. The models that evolved out of the internet era evolved rapidly. There were many firms in 2000 and 2001 that said that they were already internet- adopted. If you look at the number of business models, fantastic companies that were internet- ready in 2000 and 2001 that no longer exist or were materially taken over, it is because they did not embed culturally inside their building an understanding of what that could do to their business model. I think that this is a rapid evolution. I think this will not be a two-year or a three-year look at what we've done. Every single business model, I don't say this lightly, every single business model is going to go through a rapid evolution. There will be very substantial winners, as there was through the internet age. There will be very substantial disruptors who adopt the technology quicker show that they are more nimble on a technology stack of relevance that is safe. There will be large institutions that also adopt change and rapidly evolve their business models. They will be the winners over the course of the next five and 10 years. Any last thoughts, things to watch for, or concerns as we deploy this huge digital workforce across enterprises? Look, I worry constantly in a heavily regulated industry about governance. First and foremost, we have to demonstrate to all of our shareholders, our clients, our deposit makers, our regulators, that we're a safe organization. Staying safe on any evolution, a rapid evolution or a slow evolution, is something that you have to have as the first core premise, staying safe. I think that I constantly worry about anytime that you're enabling a large data set to be shared, whether that's inside a building, inside a division, a TopC o, you need to make sure that your rails are very solid and that the doors for entry are very solid. I think governance from my perspective, is mission- critical. Secondly, adoption. We have 90,000 employees at Barclays. If 2,000 adopt it, we're going to move slow. If 10,000 adopt it, we're going to move faster. If 50,000 adopt it, the learning capability from the data capture of usage is going to allow us to adapt our operating model this year, next year, and for the next 10 years. Adoption becomes a very important factor in the winners and the losers from my perspective. You get there when you have trusted partners showing you that direction. I use this example if any of you live in London, they introduced a 20-mile-an-hour speed limit. That's a guardrail. That was a guardrail. Many people exceeded that 20-mile-an-hour speed limit because they hadn't actually looked up and seen the speed limit. They very soon learned the lesson. Ensuring that you have appropriate guardrails that people operate inside, when you have this number of agents that are going to consume the data and take direct paths to a conclusion, and that is why the human-in-the-loop becomes critically important. I think this is a truly exciting phase for all of us in every single industry. I think I'm lucky to be involved in it. Thank you for being involved with us. Stephen, thank you. Thank you for the great partnership. Of course, looking forward to make that AI the most powerful bank leveraging AI. Well, thank you very much for having me. Thank you very much. Thank you. Thank you very much. Thank you very much, gentlemen. I would like now to invite Fernando Alvarez, Chief Strategy and Development Officer, to tell us more in detail about the market opportunity and how AI will drive business value creation. Welcome. Good afternoon to a warm London for a change. Last night, I was having dinner with one of the fireside that we will have today, and as we were getting involved in the conversation, I reflected after I came back, and I start realizing the crux of the conversation is how fast everything is moving. It is so fast, and we were discussing with one of the AI- native providers that even they cannot keep up with their own models. I think right now what we need to understand is that we're in the middle of a structural shift of enterprise creation. The purpose of our conversation today will gear around a business conversation. It's not exclusively all about technology and IT. This is a fundamental change that we're seeing in the conversations that we're engaging our clients and our partners. It's a conversation about transformation. It's a conversation about how to handle people, assets, intellectual property, an agentic workforce, and the ability to understand the industry, the domain, and the functions that you're going in order to enable growth. This makes a big difference. If you close your eyes, you tell me how many people in the industry can navigate these waters with the trust and confidence that it takes to make this happen? You see here, this is not the first wave of innovation and transformation we go through. I remember the days of client server, everybody telling you, "Your days are numbered. You're going to be out. You're not going to be needed." Came the days of outsourcing, offshore development, onshore development, the days are numbered. How are you going to adapt? Came the days of cloud and digital. I remember when it all started, everybody was talking about, "Oh, digital is just proof of concept. It doesn't scale. There's no money to be made." In every single wave, we adapted, we became resilient, and more important, we excel. What is the difference, in my introspection from the conversation I was having last night over dinner, to these waves? As I told you, things are moving very fast. In the prior wave, we always have time to adapt. Here, we have to move. The clients have to move, because the objective is the ability to address business value creation by increasing shareholder value. That's the conversation. That entails moving into the ability to deal with structural rotation of that growth that we're going through, the wave we're in, to the new wave we're confronting. Yes, that implies secular growth, and that implies also some compression. That's part of the game. The question is, what's ahead? What it takes to be ahead. What we intend today is to cover with my colleagues all the value pools that Aiman has pinpointed out to you, in which we see the upside. That's the purpose of the exercise. The most important, for me, personally, in being in this industry for so long, is the ability, for the first time, to grab a total addressable market quicker, faster, and better. I will have people, assets, and agentic workforce. That is what's going to catapult me to grab that total addressable market. Rather than, "Oh my God, I don't know how to do this," we are embracing it as fast as we can, at the speed that the industry is moving, because we have that industry domain expertise knowledge to help our clients. Let's go quickly one by one. Value pool number one. There is absolutely, everybody's enchanted with the tip of the iceberg. Everybody wants to talk about agents and agentic. Everybody gets excited. You know what? In order to enjoy that tip, you better have solid foundations. That's how we started. If you don't have your foundations solid, if you want to take advantage of 2026 technology with 2016 foundations, good luck. Somebody has to tackle that. Value pool number one. Value pool number two. In order to do that, it will require a complete reset of the technology stack. What does that mean? It means that things cannot work in silos. The success of embracing agents and an agentic value architecture, it's based upon data. If the data is segmented and siloed through an enterprise, good luck. Somebody needs to put that house in order. Value pool number two. Value pool number three. When you do this and you embrace this technology, agents become a workforce. If you've now hear, those of you who were at SAP Sapphire in Orlando, SAP has an agent control, AI control plane. Workday has an agent control plane. ServiceNow has one. Microsoft has one. AWS has one. You know who is going to govern and manage that workforce? Somebody has to, because all of the sudden, these are not people informing anymore. These are not artifacts informing. These are acting autonomously. You need guardrails, you need cybersecurity, you need guidance, you need orchestration. The reality of the business that we're in, people still have legacy applications. There are some people still running mainframes. There are some people still running COBOL, besides the SAPs and the bespoke development that is happening. How do we orchestrate that? Value pool number three, the agentic control plane. It's not all about technology. I told you at the beginning that the conversation today is our business conversation. Everything needs to land. What are the playing fields? What I told you earlier, industry, domain, functions, knowledge. How do you enable? How do you use this to make this happen? Here is where everything that we have comes into fruition. We basically land in value pool number four, playing fields. Products and services. I hope now you start putting together all the decisions that we have been making for the last years as we have been building towards the momentum going forward. We have a Capgemini Engineering. That's a value pool of a wallet spend waiting to take advantage of this opportunity. The ability to bring the physical and the digital together. There is so much data to go after. There's so much opportunity to build bespoke applications, leveraging the different models and the different platforms that is there for us to grab. You will hear a testimonial shortly after about one of our clients and how it is making happen. Value pool number four. Value pool number fifth. Here is where you all wonder, I remember not long time ago, when we announced the WNS Holdings acquisition, you said, "These guys are crazy. They're going after the BPO market space, BPO market." No. Trust me. The reason we did what we did is because we saw it coming, we came to the conclusion that the best place to showcase agents and agentics is in the business process space. That was the bet. The bet was to move from a talent-based model to a consulting-led, AI power, technology, digital-driven model. We made the bet. Today in the enterprise process transformation, it's probably our biggest upside. It was the right bet, and we have built the momentum towards that. Value pool number fifth. As we have been learning, interacting with partners and clients, this is not an electrical switch. Everybody thinks lights on, agents lights off, agents on, no more people, less people. It is a combination. It's a journey. It's an evolution. We are evolving from a talent-based model to a consulting-led knowledge domain functions. Do not forget that. Knowledge and domain functions combined with the ability to leverage the power of AI and take all our heritage of technology and digital. Combine them all together, plus advisory and plus engineering, we're catering every single wallet spend pool in the market. We're going about it, not only our technology is about resolving business problems, enabling them with technology. That's how our TAM is going to increase. That is why we're so optimistic about it. Yes, we will invest in assets. If you look, and you will hear from my colleagues, how it is we structure and the reset, the technology stack reset, that contextual semantic layer. That's where I'm going to build my assets in order to increase my value of my services. I will do it by industries, by domain, and by functions. I need to understand the pain point of my client, and I need to have the ability to manage them and orchestrate them appropriately. As Aiman was pointing out, slowly, quietly, one step at a time, we have built a strength of a very solid ecosystem of partners. It's very easy to say, I have a one-to-one relationship with a partner. I'm the number one because I generate enough influence revenue for them. When you tackle industries and business problems, you need an ecosystem of partners that can speak the language of that segment of that industry to resolve a problem. That is what slowly we have been building and enabling the company to act. What we have done, first of all, every single decision, every single acquisition had a purpose. Syniti. Syniti is a asset-led service company. It was the ability to understand how by leveraging an asset, I can create enough stickiness to elevate my value creation in the services I provide. Focus on Data and AI. Focus on the migration from Premise- to-C loud. Cloud4C. We made the bet sovereignty. Defense is a growth engine. We make the bet that the world is not only about public cloud. It is a hybrid cloud environment. An asset base, totally AI-enabled, managed service platform built on open source. It allows us to deliver quicker, faster, better, and address the hybrid environment and the sovereignty environment, and yes, the foundations to enter into a mid-market for those who are asking us to help them to get into that mid-market. The last one, WNS. I already told you the logic behind it and why we did what we did, and it is the foundation of what we call Intelligent Operations. The consulting part. Now you see the relevance of the decision we made to bring Capgemini Consulting to Capgemini Invent and bring it closer to the decision-making process of when people are making decisions in the industry, tightly coupled with the advisory work, with the technology work, which is the foundation of the conversation we're having today now that we're embracing agents and agentic. It is crucial for us to make that happen. Finally, the ecosystems of partners. I told you earlier, 68% all our bookings last year gravitated around an ecosystem of partners. Quietly, slowly, and focused on industry knowledge and expertise. Sorry. Now we're making some very acute decisions right now. You're probably going to get bombarded by everybody telling you, 'I have so many FDEs. I got 30,000 FDEs. No, I have 100,000 FDEs.' I encourage you to go and find out what an FDE means. Ask Palantir, ask OpenAI, and ask Anthropic. The skill sets required to their definition is something that we all are working on. There's one thing you need to understand and why we are working closely with all them. An FDE, one quality it has, it has access to the source code of the models. Why they were forward deployed? Yes, they're forward deployed to the client because of that knowledge, that unique, that talent, and that skill allow us to deliver the final outcome quicker because they have access to certain code that nobody else has. That's why it's so difficult to replicate. That's why we decide that we will embrace the model, we will run with the model, but we will call them Outcome Deployed Engineers because our objective is to deliver an outcome to a client. We will embrace the FDEs of all of them as one to deliver that outcome. You will see series of families of Outcome Deployed Engineers to move into this new agentic value architecture. Decision number one. Decision number two, we acknowledge that the skill set requires are very unique. We are launching AI enterprise hubs. The first one we launch in Google Cloud Next 2026 with Google. Next ones are gonna be OpenAI, Anthropic, Microsoft, AWS. The ability to triangulate among themselves by picking industries and swim lanes to scale our people and our teams according to the skills required. Finally, we are investing in agentic control planes, but we will laser focus with our portfolio teams in a very industry approach in order to be able to take those problems. I think needless to say, we have what it takes to deliver. Then close your eyes and tell me who has the consulting capabilities, industry knowledge, domain expertise, scalability, the ability to address engineering and R&D, the ability to deal with technology and digital, and finally, the ability to scale this to the levels that you see clients' testimonials today that requires for this to work and manage an agentic workforce combined with people and asset to do effectively and efficiently. Tell me how many of those are there, how many clients we'll need. There will be, in the end, a lot of help. We will be there. One thing that we will focus today is to tell you how. We will transform or Transform and Operate. We will partner with clients by helping them transform, and we will partner with clients, those who want us to transform. By the way, can you operate this for us end- to- end? Here is where Intelligent Operations shines, and here is when Intelligent Industry shines in those areas in which it merits to go after those industries. What we will do today for you is to go now value pool by value pool, explaining to you why we believe that in each one of these value pools, that I dare to say we're the first one defining each one of these value pools in the market. We will translate those to our best knowledge of what that total addressable market looks like. With the confidence, with the enthusiasm, and the spirit to tell you that we're here to stay, and there is a big future ahead of us. This level of energy is not because I’m here today talking to you, it’s because this is the best thing that I’ve ever seen happening in our industry. We will embrace it as fast as we can. Thank you. Thank you. Thank you very much, Fernando. I'm going to ask you to stay on stage as we're going to have for the last section before our first break, we'll hear of our partner perspective. Perhaps, since you're going to lead this conversation, I will let you introduce Thank you. our partner. Thank you. Earlier in his career, our guest worked in private equity and investment banking at TPG Sixth Street Partners and Goldman Sachs, then co-founded Mainstay, the enterprise business spun off Opend oor in August 2024. Today, he drives global business with governments, enterprises, research ecosystems to accelerate responsible AI adoption and large-scale infrastructure deployment as Vice President of Global Business of OpenAI. Please welcome to join me, Nate Harbacek. How are you, man? Friend. Good to see you. Good to see you. The first thing I have to say, thank you, because this gentleman landed from San Francisco yesterday, and he's departing back to San Francisco at 6:00 A.M. today. Thank you very much for being here. It's always a pleasure to see you. Likewise. It's a pleasure having me. We move quickly, but that's part of the fun. Trust me, you move fast. I have a series of things that you and I have discussed. Let me ask you some questions to see how it flows. OpenAI is known for its pioneering role in pushing the AI Frontiers, but what we are seeing now goes even beyond that to our AI systems, agentics platforms, and enterprise-wide deployment. From your perspective, how should investors understand OpenAI's evolution and ambition in the enterprise market? It's a great question. It's really great to be here talking about it with you guys because I think you guys are stakeholders in that with us. I think when you think about our role, our job is to push the Frontier of Intelligence. I think you see us doing that with the models themselves, whether it's Codex and coding recently, the work that we've done with Rosalind, our biology model, and then even last week, there was a publication that came out where we proved a mathematical hypothesis, one of the Paul Erdős problems, that hadn't really been solved for 80 years. You see us pushing the Frontier of Intelligence every day, and the mission and the mandate of OpenAI is to push intelligence for the benefit of all humanity, and so that is foundational in and is our bedrock. What's interesting, what we've seen in the past 18 months, is that is definitively not enough. You have to be bringing capability not just to a human, but to an enterprise. Whether that is scaffolding, whether that is the harness, whether that is the Codex product itself, we're starting to move very rapidly into building the platforms, the agentic systems, the context layer and the understanding to really bring the technology into the enterprise. Agentic AI is set to drive structural shift. This is, I think, a given that we all have accepted and we're looking forward to it in how enterprises create value at scale. Ultimately, the challenge is how do we scale it and how we create that value. Moving from technology adoption to end-to-end enterprise transformation. From an OpenAI vantage point, how do you see this market transforming? Yeah, I think we're still early when you think about it. To me, the analogy is kind of crawl, walk, run. The first phase in adoption was, in many cases, amazement. Really touching the technology for the first time, using it for writing or answering a question, or enabled search or otherwise. It was single player, individual use, individual productivity. The walk, which we've seen over the past 18 months, has really been about individual workflows, creating an agent that matches the capability of a given human to do a task that existed before, and effectively superhuman skills and augmentation. The run is going to be really embedded agentic workflows. Production-grade enterprise deployments of something, taking the technology and embedding it deeply in the enterprise with the context, the controls, the permissioning, and the skills that it needs to operate, and agents that are specifically not linked to an individual person pushing the technology forward. You'll start to see entire workflows re-architected around an agentic workflow. Doing things that were either previously not capable or you had to do them in a different way, starting with human-in-the-loop, but in time, even without human-in-the-loop, at a scale and a capability level that we haven't really ever seen before. Then I think sprinting is going to be when the models themselves are able to figure out what to do and able to take on problems for a company or all of humanity that have never been solved before. We're going to see the models themselves in the next couple of years get to the point where they're capable of self-architecting new solutions or pushing capabilities forward in a way that has previously been impossible to understand and to me, that's the evolution of what you're going to see. While the breakthrough potential of AI is widely recognized across all industries, I think we have been addressing that. The key question is how fast enterprise can convert that potential into scale value, which is essential. From your perspective, what are the main barriers to deploy AI at an enterprise scale? There are a couple. I think the first is, it's a new way of thinking. The models themselves are incredibly intelligent today. In reality, that Frontier in terms of model capability and the gap to what they're being used on the enterprise is wide. The drivers of that gap are, you have to harness them the right way. They have to be embedded in the enterprise very deeply. You have to give them context. Our Frontier platform has a context layer that is a knowledge base and just connected understanding of the entire context of the organization, system of record, history, knowledge, internal encyclopedia of terms, understanding of workflows. You have to give them appropriate compliance and oversight and governance. You have to do the things to the model that allow it to sit in the organization with enough context to be capable. When you do, when you take the model and bring the organization from this lower level to the higher level and put it on the Frontier, that is when the magic happens, that is very hard. There's an analogy we use that in many ways, deploying this technology for the first time in a traditional enterprise is in many ways akin to open heart surgery. The models will get better in time. That surgery will evolve from open heart surgery into something that you could do efficiently, and it will exist. That gap is broad, and to do that today requires sophisticated expertise, whether it's FDEs, as you spoke to earlier, or individuals that are very capable of understanding the organization and taking that context and organizational structure and layering it into the model. That brings us directly to our partnership and the role of Capgemini as a founding member of the Frontier Alliance. From an OpenAI perspective, what role does a partner like Capgemini play in converting from Frontier AI into a measurable enterprise value? Necessary is how I'd start. I think our role, and I alluded to this in my first answer, is to push the Frontier of Intelligence and model capabilities and to build the products and the platforms upon which the technology will sit. We need partners that are capable of understanding industries deeply, that have the trust of customers, that are able to explain to customers what this technology is as it's rapidly changing. They have the credibility to go do a transformation, and then they have the scale to help us bring the technology to the world, which is our goal. Even today, OpenAI has approximately 4,500, maybe 5,000 people sitting inside the organization. We are going to need an army of resources that both understand the industry and the customers that we're working with, who can help us bring this technology, embed it deeply within the organization, and then start driving commercial outcomes for customers. That is how we think about the partnership. Let's get into another topic, because I remember when it started leaking into the press that your formidable competitor, or maybe yourself, will be launching your own deployment companies, that that was the end of all of us, and that will be it. You basically launched recently the deployment company, which Capgemini is both an investor and a strategic deployment partner. How should investors understand the strategic importance of this move that we make with you? Thank you for that, first and foremost. There's an African proverb that I've used a couple of times to describe how we think about the deployment company, which is how we think about it. I don't know if it's how the entire industry thinks about it, but we very firmly believe that if you want to go fast, go alone. If you want to go far, you go together. The deployment company brings together 19 investors and a number of strategic partners that we believe will help enable us to deliver the technology quickly. I think it goes back to my previous answer. We have this amazing technology that is infinitely more capable than what people are seeing it adopted today in the enterprise for the reasons that we discussed. The way that I think about this is, you guys understand your customers and the industries that you operate in far better than we will for the short to middle term. It is your job to understand what customers need, to articulate that very clearly to us, and us conversely, to explain the technology to you, and then to develop solutions for those customers in those industries that are revolutionary. We are going to need every resource that we can find to bring the amount of transformation to the world that AI has the potential to deliver, whether it's for drug design or in the energy space or in financial services or just in government efficiency as it will continue to exist. To do that, we have to do that in partnership, and the deployment company was our foray into the world and saying, "The technology is ready to be deployed in the enterprise at a scale that we haven't really ever conceived before. There are hundreds of billions, if not trillions of dollars of transformation that will happen in the enterprise space. Come help us. Come partner. Let's grow what this technology can be together and bring it to the benefit of all humanity and to all enterprises. As an anecdote, I've done many M&A transactions. I had the pleasure of negotiating with Nate, which was very little negotiations. It was, for me, fascinating because we had the opportunity of meeting at your offices in San Francisco at the speed, there were 19 different investors, the speed that you and your team was able to architect and the decision-making process that was made and the concessions. We had some concerns, they were accommodated. Our main message was, whatever this ends to be, it has to be seen. We need to be part of something that provides the value that you're articulating. They always commit to that, and they always have stick to that. For me, it was a pleasure, is a pleasure to be associated with that effort. Again, from our part, thank you very much for the opportunity because I know you were very oversubscribed. You pick who you want to go fight with, and thank you for allowing us to be part of that process. Just to wrap up, if investors here need to keep one key takeaway or take one key takeaway from our discussion and our partnership as the enterprise move into deployment phase, what should it be? I'll address that in a second. I'll go to your previous point just very quickly. Part of selection is based on what we hear from customers. Customers understand who is capable of doing this. There were a number of customers that were working with Capgemini that said they will help you deploy faster and better in the enterprise. Part of that choice and selection, subscription aside or otherwise, is based on ability to deliver value to customers. For that, we are very grateful, and we're grateful for the support. In terms of takeaways, I think we're at a really interesting inflection point right now. I would say OpenAI has been around for a decade. A lot of that was innovation and amazement and proof that you could apply more compute to a problem and the models would get more intelligent. We're at this really interesting inflection point right now where we're moving from amaze, individual productivity, single-player tools to now entirely agentic systems. What we need to see is impact and meaningful deployment in the enterprise. If we do, I think the future's going to be an incredible mix of technology, human enablement, capability, sustainability, and then just every week we're going to see magical outcomes, whether in the drug design space, energy, technological systems, computer design, or otherwise. Every day inside of OpenAI, we see magic. I think our job right now with partners is to bring that magic to the world. We're at this point where it's going to shift from individual use and amazement to real enterprise deployment and scale, and I'm very excited to be a part of that and grateful that Capgemini is here to help us. I really thank you very much for you being here. Yeah. Thank you very much, gentlemen. Thank you very much. I suggest we now take a break and we'll reconvene here in 15 minutes. Like that? Okay. Welcome back for the second segment of this Capital Markets Day. I have the pleasure now of welcoming on stage Roshan Gya. He's the CEO of Northern Central Europe and Chairman of Capgemini Invent, and he's going to address the next two items on our agenda. First, agentic AI technology and governance, and second, agentic AI product and services. Roshan Gya. Hello everyone. In this section, I will now deep dive into the value pools that will shape the future growth of Capgemini Group. The slide that you see here has been presented by Fernando, by Aiman, will be the compass for this section. We will now start with the enterprise technology modernization part. Every enterprise today, they want to be agentic. First, before, what they need is to become AI-ready. The reality today is that most enterprises are not ready. Over decades, companies have accumulated massive technical debt, monolithic applications, fragmented data, obsolete architecture, challenges around interoperability, and the blunt reality, and we've heard it today, you cannot scale AI on top of this technical debt. This is why we, Capgemini, believe that the very first major value pool in this new AI era for us is modernization. What has changed today is that GenAI, agentic AI itself, now makes modernization faster, cheaper. It is economically viable with strong business case, and it becomes mandatory if you want to go agentic. Programs that were previously considered to be too expensive, too risky, too complex, are now becoming feasible at scale. This include COBOL modernization, mainframe modernization, data foundation projects. Most importantly, what I want you to note is that this is not a short-term cycle opportunity for us. AI today is not eliminating enterprise complexity. It is increasing the vital need for transformation underneath. Fernando mentioned the iceberg. Very often we look at the tip of the iceberg, but below, when you look at all the complexity that has to be managed, it's a lot of work. This is why, for us, this value pool, we believe that we are entering a multi-year modernization super cycle for a decade. Today, clearly, we see a growing funnel in North America, in Europe, on all these modernization topics. Now, let's move on to the value pool two, the agentic technology stack. Since end of last year, we have seen a paradigm shift. A shift has operated. We have moved from AI that advises, generate content, create photo, even generate recommendation based on your profile to go to a good restaurant, to AI that executes. These are AI that execute workflow, interact with a system that reasons, that plans, and can act autonomously. Tomorrow, this will evolve even further. We are now hearing about physical AI with humanoids, industrial AI, autonomous learning system. This shift to agentic AI represents today a breakthrough in enterprise execution system. This is giving rise to a new hybrid workforce, where humans and agent will collaborate at scale. This is leading to a reset. A reset of the enterprise tech stack, because the agentic world will run on a new tech stack. Now to understand better why this reset, what it is, what's the rationale behind it, let's have a quick look at a video. For decades, every enterprise has followed the same formula. Humans decide, humans execute, systems support, systems record, AI provides insights. Until now, that worked. Something fundamental has changed. AI is no longer just assisting work. AI can now execute work. The current enterprise stack was not designed for AI-driven execution. This is where the stack begins to break. The fracture appears in three places. First, orchestration. Modern enterprise systems are still largely divided into silos. Humans bridge these gaps every day. AI doesn't think in silos but operates across the entire value chain. The current stack was never designed for this level of orchestration. Second, hidden logic. Most enterprises do not only run on explicit logic, the implicit lives inside people's heads. AI, operating at scale, requires that logic to be made explicit, consistent, and machine interpretable. Third, governance. Today, humans are the governance layer of the enterprise. They approve, validate, escalate, intervene, handle exceptions. When AI executes millions of decisions in real time, governance must evolve as well. It can no longer be manual. It must be embedded. Not after execution, but during execution. The challenge is not deploying AI. The challenge is that the enterprise itself was never architected for AI execution. It was designed for humans to execute work with software supporting from the sidelines. We are not simply adding AI to the enterprise stack. We are entering the moment where the enterprise stack itself must be reset. What is this new tech stack we are talking about? Think of it as the new operating system of the enterprise, the new OS. At the very bottom here, you have a data application and the infrastructure. It's a foundation. It's a system that runs the business today. On top of that, we need to add three new AI- native layers. We've been talking about context and semantics. This is the layer that teaches how the business actually work. Then you have the model and enterprise integration. This allows you to choose whatever model. Is it a large language model? An SLM? You can use several models and connect it back end to the front end. Then you have the action layer. This is the layer where the agents work. We orchestrate the processes. On top, there is the existing user experience layer, web, mobile voice, that will need also to be transformed to move more towards outcome-centric. What we see today is that technology vendors are building horizontal AI tools and technologies. The enterprise execution is deeply industry- specific. It's vertical. A bank, a pharma, a manufacturer, a telecom operator, they execute differently. Industry context becomes strategic. This is the semantic and the control layer here, it is key. The workflow are different, regulations are different, operational risks are different, and the economics are different. This is where Capgemini create value. Over decades, we have accumulated workflow knowledge across all industries, all segment, very deep operational process understanding, engineering expertise, industry semantics, domain knowledge. We at Capgemini, we bring contextualization to life. Our mission is to instantiate the technology in enterprise context to deliver business outcome. There is one motto for us, without context, there is no intelligence at scale in enterprise. The context is our first moat. Also, there is a question today is, it's not about, okay, which model you will use. You have seen, these models are super intelligent. We are not even leveraging 10% of what they can deliver. The question is how do you turn this intelligence, which is abundant, it is accessible, it's democratized, to business outcome at scale and create value? One of the biggest misconception today that we have is that the belief is that the value primarily sits in the model itself. Choosing the model, whether it's OpenAI, it's Gemini, it's Anthropic, it's Google, it's just the beginning of a story. In the enterprise stack that you see here, the model is just an ingredient. Okay. It is very important, but the value sits around it. Enterprise do not and will not operate on model alone. They will need architecture, they will need orchestration, they will need API, integration, semantic business layer, data foundation. What is our play as Capgemini in this value chain that has emerged? There is a new value chain that has emerged over the past years to date. We as Capgemini, for decades, we have been helping our clients in building and operating their IT states. Now in this new technology era, the new agentic era, vendor will provide components. This will not change for other new stack. We, Capgemini, we will architect, we will assemble, we'll put the glue on all the stack, we will contextualize it, we integrate it and operate it. On top of our core strength that you already know, process expertise, data foundation, integration at scale, transformation capability, we'll accelerate and de-risk this journey for our clients with our asset, our frameworks, our IP, with new engagement and commitment model. What I would like you to take away on this value pool are several messages. The first one, it's not a one-off migration. It's a continuous evolution. The stack is not like a one-time cloud migration and it's done. It will evolve. It will take time. There will be new features that will come also that are still not here. If you look at physical AI, all the new features that industrial company will have to embed in this stack, it's still not here. There will be new innovations. There will be the extended enterprise. Meaning, it's an evolution. Secondly, there will be no big bang. There will be no switch off and switch on moment. Traditional application and agentic execution will coexist for years, even for decades for some clients. As the previous transformation cycle that we have seen, meaning we have different profile of client and different profile of industries. You have clients that will be more risk-taker, that will have more appetite for transformation. There will be those who will be more conservative. Meaning, it will be an heterogeneous of portfolio of clients to serve with different tech stack profile and maturity. Important message is the context layer, which our partner, OpenAI, was mentioning, it is never finished. Business context changes. It will be required to be configured once, to be updated, product regulation, ontology, knowledge graph, everything will have to be maintained continuously. This create a durable value pool for us for Capgemini. The stack will keep on evolving for decades, and our client will require a long-term partner to help them in this reset. In a nutshell, technology provides the intelligence, and we as Capgemini, we are the bridge between technology and business outcome, as we have always been doing. Let's move on to our new moat, our third value pool, which is the agentic control plane. Fernando has been mentioning it, Aiman also. So far, we have been discussing about the new enterprise architecture. We've seen that we need to do a reset of a stack. Now when AI is moving from generating content to executing work, a new reality is emerging. This reality is today we will have to put under control these thousands, hundreds of thousands of agents. Same as for the semantic layer, without context, there is no intelligence. Here, without control, there won't be AI at scale. It is about the agentic control plane. By 2029, we saw the figure, there will be more than 1 billion AI agent that are expected to operate across enterprise. We need to imagine the future with a digital workforce that we will need to register. We'll need to understand their role, their skills, their access right, the definition of what they can do, they can't do, and finally, very important, the cost, the token consumption. As such, we mentioned that there is a first reset that has to be operated around the tech stack. There is a second reset that has to be operated, and this is the reset of the system of control and governance. Without this reset, it would be comparable, and Aiman always says that, to having a huge workforce with unlimited rights and unlimited payroll in your organization. Imagine how chaotic it can be if you don't put that under control. For our client, this agentic control plane is the CHRO, the COO, the CFO, and the CIO of the agentic workforce. It's not a technical overlay. It's not a dashboard. It's not just a layer. It is the central nervous system of the agentic enterprise. What about Capgemini now? How do we play on this control plane? Over the past two years, the discussion, the narrative on the market was, okay, ChatGPT versus Claude versus Gemini, who has the strongest reasoning quality? We know that now these model are far beyond what we can leverage right now. One thing is when you shift from adoption of ChatGPT and Claude from your phone that you are using, and you start thinking about integrating it in a complex and living enterprise architecture, the question being asked are completely different. Who decides what are the agents allowed to do? Who manages the agents across the enterprise? Who stop them when something goes wrong, and who is managing the costs? This is not a model problem anymore. The model, as I mentioned, is just an ingredient in the stack. The reality is that there will be Claude, OpenAI, Gemini, SAP Joule agent, you will have robotic agent, you will have Edge AI agent, you will have thousands of agent from several providers. Therefore, what is important for our client is that the strategic capability is the one that governs identity, permission, policies, orchestration, auditability, and costs. In this value pool, our mission is to help our client, as a trusted and neutral partner, to reset the system of control and governance. We will build it, we will integrate it, and we will help our client to operate it. As a conclusion on this part, our role in this new agentic era is crystal clear for us. As you can see, for our client, the new scarcity is not coding hours anymore. We know it. It is governed and trusted execution at scale on a new tech stack that has to be built with a new central nervous system, and how to make this operating system economically viable. We mentioned one important thing, it's adoption. It is to have a Human- AI Chemistry. Building this agentic enterprise is not a technology transformation, as Aiman mentioned. It's a business transformation. It's a systemic, complex business transformation, and we require a new playbook of transformation. We've released our framework of transformation, which is the Resonance AI Framework last year, is to help our client to move into this agentic enterprise. This transformation, when you look at it, we have been accompanying our client for decades in several types of transformation. It's both an art and a science. It will require to get the right balance between what is the right architecture, what is the right operating model, what is the right governance, and what is the right economics with the proper cultural transformation, blending human and AI to work at scale in enterprise. For us, our mission is to be the trusted architect of this new tech stack, to build it, and to operate it for our clients in this new era, and we Make it Real. Now, I've finished this part. We will now move to the next section, exactly. We'll move to the next section of a presentation, which is the next segment of a program. I will focus on the agentic product and service. On this value pool here, what you see is that we have the foundations. We are building the tech stack. We have a control plane. How do we create value on a play field? Now, when you bring all together this stack, the technology and data, and when we instantiate it on the agentic product and services, we have this. Six years ago, we have launched our Intelligent Industry play. It was a new concept on the market. At that time, we had a conviction. It was the future of industry is the convergence of the physical, digital, and biological world. It is about the systemic and concurrent transformation of both core industrial processes, engineering, supply chain, manufacturing, and the connected services and the product itself, making the product more intelligent, autonomous, with hyper-personalized experience. You can have the example of vehicles, network, medical device, industrial machine, infrastructure. When we look at the acceleration, what is happening right now, if you just pause and you step back and we project ourselves, what we see clearly today is that the compounding effect of technologies, I'm not just talking about agentic engine AI, because all the accessible technology around 5G, around connectivity, AR, VR, simulation, digital twins, when you blend it now with agentic engine AI and soon physical AI, robotics, humanoid, this unleash a whole new waves of value in this field, the Intelligent Industry field. With that, we can clearly envision a very different future in the way you, we, human beings, will move on the planet, will shop, will talk, will attend to our health, will work. This is opening for us, as it was mentioned before, the biggest opportunity for Capgemini in 60 years. How we will tap into this opportunity? The first wave of AI we've seen was about prediction. The past years, we were talking about predictive maintenance a lot. This was prediction. The second wave is about generation. GenAI, it's execution with agentic AI. The third wave, which is coming very soon, is around autonomous optimization, which will be catalyzed by physical AI. In Capgemini, we call it the Industrial Learning Loop. We will communicate soon on the market on this new value pool. What is important is that this value pool is not a generic AI opportunity. Understanding mission-critical environment is something that cannot be faked. We are talking here about connected defense, self-healing supply chain, agentic engineering for capital project management, dark factory. In the Intelligent Industry world, there is no tolerance for error. Mission-critical environment cannot be addressed with generic model. A factory cannot hallucinate. A medical device cannot guess. Today we have stochastic model like Anthropic, OpenAI, Gemini. They are not enough. Alone, they won't do the cut. To create value in the Intelligent Industry space, what we need is that AI need to understand clients' data. We need to understand the core process, the engineering constraint, the physical reality, with the regulatory framework. We need to do reinforcement learning. We need to do complex simulation. Today, Capgemini is the only player in the market that has this breadth of capability at scale. Engineering, scientific. We have 30,000 sq m of labs in Cambridge. On top, our Industry Dev, we couple it with our large-scale transformation team, design, tech, and data to capture this growth. In this value pool today, we are uniquely positioned to tap into these massive business opportunities. As a conclusion, this is my personal conclusion, is that in the two decades in the industry, I've rarely seen a shift of this magnitude of impact. I strongly believe that this transformation will create values for years, if not decades to come, for our people, for Capgemini, and for our clients. Thank you. Thank you. We now have another client perspective we're going to hear. I'm going to welcome on stage Étienne Grass, the Global Chief AI Officer of Capgemini Invent, and he's going to welcome on stage Olivier Charmeil, the Executive Vice President, General Medicines of Sanofi. Just a few words about Olivier Charmeil. He began his career in mergers and acquisitions before joining Sanofi in 1994. Over more than three decades, he's held series of global leadership roles across Europe, Asia, China, covering vaccines, general medicine, and emerging markets. Very recently, at the beginning of 2026, he served as Interim Chief Executive Officer. Today, Olivier joins us to share his perspective on how AI is transforming the healthcare industry. Hello, Olivier. Thank you for being with us for this important moment. I will begin with having your thought, your reflection on the shift that Roshan shared, especially the one on how we read at the enterprise stack. Also, this huge shift on the new generation of products and services that are coming. What is the perspective of Sanofi on that? Everything that Roshan has been saying resonates very much with Sanofi. We don't see, of course, AI as another tool. We see it really as a way to transform the way we work and truly across the value chain. Of course, we are in the industry innovation. Our objective is to bring to patient new drugs. It start, of course, on the research side, on the development side, because a significant part of what we do is developing and making clinical studies. Of course, manufacturing. Across the value chain, we see what's going to be the contribution of AI. What are the implication? The implication are quite clear on the technology foundation, which is making sure that we have trusted data, that we have integrated system, that we have, of course, the right governance and control, that if we want to leverage AI, we need to make sure that we are controlling and at the end of the day, we have the management human oversight. In a company that has been built by acquisition. To say that we have integrated system and we have trusted data is, of course, something that is not easy to say. The ability to orchestrate agent with human oversight is something that is key. The second dimension, when you make sure that you have the tech stack, is really the part related to business transformation. In order to have new product, new services, you need to make sure that you have, of course, solid foundation. It's about redesigning your processes. I would even argue across the value chain, across R&D, across manufacturing, but I would go even beyond. I think our objective is definitely not to digitalize the past. It's to reinvent and to build new processes that will allow us, of course, to extract the new value. It's really creating new AI-enabled products. I can take a few example of things that the human brain couldn't do. It's clear that using AI will allow us, I have a couple of example in mind, especially on the manufacturing side of the business, where we can give answer to questions that we didn't know even that we should raise those questions. I have a couple of example, in terms of yield improvements, because there were a ton of data, we couldn't do that. Once the foundation are in place, we can become very concrete, we can become very concrete especially on the R&D side of the business. Yes. That's my question actually. Let's focus on R&D. We do together the Act for Patient program, which an important program actually to speed up the clinical development. How do you see AI speeding up the access to drugs, making it faster, more patient-centric, too? This Act for Patients program is an important program for us. For us, I mentioned a little bit earlier, our purpose is to make sure that we bring innovation to patients. We bring innovation to patients, there are a lot of debate and a little bit theoretical debate. Should it be best in class? Should it be first in class? I think at the end of the day, the faster you move, I think the better it is. Making sure that we live, of course, in a very competitive space, and there is also this purpose on bringing new innovation to patients, so to make sure that patients have access to treatments. Where AI change fundamentally what we do is, of course, on the clinical development side. Here, in order to be concrete, we work collectively with Capgemini on the protocol design. You know that 80% of the cost of inventing a molecule are on the development side of the business, performing clinical trial, massive clinical trials. It might change, in the next 20 to 25 years, where I think there will be more and more synthetic clinical trial. For the time being, those clinical trials are done on humans. It's protocol design, it's making sure that you identify the right patients. We segment population in order to show the efficacy of the patients on the right population, it's getting, of course, increasingly important. At the moment, there is so much strain on financial expenses and healthcare budget, to make sure that the right patients are treated with the right drug. It's about patient identification. It's also about site identification. I mentioned a little bit earlier that in order to move fast, you need to make sure that you identify the right sites. There is competition across the industry to get access, for example, in the field of oncology, to make sure that you reach the right site so that you can enroll pretty fast. The last dimension of development, and which is very important, is the regulatory preparation, the regulatory dossier. There is a lot of work that is being done, and here it's AI, and we have a couple of programs in place, and we are starting to see things that are concrete. It's hundred of thousand of pages a dossier. Here it's obvious that AI is going to be transformational. Beyond that, it's not a process. It's a continuing of process. Making sure that we get out of those fragmented process to ensure that we are more data-driven at each and every step, to make sure that all those processes are connected. I take an example. If you reverse back and you realize that, for example, you are not enrolling fast enough, it might mean that you have not selected the right sites. At some point, we did in this industry what we call futility analysis, to analyze a little bit before the end of the clinical trials how things are moving. It might reorient the way you are going to recruit patients. With Capgemini, what we do here is really combining AI, data foundations, process redesign, and I think there are healthy debates from time to time because one of my observation, we are, of course, in an hyper-regulated industry. I have often said that I feel that the agencies are more conservative than us. You understand what I mean? To have people that can challenge us, of course, in a constructive way, I think has a lot of value. You see more and more AI factored into our day-to-day practice, and I think it's only the beginning. I think it will have significant impact in two direction. The first one is we will be able to move faster. You understand what I mean? The second one is, I think at the end of the day, it's going to be more efficient. You understand what I mean? It's not only a productivity gain. I think it goes beyond productivity. We have not been talking about the research part, which is also important, which is probably more 5 to 10 years. One of the issue of this industry is that our return on investment are far too low, because you develop drug and you realize very late in the process when you have invested EUR 100 and to say billion, that they fail in phase III. Here, I feel also that even if it's more long-term, better target identification, better understand of the biology, I think it's going to be very transformational, and we want to be part of that game. Thank you, Olivier. Sanofi is also really transforming the way you manufacture medicines, actually, and it's also an area where we work together on smart operations. It was actually highlighted by Roshan. How do you see the role of AI, agentic AI, physical AI, digital twins in, I would say, the design of more intelligent, resilient, and controlled production systems? More and more, and I started my career, I've been with Sanofi for 30 years, and of course we were in small molecules, a little bit on vaccines. More and more, of course, we are in what we call biopharma. What is very specific with bio is, of course, it needs to be scalable. It needs to be reproductive. It's much more complex than any chemical process. For us, at a moment of time where we are bringing to patient new drugs to make sure that we are able to scale capacity from the lab scale to something that is much bigger, is of course of paramount importance. Keeping in mind that we are in an industry that is highly regulated, and everything that we do needs to be documented. When everything is documented, we need to be compliant, of course, with everything that is part of the process, and the process is often known as a product. We have a very high ambition in terms of with our smart operation ambition. We see the next generation factories as being significantly different from what we have today. I'm just back from China. I've visited many manufacturing facilities. We are going to build a new manufacturing insulin factory, and we see both physical and virtual AI playing an important role. I was thinking that probably, it's not five years down the road, I think it's more two or three years down the road, we will have dark factories, we will have humanoid robots. One of the big things that we do in order to get prepared for agency inspections is definitely our own inspection. In the world of tomorrow, and I think at some point the FDA and EMA will do the same, we will have humanoid robots that will work, would spend a couple of hours, 10,000 of sensors, and you will get good understanding of what is the situation in your manufacturing facility. The last point is, of course, digital twins, which allow you to test what you're doing, to test the design, to make sure that you build something that is fully optimized. This is what we are doing. We have two manufacturing facility that also are twins, and one in Germany, two to one, and one in China. We want to make sure that we leverage our digital twins in order to make sure that we do right for the first time. Here, it's really about combining the expertise, and our expectation are very clear. It's not so much working with smart people that have good engineers, and Capgemini, of course, has full of engineers. It's really making sure that we embed those capabilities into the day-to-day operations. There is a big business transformation. The most difficult is not technology. I'm sorry to say that. The most difficult in a company like us is to make sure that we transform the way we work. Easy to say, much more difficult to do. That's exactly the question I was going to actually say. What you shared is that agentic AI touches on very critical part of the company business: R&I, R&D, manufacturing, data platforms. What do you expect for a partner like Capgemini to go from use cases to being a trusted transformation partner in your journey? I think it's a very important question. I am not expecting so much on the technology side. Of course, we know that here you are state-of-the-art. What we are expecting is a good combination between your technology expertise, your engineering expertise, but also life science. It is a combination of both that is very complex, that create a lot of value. We are, again, in highly regulated environments. We have very critical processes. Those processes need to be end-to-end, and you need to have an understanding of each and every step of the development of a new drug, and the manufacturing of a drug. We need more than a technology provider. We need really a transformation partner, and I have just mentioned the importance of changing things. What matters is really the life industry expertise, the life science expertise that you have. It is also a deep understanding of our context. Again, one of the key points for us is to make sure that we improve the productivity index so that we generate a continuous flow of new products. This industry is about innovation, making sure that you have a continuous flow of new product is of course very important. The last point that is important, I touched a little bit earlier when I mentioned not digitalizing the past, it is really making sure that we embed AI into our workflows, that we accept to change our workflows when it is needed, and this is something I insist a lot vis-a-vis my team. It is again, not digitalizing the past, it is making sure that we have the right workflows, data, and operations that will allow us to go to the next level. Last point, which is key, and I think AI can be really transformational here, it's the end-to-end capability. I think here, AI is unique. There are a lot of activities that were done manually in the past, and now, AI, of course, is truly transformative here. I think it's true across research, development, manufacturing, and I would argue even around commercialization and segmentation of patient. In order to be able to move fast, there is one dimension that is noteworthy. It's really the dimension of trust. Nothing is feasible without trust. I think it's true in any industry. I would argue that it's even more important in the pharmaceutical industry, of course. Why? Because we work on human. Warm thanks, Olivier. I think that has been very powerful illustration of several message that has been conveyed on the reset your stack, on the importance of the context, actually, layers, also on the need of a control plane on manufacturing critical activities. Stressing also the importance for us of having a deep industry expertise and, of course, of trust. I think thank you for being there. It's also a good illustration of trust. Thank you, Olivier. Thank you. Thank you, gentlemen. Thank you very much. We've seen how agentic AI creates value. Let's see now how it transforms enterprise operations. For this, I'm going to invite Franck Greverie. He's the Chief Technology and Portfolio Officer and Leader of Business Lines Capgemini. Franck Greverie. Hello, everyone. Now we'll speak about the value pool number five. Everything we have discussed so far matters, but you know for us, transforming enterprise operations with agentic AI is the biggest AI shift for the enterprise. Why is that? Because this is where AI stop being a promise. This is where the value of AI becomes clearly visible and measurable for the CXOs. It is key to understand that transforming enterprise operations with agentic AI is a business transformation. You know, it's not a simple technology upgrade to better access data with AI agents or to automate with AI agents a few tasks while keeping the same processes that were built for humans. Actually, when we are transforming enterprise operations with agentic AI, it means three things. One, we reinvent the business process itself, so as it can be operated efficiently by a Human-AI workforce. Two, we build agentic systems that operate the business, and it's very important to understand that. The agentic system is operating the business. It change how the enterprise works. It means systems that understand the context, make decisions, and take actions autonomously based on business rules and context understanding. Three, the result is not just efficiency. We create a step- change in business outcomes the company couldn't achieve before. Put simply, our clients are moving from a company run by people and supported by software to a company run by a Human-AI workforce. That's the shift, and that's where the value is. For Capgemini, we see two business opportunities, two distinct businesses. The first one, we call it transform, as it was mentioned by Fernando earlier. This is for clients who want the upside of Agentic Transformation, but they want to keep running the process itself. They don't want to outsource because sometimes, you know the process is too strategic for them, or sometimes they are not ready to do it. In this case, our accountability ends at go live. After that, we stay involved for maintenance and continuous upgrade. The second type of project is different. We call it Transform and Operate or Intelligent Operations. Clients give us the transformation and the operations. Why? They want to move faster. They see us as an accelerator, the fastest way to capture the value that AI makes possible. In this model, we are fully accountable for the value and for the outcomes. We have skin in the game. Our pricing reflects that. It's tied to result. One important point that was mentioned before, Intelligent Operation is not what you know as BPO. It's fundamentally different. Let me show you on four points, BPO on the left, and Intelligent Operations on the right. BPO was about cost saving. Intelligent Operations, it's about value creation. BPO was about process lift and shift. Intelligent Operations is about process redesign, and it's a move from a Human-based model to a Human- AI-based model. It's very different business. That's a different model. That's a different kind of value. Now, we have all been talking about value, where does it actually come from? Agentic Transformation unlocks two sources of value, these are not incremental gains. They're a step- change for the enterprise. The first source of value is business outcomes that AI now makes possible, outcomes that didn't exist before. Few examples. New customer experience, very different from what was known before. New revenue stream, faster execution, strong cash generation, and more Compliance and Quality by Design, and a whole new way of making decisions. The second source of value is cost optimization, specifically the total cost of tech and operations. Agentic Transformation change the cost equation. It's from mostly a human workforce to a Human-AI workforce with significantly lower cost, even after factoring in Agentic Transformation and AI tokens. Two sources of value, both transformational and a real step- change. Now a word on where we play. We go broad across horizontal processes on finance and accounting, HR, procurement, supply chain, customer service, and marketing, and across more than 10 industry-specific processes such as banking, insurance, travel, logistic, where we have deep domain expertise, significantly amplified by the WNS acquisition. Now to make it concrete, I will share with you three examples of projects that we are doing with clients. The first project is a transform project. It's a global bank wanted to reinvent its call center. You know, not the usual story, not adopting cloud-based call center technology or deploying a chatbot. They wanted something much bigger. They wanted an Agentic Customer Engagement Center. An agentic system that solve customer issues end- to- end with humans stepping in only when it matters, while obviously proposing at the same time, proactive services to their clients. They didn't want a tech upgrade. They wanted a business Agentic Transformation, but they were not ready to outsource the operations. Here is the picture before the transformation. Customer service was mainly reactive, almost no proactive engagements. Call center agents had limited context, no clear view of the customer, no clear view of their journey, and high attrition made even worse. You know what? We have all been there. You call, you wait, you re-explain everything to each new call center agent, and you have no idea when it ends. Here's what we unlock with the Agentic Process Transformation is first new outcomes, a much better customer experience through omnichannel, hyper-personalized engagement, and self-service that actually works. We target with our clients a 15%-20% lift in customer satisfaction. On top of that, customer service becomes an upsell engine, proactive, personalized. Here we target $300 million upsell over five years. It's very significant for them. 55% efficiency gain in the process operations, translated into EUR 150 million in savings over five years. This is a complete reinvention of the customer service with huge net benefits for the bank. Now, let me share a second example, this one on Intelligent Operations. You have an outsourcing dimension into it. It's one of the largest logistic company, they came to see us with a real challenge: managing invoice disputes with their customers. Every day, thousands of customer disputes because of wrong invoice, late deliveries, damage or lost shipment, or many other reasons, every dispute blocks cash collection. The volume was just huge, between 15,000 and 40,000 cases a day. The causes are complex, highly variable, across many languages. Historically, it was mainly a manual work. The result for the clients, only 55% of cases resolve, leading to frustrated customers on one side and growing write-off on the side of the clients. What did we do? We redesigned the process for AI. We built an agentic system to resolve autonomously disputes, we operate it for them. Here's what we unlock. First, business outcome. Globally, their case resolution moved from 55%- 85%. It's very substantial. A strong improvement in customer experience with a net promotion score up by 20 points, and EUR 80 million a year recovered on revenue leakage. Just imagine for five years, close to EUR 400 million. Second, it unlocks a global optimization of their cost of 75% efficiency gain in the process operations, translated into 40% net head cost savings across the entire finance and accounting function. You can see how important the benefits are just for One Finance process. Now imagine the potential at scale, and that's what I will show you on the last example. It's another example on the Intelligent Operations project, this time for a multinational car manufacturer. The scale, 100 countries, more than 10 global delivery centers, teams speaking 40 different languages. What was their goal? To gain a competitive edge over their peers by reinventing four functions at once and getting them to work together, finance and accounting, HR, procurement, and supply chain. What did we do? Same playbook as I've said before. We designed the process for AI, built an agentic system across the four functions, but with a value chain approach across the process, and we operate them. Here's what we unlock. First business outcome, around EUR 1 billion over seven years. A very substantial outcome. Let me give you some ingredients of it. EUR 250 billion in supply chain, EUR 150 billion in procurement, EUR 100 billion in warranty optimization, leading to less revenue leakage. Second, on cost optimization, a 50% efficiency gain in the process operation, translated into EUR 1.3 billion in saving over seven years. Look what happens when we combine four functions. Total enterprise value created over EUR 2 billion in seven years. That's the power of Intelligent Operations with a value chain approach. You have seen the example. You can see this opens for us a whole new market, a major opportunity for Capgemini. Let me tell you three things about that new market. First, the market is already strong today. Clients are investing, and we've seen an explosion of our business opportunities over the last few months. Our pipeline of business opportunities already exceeds EUR 12 billion, and it's grown 30% year- to- date. Our clients are investing for two main reasons. One, this is a breakthrough in value creation, and you have just seen the proof on the three examples I've presented. Two, and I think it's a very important point, agentic system give clients a new kind of flexibility, a more variable cost base, the ability to scale operations up and down with market conditions. Second point on the market, it keeps growing. Demand will only accelerate in the coming months. Imagine our clients are already seeing value on a few processes. Now they want more. They want to extend it across most of their processes. They want tens of thousands of AI agents working for them, and for us, that's tens of thousands of AI agents to build, manage, and evolve. The third point on the market is also very important. This market is very dynamic. It's not a static market. These Agentic Transformation are not one-time projects, just the opposite. AI processes, business priorities are continuously evolving, which means, for us, a continuous demand for continuous transformation to get even more outcomes on optimized cost. All that looks great. Now, how do we Make it Real? We have built a methodology, a five-step approach to deploy and operate agentic system at scale. Let me focus on the two first steps, and especially on the most important points. Step one, and it has been said several times today, we reinvent the process. The idea is simple. Business process were built for humans, sequential, step- by- step. We reinvent them for AI. It means that we eliminate and compress task. We run them in parallel, and we add feedback loops for continuous improvement. Step two, we design the Human-AI operating model. Two things here. First, for every single task, we decide who does it. Is it fully autonomous AI with human oversight, humans assisted by generative AI, or humans alone? Second, obviously, we redefine the roles of the people. This step, it has been mentioned by Aiman earlier, is only possible because of one thing, our deep operations expertise. We are running the process, and that expertise has been significantly amplified by the WNS acquisition. We have now a team of 100,000 people specialized in operations. That's the foundation of Intelligent Operations. Now, the last three steps. They are more technical. This is where the magic happens. Let me start with the challenge of using an agentic system to operate a process. An agentic system is stochastic by design. It's probabilistic, which means out of the box, you can't use it to run business processes that need predictable controlled behavior. Intelligent Operations requires exactly that, predictability, control, governance, and audit. How do we solve it? We inject three unique ingredients across the three last step of our approach. In the step three that you can see on the slide, we build a solid data foundation. On top of it, a semantic layer where we encode with our deep knowledge the intelligence of the business, and it correspond to the two bottom layers on the screen. In step four, we design enterprise-grade agentic system. Not toy AI agents that you can build in a few minutes, but reliable orchestrated AI agents that can take actions autonomously. In the step five, the control plane on the experience layer, we add values in two ways. One, in the control plane, we are implementing the AI agents' control mechanism to manage AI agents like we manage humans. We assign missions, we set guardrails, we monitor performance, we enforce compliance, and we optimize the token consumption. Two, on the experience layer, we are implementing our human-in-the-loop approach, so user can oversee the AI agents when it matters. The result is very important. It's a controllable, predictable agentic system with governance and audit by design, one that enterprise can truly rely on. As you leave today, I think two things to take away. Transforming enterprise operations with agentic AI is the biggest AI shift for the enterprise, and the market is already strong today and continuously accelerating. It's a major opportunity for Capgemini. We do think that we have the business process knowledge and the breadth of capabilities at scale to Make it Real. Thank you. Thank you very much, Franck. To discuss further how Capgemini transforms enterprise operation, let's move on to our next fireside chat to hear another client perspective. For this, I'll invite Aiman Ezzat to introduce our next guest. Thank you. As you just said, agentic AI is reshaping enterprise operations. Let's bring this to life with one of our more strategic client partnerships, Unilever. Unilever operates at massive scale, reaching 3.7 billion consumers in over 190 countries with a clear ambition to scale AI across the enterprise. For that, Capgemini is a key partner, helping turn that ambition into industrialized Intelligent Operations. I'm very pleased to welcome J.C. Parada, Chief Global Business Services Officer at Unilever. J.C. leads a critical engine-driving transformation across Unilever's operation, and he's going to be joined by Rob Walker, Managing Director of Capgemini U.K., and together they will share how we are moving from AI ambition to real impact and scale. Hello. Hi, J.C. Thanks very much. Thank you for joining us. As we've heard today, it's all about agentic and AI and Intelligent Operations, but provide us a little bit of context around global business services at Unilever. What does that mean to you? Well, hello, everyone. Thank you, Rob, for the invitation. A privilege to be here. If you think about the context, let me say quickly that Unilever has decided to focus on three main priorities. When it comes to brands, it's about driving Desire at S cale. For organization, it's about playing to win. When it comes to optimizing the enterprise, it's about making the company fit for the age of AI. Intelligent GBS is, in this context, of course, we touch everything because of the size and the breadth of what we touch, but we have a protagonic role in making the company fit for the age of AI. Our objective is to focus on creating end-to-end intelligent workflows that can make the connections that create business impact and creates the conditions. Scale up agentic to maximize the benefit. We have the unique ability to look across processes and across functions to find these meaningful connections that will deliver impact. In the end, our aim is to liberate our frontline colleagues from all this administrative burden, provide them more intelligent data and information, and in the end, enable them to focus on winning in the marketplace. Very good. Investors hear a lot about AI nowadays. You can't do anything without hearing about AI. Less so about how it's being industrialized inside organizations, and particularly industrialized at scale. Maybe to start with, how are you thinking about that shift towards agentic AI at Unilever? Well, I think it comes back to, first and foremost, to create these intelligent business processes, focused on business impact. Drive significant levels of standardization and harmonization, which we do a lot through centralization and hubbing in partnership with Capgemini, as you actually run a massive amount of our operations. That creates the base to infuse technology in a way that is scalable, not fragmented. That's the key point. Of course, as you can imagine, this brings a significant change in the way people would work. I think the way you can think or we think about that is progressively, people used to do the work assisted by machines. In this context, what we are doing is really having the machines doing the work, assisted, guided by people. That creates a massive shift in the skills and the role of our people. Bottom line, what I'm saying is intelligent processes at scale and bring people along to be able to lead in the new world. Very good. In our conversations and in here, we've heard a lot about domain understanding. Why is that so critical as we move towards agentic models? Well, I think, I believe the solutions are as good as the quality of the definition of the problem. The key issue that typically we have in the industry is that the problems are not properly defined and not properly articulated. I believe this is the central point. The AI solutions will be as good as that scope, as that problem definition, and as that articulation. That's really key. This is when it comes one of the many elements of the relationship with Capgemini. We can have those discussions based on deep domain expertise and understanding of our operations among the two companies and put our intellects together to find this solution. That's why domain expertise is so critical to have the proper, meaningful conversation at the right level. Very good. Thank you. Unilever is a huge global organization, as Aiman Ezzat mentioned, and you work with many partners over the years. How would you characterize Capgemini's role in the transformation that you've driven, and how is Capgemini helping Unilever navigate AI transformation across its global operations? First, I think we define Capgemini in the reinvention of our process operations as a really critical strategic partner. Specifically, we see Unilever as a strategic advisor and helping us to execute operations at scale. With this in mind, with my own words, the way I see it is you have three, among many things, but you have three strategic and powerful arms. The consulting and expertise arm that basically provides all this domain expertise that you were exploring. The operational element that basically you can manage operations globally and with depth, and you know my operations by heart. Then your ability to develop and infuse technology. If you keep that in mind also, Capgemini bring us perspective and help us to sharpen also our strategic thinking. What I mean is Capgemini with these two elements, the expertise and the operations, you can go really deep within our operations and help us to optimize operations in detail internally, but also can help us to orchestrate, to find these meaningful connections, particularly leveraging the hubs. Even beyond that, you can give us CPG industry perspective, which I use frequently to make sure that we are staying awake and up to speed with everything that is happening in the market and having a critical view in terms of creating competitive advantage. That's the way to see it. The way this happens in the day-to-day I think is very interesting. We don't focus quickly in our day-to-day discussions in going directly to the answers. We basically say, "Well, what is the problem?" This notion of defining precisely the problem and articulating it properly. We really engage in discussions like, "What is the real problem we're trying to solve? What is what the customers or the stakeholders are not happy about? What is holding us back to scale more or scale faster?" I believe that's a very rich discussion. That can only be done because of the nature of the relationship and the depth of your expertise and operations. Basically, in short, when you put these three elements together, what we are seeing in Unilever is the ability that we have created as an industry model together that is capable to drive scale and agility in a way that we hadn't seen before in Unilever. Yeah, thanks. It's fascinating because there's a combination of that domain content with operational scale embedded by technology and innovation. I think the other aspect that we've spoken about, it's not just about back office or mid-office, actually, it's really moving into the front office and operations and really customer-facing activity. Maybe help us bring this to life with an example of where you've seen that working at scale. Sure. Well, we have been working now for a few years together, and there are different examples, but let me refer to one today, which is we started with a project called iOps or Integrated Operations. That has been the precursor of a new function that we created 3+ years ago, which is Customer Operations. Basically, with the Customer Operation, we have created an end-to-end flow that goes across functions, focused on business impact, and has delivered disproportionate impact for Unilever in cost, cash, productivity, service, and all the areas in scope, and has allowed us to infuse technology and scale, becoming today our most advanced model. The others that you are also helping us in marketing and in other areas are also scaling up, and in finance, are scaling up quickly. Basically, this model is an extended version of the traditional go-to-market supply chain. We combined elements of sales, operation, promotional planning, and execution. We connected these with the whole flow of supply chain from demand planning, supply planning, et cetera. We went to order management, delivery planning, delivery execution, customer delivery, and basically collection. We put together pieces of supply chain finance and sales in a single workflow. We put this in hubs, which you run for us, and then basically you help us to consolidate that operation to bring rigor in much faster than what we could have done in decades. You help us to drive standardization and harmonization and creating these flows that pave the road to then, with a very solid tech stack, to be able to make the data flow in a consistent way with more consistent processes. We have levels of standardization now reaching 90%, which we only in our dreams in the past, but basically today they are real. Then with that, we have been able to scale solutions, both in the vertical processes like supply planning, materials planning, or delivery planning, but also in orchestration and end-to-end flows. I think one of the key points is, of course, we have delivered significant value both in what we expected, productivity, service, optimization, automation, all the things that we know. To me, the most important thing or the most interesting thing, what I learned is we also found a way that a lot of the value comes not from what we were optimizing, but from the questions that we didn't know that we could ask, what we found that we were not expecting. Number one, we found questions that we didn't know before. Number two, we have not been able to answer these questions before. Today, even if we know the questions, our human capabilities fall short significantly of what it takes in terms of iterations and details to be able to answer those questions. I'm talking about millions of decisions per day that we didn't know we could make. Now we make it. We're covering 75%-80% of those millions of decisions, which are simply impossible. We couldn't go beyond 3% of those questions now that from a human standpoint, now that we know this question. Before we didn't even know about those questions. That's what we have brought to reality. I like that. It's a nice example of unknown unknowns, so to speak, and what it allows you to unlock now. When you look back, what do you see as some of the critical success factors around enabling this? I think the success factor, I think it goes around three elements, I would say. Mainly, it's about creating and putting in front of the organization a clear business benefit. I think the second is to create a model that in an organization like Unilever, make sure that you can scale quickly at global levels because that drives credibility and consistency. The third point is to be able to manage properly change management and user adoption, because that in the end is the make it or break it, because technology and solutions are as good as what the users are willing to take. Honestly, those are the three elements that would ensure success and which we have learned along the way to do better and better. It's very simple then, I mean, very straightforward. I wish it was. The perspective of start with the business outcome. That was a critical component. Look, we often hold up Unilever as being one of the most innovative companies, particularly in this Intelligent Operations. Even where you are, what is ahead for you for the next 12 months as you're looking for even further innovation? What excites you most about the next phase? Well, I think maybe if you refer to where we started, I think the most exciting thing is that we are really transforming Unilever operations at scale. Basically, my key point is we are in the middle of a pivot. What excites me the most is the ability to move GBS from the traditional being a cost center, or a purely operational engine, to be a strategic partner of the business to create value. That is what we are finding more and more, value streams. We are standardizing more, we are giving more meaning to technology to scale it up, and we are guiding even our HR function in partnership with them to really transform the way we upskill people and we manage the operation successfully. We're having some successes there, a lot of them, but we have a long way to go. In short, doing this transformation, the key legacy that I would like to create is really landing this will make Unilever fully fit for the age of AI, which is what we're looking for. I really appreciate that the partnership that we have with Capgemini, with everything that I said, is really enabling us to move there at pace. Well, J.C., I couldn't think of a better set of examples where you're talking about business outcomes, value streams, and accelerating the opportunities with what you call iOps, Integrated Operations, but we're now developing into Intelligent Operations. Thank you very much for joining us. Thank you, Rob. Indeed. Thank you, gentlemen. Thank you very much. I'm now going to invite Karine Brunet, she's the Chief Operations and Delivery Officer, to tell us about the adoption of agentic AI in Capgemini. Good afternoon. You've understood from the previous speakers the value pools that agentic AI is creating for us. This section is going to be focused on how agentic AI is transforming delivery. As you understood, throughout the past 25 years, we went through many evolutions. We transformed our infrastructure business to make it fit for the cloud. We adopted agile and DevSecOps at scales for software engineering. Obviously, where today, cloud and digital represents about 80% of our business versus close to 0% 10 years ago. This means that over the years, we have developed muscles on how to manage those technology evolutions, and practices on how to embed them in our delivery team. AI is definitely an inflection point in the industry, and it is impacting our delivery in two main dimensions. First of all, how we deliver services to our clients, but also what we deliver for those clients, since you understood that we are building, deploying, operating, and maintaining the new agentic AI enterprise stack. Now let me focus on the how, and I'm going to focus on the technical delivery side of the house. First of all, we can already see today very concrete, tangible benefits of agentic AI in delivery. The first one is that it enables us to reduce the cycle time. This means that today, thanks to agentic AI, we can develop applications 30%-70% faster, depending on the type of application. Application migration and application refactoring can be executed today in half of the time. Obviously, it also enables us to increase the availability and the reliability of the Information System. By being able to predict incidents before they occur, we can reduce about 20% of those incidents and the associated downtime. Today, thanks to agentic AI, we are able to solve about 30% of incidents without any human intervention. Overall, the mean time to repair an incident is decreasing by more than 10%. Obviously, with agentic AI, the associated human effort is decreasing for a given task. Agentic AI has a cost, a token cost, and it can be significant. Therefore, effectiveness has to compound the decreasing cost of labor and the increasing cost of the tokens. Transformation is happening and is occurring, but as you heard before, it's going to happen over many years. CXOs are actively pushing for the adoption of agentic AI within the enterprise. The enterprise adoption of agentic AI is still slow. Companies are still selecting the technology they want to go with, defining their governance model. Are they going to make sure that those agents that we are building are ensuring compliance with regulatory frameworks? As you also understood, large enterprise have got very complex Information System that are also slowing down the adoption at scale. Those Information System s are made of hundreds of applications that have been developed over the years that are tightly or loosely integrated. Which means that to be able to get the power of agentic AI, we need to make sure that those agents can interact with the existing Information System, and that complexity is slowing down the adoption at scale. Now, this slide is a simplification of our delivery capabilities. As you understood today, we have much more breadth. For the purpose of this, I just try to regroup in four main categories. The purpose of these slides is to explain to you that to adopt agentic AI in delivery, it is much more than just providing technology and training to a workforce. First of all, as you heard from Fernando, over the past six months, we have signed very significant agreement with all the key AI technology provider, making sure that our workforce has got access to the broad set of the capabilities and technology that agentic AI is providing. We are increasing our already significant effort in developing Capgemini delivery platform. Obviously, we are focusing on the upskilling and the reskilling of our workforce. We are delivering more than 25 million of hours of training on data, AI, and agentic AI. We have redefined individual KPIs for our employees, making sure that we can measure how well they are using agentic AI in the delivery of their daily task. To give you an example, for a software engineer, we are monitoring the quality of the code that they're producing thanks to agentic AI. Also the type of model they are using. More importantly, the number of tokens they are consuming on a daily basis. As you may know, a given developer can spend in one day, in terms of token, what another one will spend in a month. Therefore, managing that delivery cost is becoming very important. Obviously, we have also revisited the way we recruit and our selection criteria for candidates. For young professionals, we are focusing on problem-solving, curiosity, and communication. Three dimensions that are key in the adoption of agentic AI. More importantly, we have redefined all our operating processes, all our practice and methodologies. We have redefined the way we develop software using agentic AI, the way we do product engineering, the way we run infrastructure on behalf of customers, and as you've heard as well before, the way we combine human and digital to deliver enterprise business processes. What does it mean for our workforce? On the left-hand side, you have the current split of our workforce, again, in a very simplified manner. On the right-hand side, it's an attempt to project what that workforce actually will evolve over the years. In the middle, what is changing. Listening to our previous speakers, you understood that there is a growing need for business process consultants to redefine and reinvent the enterprise business processes. The role of the software engineer is changing and is expanding throughout the software development life cycle. It now encompasses business analysis, development, and testing. Software developers are working in a much smaller team that is usually now pluridisciplinary. To take a music analogy, the software engineer is becoming a conductor. Now, to be able to orchestrate human and digital workforce, that software engineer has to be an expert in music score. Agentic AI is also creating new roles within Capgemini. Agent Enterprise Architect that can define how those new developed agents will interact with the existing Information System. A growing demand for data engineers, but also people who can manage the Agent Identity and their Access Management. Engineers that can test the quality of those agents, engineers that can orchestrate the agents within the delivery system, and obviously, in the control plane, we will find new engineers that will operate the control plane, as well as experts in the token economy. People who will monitor on a daily basis the consumption and the quality of the token. In our application maintenance and in our infrastructure management capabilities, we are obviously automating all the repetitive task. The need for junior is decreasing in those categories. However, the demand for expert is increasing because we need an increasing number of level 3 engineers who can tune and moderate the agents in production. In the business process area, we obviously see a compression when we orchestrate both agent and digital workforce. However, the growing demand for Intelligent Operations make us think that the workforce evolution will be very moderate. I would like to take some examples on how we see the concrete benefits of agentic in delivery. The first one is a car manufacturer. They have a legacy application running on mainframe that was managing all their car sales and car orders worldwide. The objective for them was to decrease the cost of technology and to move that application from mainframe to a new infrastructure. Thanks to agentic AI, we did not only migrate that application, but we completely refactored it in half of the time that was initially planned. We went into production with a new code with zero defect. We improved the performance of the application by 90%, and our efforts has been reduced by 2/3. The second example is in life science. It's about an employee IT service desk, where we are developing agentic systems to replace some of our service desk agents, but also to assist them when they fix issues from the employees. 85% of the chats today are solved without any human intervention. Overall, the ticket resolution time has been decreased by 20%. The customer satisfaction has been increased to 4.9 out of 5, and the end-to-end cost of delivery has been decreased by 20%. The last example is not a customer example. It's about a platform that has been developed by Capgemini called Malcolm. That platform is an asset that is today used by seven customers. It is for the shipping and the trucking industry, and it is agentifying the process from ordering through invoicing up to receivable. Today, 27 million transactions managed. 50% of the bookings are made without any human intervention. The invoicing accuracy has been increased from 90%- 98%. More importantly, this means that our customers have experienced a decreasing DSO of eight days. Quite a significant impact for them. Obviously, we consider Capgemini as the customer zero for agentification. This means that we are looking at all our key internal processes, eliminating the task, simplifying the process, and reinventing new process that can be agentified. With our technology partner, we have worked on a very innovative IT architecture for ourselves, where all our enterprise data are consolidated into a unique data hub that is collecting the data, transforming them and cleaning them, and making them into data products that can be consumed by agents. Today, 40% of our customer proposals are deployed thanks to agentic, and we believe that by the end of the year, we will reach 80%. One of our core processes is our supply chain. It's our ability to match the profile of our workforce with customer ask. In that area, we are also heavily using agents to identify the profiles, to prepare the CVs, to make sure that the match is done in a very rapid time. On average, we used to staff individuals within 20 days. Thanks to agentic, we have shortened that cycle into five days. Going back to the value pool, you understood that the agentic technology stack and the control plane are two very significant value pool going forward. Where I explained in the first part how agentic is changing our delivery, I'm not going to focus on the what. We discuss about those value pool, and as you can imagine, this means that we need to build new capabilities to be able to build and deploy those value pools for our customers. What are those new capabilities? First of all, we need to develop those agents. As mentioned before, we are creating ODE, Outcome Deployed Engineers. Those people are usually on-site, working closely with the business side of the customer. They usually work in pairs with two type of profile. One ODE that will have a very deep functional business expertise, and one ODE that is much more hands- on -the- keyboard developing the agents. To make an analogy again, if you are in an insurance company and we're trying to agentify their process, one of those ODE will have to understand very well what an insurance claim is and how it is managed, so that we can take the functional understanding and embark it in the agents we are developing. Once those agents are developed, they need to be enabled in the existing Information System. Here we have an enablement team, which is very classic to what we're doing today with the growing needs for data engineers, but also infrastructure experts, network experts, SaaS experts, software experts, people that understand the Information System and can make those agents interact with the rest of the Information System. Once those agents are built and deployed, then they need to be operated. Going back to what was said earlier, we expect that going forward, any enterprise employee will be assisted on average by 10 agents. If you think of the size of the company, this means that every company will have hundred of thousand of agents to manage, you understood that the control plane is that capability and that capacity. In the control plane, we manage the user access of the agents. We give them an identity like humans. We are also defining what type of actions they are able to do and defining the guardrails that are so important for an enterprise to build the trust. We are orchestrating those agents among themselves. Obviously, we are monitoring them very closely, making sure that they act on purpose and that the output they are delivering and the action they are taking is appropriate. We also make sure that they are protected from any cyberattack. Finally, in that control plane, we have also experts of the token economies, managing on a daily basis the number of tokens consumed by those agents and tuning and adjusting so that the cost of the agentic can be optimized. This concludes how delivery is transforming within Capgemini in order to make sure that we harness the power of agentic for us, but also for our clients. Thank you. Thank you very much, Karine. I suggest we take a break. We'll take a quick five-minute break and reconvene here for the last stretch of our Capital Markets Day. Welcome back for this last and third part of our Capital Markets Day. To begin this last part, we're going to hear another client perspective, and for this, I'm going to ask Aiman Ezzat, our CEO, to come on stage with our guest, Lukas Paravicini, the Chief Executive Officer of Imperial Brands PLC. Lukas Paravicini has a proven track record in multinational consumer goods companies around the world. He joined Imperial Brands in 2021 as CFO. Previously, he was the CFO of ED&F Man Holdings, an agricultural commodities and brokerage group. Earlier in his career, he spent 22 years with Nestlé in various senior finance and general management roles, allowing him to acquire deep knowledge of technology and its opportunities to enable change. Gentlemen. Lukas, thank you. Thank you very much. Thank you for being with us. Thanks for having me. Maybe a simple one to start with for the audience in the room, maybe you can give us a brief introduction on Imperial Brands and the industry. Sure. Yeah, thank you very much. Thanks very much for having me. Good afternoon to everyone. Imperial Brands is the fourth largest tobacco company in the world. We operate in over 100 markets. This is a very dynamic business. We have a product portfolio that spans from tobacco all the way to what we call next-generation products. Anything from vapes, oral nicotine pouches, heated tobacco. As I said, a very dynamic industry. The transition is really driven by consumer insights, consumer demands, lots of innovation, and as you could expect, lots of regulation. We are known as the fourth largest, as I said, and we are a true challenger business. We focus on the consumer. We are choiceful, we make choices, and we want to remain agile. Our business is anchored around three things. One is drive sustainable value through our combustible business. It is about building a meaningful NGP business, a next- generation business. It is to transform ourselves for tomorrow, to be ready for tomorrow, which is really twofold. Gaining efficiency, being more agile, but also more data-led and being more consumer-centric. We hear a lot about AI, much less about why it's actually being industrialized inside organizations like Imperial Brands. To start, could you share how AI fits into your broader strategy and transformation journey towards 2030 goals? Ultimately, what are you trying to achieve and why this is critical for your business? Yeah. I think, as a CEO, everybody talks about AI, and everything is AI nowadays. For us, it is really important to be a true challenger, being choiceful and making sure that we focus on that element of AI and agentic AI that really helps us deploy our strategy. In broad terms, there are probably two avenues we're looking at. One is purely the efficiency. How do you become more efficient, more agile? Which is, for me, the ticket to the game. It's the entry. The real uptake for us is how do we make sure that we actually deliver our strategy better in terms of how do we capture more opportunities for revenue growth by becoming more consumer-centric, having better insights, leveraging AI for that purpose. That's really where we see the bigger opportunity in a very focused way, and where we see our partnership with Capgemini bring a lot of opportunities, both in the efficiency, but mainly also in the revenue growth opportunity. Based on the strategic importance of this transformation, so as you say, partnership can become critical. When you think about the journey you embark on, what do you look for in a partner, and what kind of partner can truly operate at your pace and deliver impact at scale? Yeah. I think for us, and again, we do that in other areas as well, like innovation, et cetera. We are a big company, but we are the fourth largest. We are a true challenger, and so we look at our partners as true strategic partners. We don't look at partners as service providers. That's not where we believe the value is. It's really in the strategic partnership. I think Capgemini has really shown us in the short time that we've been together on this journey, is that you really understand and care for the strategy, and excuse me, and importantly, know how to translate that into a true business outcome that leverages your industry knowledge and also your industry expertise. Sorry, let me just continue. That's the first piece. When we looked at Capgemini as well, what we could see is we are not trying just to move a certain element of our processes to a service provider. We're looking for a strategic partnership that, in your case, showed very good knowledge about the end-to-end capability and how you can operate that across a full- value chain and show us how you integrate technology data into our operations. That capability to bring that into our operations was very important to us. Finally, in the few months that we've been working together, there has been a solid build of trust, which is, for me, still a very important ingredient of any relationship. That trust is on an individual level, but also it is on an organizational level that you have created that with us. If you summarize in two or three words, what really is distinctive about the partnership with Capgemini in the transformation journey? I think, as I said before. Yeah for us, the importance is the capability to look at you as a strategic partner, is making sure that through the understanding that you have shown us and the interest in our strategic intent, you have really been able to translate that into a business outcome. You do that through two things. Yeah. Which is basically through leveraging your expertise in the industry and technology, but also by leveraging your ability. Sure. Operate across the full- value chain. Those two things allow us to accelerate the transformation, accelerate the adoption of technology. Accelerate the opportunity to simplify our organization, but also accelerate the opportunity for us to double down on our consumer capability, thanks to your input. With that, double down on our commercial opportunity. Okay. Just go to the next, please, questions. Looking ahead next 12 months, what is really exciting you the most about the next phase? What are your expectations in terms of what needs to be achieved? Listen, I think firstly, I want to acknowledge that even though we only started and have signed a contract, I believe it was last February. Yeah We had a very good start. We were very quick out of the blocks. Signed the contract in February, and by the 1st of April, we had 400 roles transferred to Capgemini, and actually, very pleasingly, with a 99% retention rate. Which is very important for us, is to show a good start. What I'm really excited about is the opportunity to accelerate that process now. It's going from building the foundation to really scaling up the opportunity to simplify our business, to embed AI. Again, for us, it is not just about using the technology. The technology itself is one thing. It's how to use the technology. How do you find that technology that helps us develop our strategy and bring value to our consumers and obviously also to our shareholders. I think I'm excited that we could accelerate that, we could focus on the execution, and we could focus also on deepening the relationship, which is at the very beginning and will go for many years. Of course. When you look at a partnership like that, how important is cultural fit, collaboration? How would you characterize that and how important it is for you for this to be successful? It is very important to us, in the sense that for us, I think the what is one element of any partnership. We've worked hard to define shared objectives, how do we make sure that our strategic intent also is linked to the incentive in the partnership. All these things work really well. We have mutual benefits, all that. I think where it really matters is that you do understand our culture, who we are, what are we trying to do. Again, it is about how do we win with the consumer? How do you understand our culture, how we do things? What is our challenger mentality? That is very important. You have shown a lot of interest and care throughout the process to not just deliver us an outcome, this is a solution, this is the way we do things, really understanding what we are looking for and adapting your solutions within obviously the benefit of the scale to our needs. That has been very helpful for us. I think that is key for us and for the success going forward. Great. Any final thoughts around basically level of expectation? What is the level of change you'd expect at the end of the day in your organization? You said you had to challenge your position. Is your ambition through that to be able to leapfrog in some areas some of your competitors, and you think that this kind of really large-scale transformation, with its level of complexity, of course, can really help you achieve superior result and really move forward the needle significantly? A couple of things. For us, the transformation is quite significant. As a lead team, we don't look just at the efficiency. Okay? I think the efficiency is hugely important. I've never met any consumer who's willing to pay for inefficiency. I think it is only half of the story. What we are looking for is the added value of getting more revenue growth at scale and at pace. It is a significant impact, and it is important that our employees understand that this is not a pure shared service and efficiency. This is not what we are trying to do. I think Capgemini has understood this very well, and it's helping us really to translate this opportunity in how do we unleash the potential of our great talent we have. Right now, our people are hugely motivated. They are very talented. We've brought hundreds of people from outside in, combining them with our knowledge in tobacco and NGP. They come in a company where the processes are broken. We can't really unleash our potential. If we can show our people that what we're really trying to do is helping them to deliver their best every day and win with our consumer, that's really what we want to do. Everything else is a consequence of our people, with your help, with your knowledge, and the capability of accelerating that to do that quickly. Just to finalize, we've discussed this in previous discussion. If you think of what we're trying to do, many other companies have gone this journey. They've probably taken 10, 15, 20 years. We're trying to get to that point in two years by doing an end-to-end design. We could not do that without your help. Lukas, thank you very much. Thank you for taking the time. Thank you for the partnership, and of course, looking forward to great achievements. Thank you very much for having me. Thank you very much. Thank you. Thank you very much, gentlemen. I'll ask Fernando Alvarez to come back on stage in order to share with us in more detail the evolution and expansion of total addressable markets. Good. As you do, we do the same. We work on models to justify the decisions that we made, and we put a lot of the energy in defining these five value pools that we share with you today. What I will share with you is an effort that we have been working on for quite a while to get to some of the conclusions that we have shared with you. Rather than depending upon ourselves and our sources, which we do, we have a team to do that for us, we decided to partner up with McKinsey. When we work and approach McKinsey, we basically ask them what source of data and information could we rely on and share together to define and accelerate the definition of these five value pools. We decided to leverage the work they recently did for NASSCOM. When we realized and we studied that analysis they did, we realized two things. That we're missing information on advisory capabilities, which is everything that has to do with the consulting part that we have been drumming on today. It's very important. We also realized, as we have been realizing in dealing with other market analysts, that there was not a lot of coverage on what we call ER&D, engineering and research and development, because our objective was not only defining the five value pools that we share with you today, but also try to quantify the wallet spend pool that we were going after. We are trying, as what we have done today, is to create a very clear laser focus framework that defines where our upside is in terms of we move forward as we embrace an agentic value architecture. As a consequence of that, this is the result of that work. First of all, it's a global business and technology service market. That means we win, and we set the stage to what we do. We're focusing on the service market. We left a lot of content, including infrastructure as a service. It's not included as part of this TAM. That's the first thing. Then we did the analysis for the period of 2025 to 2030, and we came to the conclusion in the work going back and forward and challenging each other that we expect this particular TAM to grow at a 4%-5% CAGR during this particular period. Under the same analysis, we were very laser- focused in understanding the same way that there is an AI expansion, there's also AI compression. We were trying to make an assessment, what does that mean? We were in a very detailed modeling, what's behind each one of these value pools in order to quantify versus the raw data that we were getting in terms of taxonomy. We came to the conclusion that the AI expansion and the AI compression at a midpoint level is a net neutral during this particular period. Give and take, as you can see there, depending on the fluctuation. This validated for us the following, this is what I was trying to articulate and my colleagues tried to articulate throughout the presentations today, that there's a structural growth driver rotation in terms of the rotation of what secular growth is versus the AI compression. We see that now it's what we are trying to articulate with the five value pools of how this is evolving. That led us to understand also as well, that the agentic value pools that we are describing here in each one of the five are growing at an accelerated pace, ultimately surpassing the secular growth. The assumptions start validating that the secular growth will continue. Yes, there is some AI compression, but the speed of the five value pools around agentic AI architectures that we have defined is surpassing. The AI expansion during this period of 2025 to 2030, it's give and take at a midpoint range, EUR 420 billion versus EUR 400 billion. It's a pretty conservative approach that we took rather than the aggressive approach in order to justify the decisions of where do we go in each one of these value pools. We decided we were challenged. What about after 2030? One of the arguments that we constantly, in this particular rapid pace market that I was describing to you, it's rather difficult to get our arms around three years. We went five, and that's what happened beyond. We took the models, we took the standard deviations that we have agreed upon ourselves to give us the validation, and we decided what that looks like beyond 2030, just for the sake of the argument of the analysis. It came to the conclusion that the agentic value pools drive long-term market growth beyond 2030, according to the information that we were calculating on AI. Compression will not be as aggressive during the period 2030 to 2035 as it was in 2025 to 2030. If you give and take and you take the mid-range point of this, the AI compression between 2025 to 2030, more or less is 20%-25%. If we do the same analysis beyond that point, 2030 to 2035, it takes us to between 10%-15%. The point at hand is, it's not an electrical switch, boom, off, we move on. It's nowhere to be seen. It's an evolution, but we see that the compression start diminishing as AI starts taking over the different elements of how do we address the business. Concluding in the analysis that the business and technology service market, again, with the conditions and the if, buts, and maybes that we define, especially we took infrastructure as a service out, should be growing between a 6%-10% CAGR. You take the mid-range point is around 8%, moving forward. That needs continued more analysis, continued more modeling, learning from it, validating the sources, but that is more or less the analysis that we have done, and that help us guide ourself the definition of the five value pools and the intensity of what we see in the short term, midterm, and long term in the enablement and the growth of these five value pools. Based upon that assessment, then I will leave Nive to come on stage and share with you her visions or our vision of the ambitions moving forward. Thank you very much. Thank you, Fernando. As you said, it's now time to welcome Nive Bhagat, the Group Chief Financial Officer, to share the group's 2028 financial ambition. Hello, everyone, thank you, Adrian. Thank you so much. Thank you all for having come today and joining us. Let me start off with our performance over the last few years. As Aiman has said before, we have a track record of successfully navigating various technology waves. Over the past few years, we have demonstrated just that. Since 2020, we have built a high-quality portfolio that has positioned the group at the center of enterprise digital transformation across enterprise management, customer- first, and intelligent industries at the back of our leadership in digital and cloud. We have built real strength around cloud, supported by deep strategic partnerships, and at the same time, we invested early in Data and AI, which was one of the pillars of our 2021 strategic framework, positioning the group well ahead of the next wave of transformation. The results from this are clear. Between 2020 and 2025, we delivered a 7.3% revenue CAGR at constant currency. Excluding acquisitions, our organic growth was slightly above 5% CAGR, reflecting our ability to capture demand across the cycle. Over the same period, we managed to expand our operating margin by 140 basis points, and we are now one of the few global players to operate at a level visibly higher than the pre-COVID profitability. This reflects a combination of factors. We have increased our exposure to higher value services with our clients, successfully addressing new playing fields within our clients' operations. At the same time, we have maintained strong operational discipline, ensuring that our model remains resilient in what has been a fairly volatile environment. This has translated into solid cash generation. Over the past five years, we delivered EUR 9.6 billion of cumulative organic free cash flow. This has allowed us to continue investing in the business, reinforce our capabilities through targeted acquisitions, and return capital consistently to shareholders. Now, as we enter the next phase, we do so from a position of strength with a balanced and disciplined financial model that continues to support both performance today and investment for the future. Now let me turn to the evolution of our profitability framework, which is an important element of how we communicate our performance. We have listened, and this evolution reflects the dialogue we've had with you, our investors. As our business continues to evolve, it is important that the way we measure performance evolves with it, and our headline profitability metric needs to provide a clear view of the group's all-in operating performance. Currently, we present a number of our other operating income and expenses below operating margin. We're now simplifying this by moving most of these lines above the line. These items go above the line. They will be included directly within our operating cost and sit within the operating performance we manage on a day-to-day basis. This includes items such as share-based compensation and restructuring costs, which are a part of running and transforming the business over time. These, together with the other items that were previously reported under other operating income and expenses, will now be included directly within the relevant operating cost lines. We will therefore introduce this new refined view through a new metric, adjusted operating profit. This metric is simply the IFRS operating profit before acquisition- related expenses. As such, it only excludes integration and acquisition costs and the amortization of intangibles linked to business combinations. By isolating these items that are directly linked to past acquisitions and that are mostly non-cash, it provides a clearer and more consistent view of our all-in operating performance. I now come to how our 2025 results would look in the context of this new profitability metric. As I outlined earlier, share-based compensation, restructuring, and other items are reclassified into the relevant operating cost lines. That's cost of services rendered, G&A, and selling expenses, resulting in the new metric of adjusted operating profit. Below this are the acquisition-related expenses to then arrive at the IFRS operating profit. You can see that from this in 2025, our 13.3% operating margin translates to a 10.8% in the adjusted operating profit metric. Moving forward, and in terms of reporting for 2026 in particular, nothing changes. Our outlook and the way we report our financials for 2026 remains unchanged. We will apply this framework to the 2027 outlook that we will present next February, and it will be reflected in our financial statements from the first half of 2027. To summarize, this change strengthens the relevance of our key profitability metric and provides a more comprehensive view of the group's operating performance. Let me turn to our 2028 financial ambition. Our ambition is clear: to combine stronger growth with higher profitability while maintaining disciplined cash generation with the same execution rigor that has underpinned our performance in recent years. Starting with growth, we are targeting a CAGR of 5.5%-7.5% at constant currency through to 2028. This includes around 2 points from M&A, primarily from WNS, which is already embedded. This ambition is grounded in a reasonable view of market dynamics and reflective of our positioning across the key transformation opportunities, as you've heard from all our other speakers today. On profitability, we see a clear opportunity to further improve our all-in operating performance, and note my point on all-in operating performance, over the next few years. We are targeting an expansion of adjusted operating profit margin of 130 basis points to 150 basis points versus 2025, reaching between 12.1% and 12.3% of the revenues by 2028. Just to reiterate, this is our new measure of profitability, which incorporates the majority of what was previously reported under other operating income and expenses, and therefore provides a more complete all-in view of our operating performance. We also expect to maintain a strong level of organic free cash flow generation over the period, consistent with the discipline that we have demonstrated historically. This translates to over EUR 6 billion of organic free cash flow over the next three years. These ambitions are supported by a combination of growth drivers and operational improvements, which I will now walk you through. Turning to our revenue growth ambition through to 2028. Our objective is to accelerate growth, and that rests on capturing the value created by AI. All our speakers have spoken about that. Agentic AI enterprise transformation will grow to be a meaningful chunk of our business. It's important to note that the non-agentic demand will not simply disappear. Even after taking into account the transitory impact from the ramp-up of AI-native delivery, it will still be an important part of what we will do in 2028 and beyond through 2030. AI is both expanding our existing opportunities to create value and opening up new ones. The key aspect to note here is that we see AI will be a net incremental tailwind to our growth, and this will be across a number of value pools. First, we will benefit from the acceleration of technology modernization. We're seeing this here and now as some of this is a clear prerequisite to building the agentic tech stack. In this agentic AI enterprise transformation landscape, Intelligent Operations stands out today, and we already see material traction on this value pool with double-digit growth across our digital business process services business and across both, of course, the Capgemini and WNS scope. We expect the other AI value pools that Fernando talked about to increasingly pick up through this period and continue to support our acceleration. This covers the agentic tech stack, the agentic control plane, agentic products and services, and they will increasingly be fundamental as organizations fully embrace the AI transformation. Taken together, these drivers support our ambition to deliver the 5.5%-7.5% constant currency revenue CAGR through to 2028. Let me turn to margin expansion and to what gives us confidence in our ability to deliver it. Historically, margin expansion has come from delivering higher value services that continue to be a key element. As AI embeds into enterprise transformation, our work becomes more strategic, more integrated, outcome-led, and supporting a richer mix of services. Our margin trajectory assumptions are therefore based on a number of factors, and a significant part of it is supported by levers that are already underway. Alongside developments within AI and the portfolio mix, as I just talked about, the Fit- for- Growth initiative that we announced earlier in the year will support the margin as we address the pockets of under-absorption in Continental Europe and to accelerate the transformation of our capabilities to support our growth agenda. WNS will contribute through its structurally margin-accretive profile and the synergies we expect over time. We will continue to maintain strict discipline on SG&A, which remains a clear priority for us. Now, investments are an important part of this equation. We are increasing our investments in offerings and selectively in the capabilities that strengthen our positioning and support future growth. All of this while retaining the flexibility to adjust as the market conditions evolve. Taken together, these drivers provide a clear and credible path to expand adjusted operating profit margin by 130 basis points to 150 basis points between 2025 and 2028, reaching 12.1%-12.3% of revenues. To summarize, and it's important here, our margin expansion is broader than AI alone. It combines increasing client value creation with visible execution levers, which are not just dependent on external factors alone. Let me bring this together and show you how this translates into a clear path for strong earnings growth through to 2028. At the core, earnings growth is driven by our operating performance. Our revenue trajectory, supported by a combination of structural demand, the expansion of AI-driven value pools, provides that foundation. This is complemented by the expansion of our adjusted operating profit margin, which incorporates several levers, which I just talked about. These levers are further complemented by the mechanical tailwind from the lower acquisition-related expenses as we exit the integration phase of WNS. At the same time, we factor in some headwinds with an increase in our effective tax rate and our financial expenses higher than in 2025, as we now have the full impact of the WNS acquisition debt. Taken together, these elements support a high single-digit earnings CAGR over the period between 2025 and 2028. I'm going to pause here for a minute and tell you what this implies. If you start from your consensus estimated earnings for 2026, I am talking about a solid double-digit CAGR to our 2028 earnings. What matters most here, of course, is the quality of this growth. It comes from a balanced mix of top-line growth, Capgemini specific execution levers, and very disciplined financial management. The message here is simple. We have a clear path to the high single-digit earnings growth, supported by our distinctive positioning in AI-led enterprise transformation, aided by the discipline of our operating model. Moving on, organic free cash flow generation is a core strength of our model and remains a central pillar of our financial framework. This is underpinned by our strong operating performance and is, of course, driven by the acceleration of revenue growth and the expansion of our adjusted operating profit margin and our continued discipline across non-operating items. Together, these elements translate into a strong cumulative organic free cash flow generation of above EUR 6 billion over the period, with free cash flow consistently above net income, a continued area of focus for us. This provides us flexibility to continue investing in the business while returning value to shareholders. Let's come to capital allocation and capital allocation through to 2028. We will maintain a balanced and disciplined approach to redeploying our free cash flow across four priorities: dividends, share buyback, deleveraging, and acquisitions. Starting with the direct returns to shareholders, our dividend policy builds on a clear historical payout ratio of around 35%, which today represents an attractive yield of over 3%. In terms of share buyback, we will continue to execute our EUR 2 billion program, which we launched last year, and this will clearly reduce the number of outstanding shares. Taken together, these two elements represent around 5% annual return to the shareholders based on our current share price. Deleveraging remains a priority over the next three years, with a lower level of acquisition activity providing us the flexibility to pursue deleveraging. At the same time, our approach to M&A remains disciplined, as implied by our growth CAGR ambition that embeds about 2 points from M&A, of which 1.7 points are already embarked. We expect to tone down M&A in the short term. We will continue to pursue selective bolt-on acquisitions where appropriate to strengthen our capabilities and enrich, of course, our offerings. Overall, this framework reflects a balanced and consistent approach to capital allocation, combining continued investment in the business with attractive and sustainable returns to shareholders. Let me now conclude by bringing all of this back to what matters most, which is shareholder value creation. There are two core components behind the value we intend to deliver. Number one, the first, of course, is the earnings growth with a high single-digit net income CAGR from 2025 to 2028 period. The second component is direct cash returns to shareholders. Through our dividend and share buyback program, we'll deliver around 5% per annum to shareholders. Taken together, earnings growth, dividends, and share buybacks result in a total shareholder return profile that is both attractive and well-balanced. This ambition we have built on reasonable and transparent assumptions. It does not rely on significant external tailwinds or aggressive scenarios. It is grounded on what we can focus on, our growth profile, our margin trajectory, our cash generation, and our discipline in capital allocation. As you've heard throughout the sessions this afternoon, we see agentic AI as a clear structural growth opportunity and a catalyst for our margin expansion. With this, we're confident in our ability to deliver superior shareholder value supported by a balanced and a disciplined financial model. Thank you very much. Thank you very much, Nive. Thank you very much. I'm now going to invite Aiman back on stage for his closing remarks before we move on to the Q&A session. Before we move to the Q&A, and I'm sure there will be many question, let me leave you with a few closing thoughts. It's been a long CMD, but I trust you agree agentic AI warrants the extra time. My message to you is extremely simple. First, agentic AI is not just the next wave. It is a step- change. It will unlock significant value for clients while introducing new levels of complexity, and navigating that complexity is exactly what we are built to do. Second, to those who see this as a threat to our industry, our growth, and our margin, we will prove them wrong quarter- after- quarter. The five value pools we shared are not theoretical. They are tangible, quantifiable, already underway, and they expand our total addressable market. Third, we're not entering this new era unprepared. We're entering from a position of strength, deep industry expertise, end-to-end capabilities, a powerful partner ecosystem, and years of sustained investment in cloud, data, and AI. Let me end on a personal note. I have been in this industry for a long time, a very long time, and witnessed many shifts and transformation. I have never been more energized, more excited, and more confident about Capgemini's future than I am today. Thank you. Thank you very much, Aiman. We're going to set up quickly the stage for a Q&A session and have you back here with Nive and Fernando. Allow me first to share with our audience the following indications for the Q&A session. In order for as many of you as possible to ask your questions, I would ask you to ask just one question, one question at a time, and to begin by stating your name and your company name. If you wish to ask additional questions or follow-up questions, please rejoin the queue. This will allow us to take questions from as many people as possible. Let me add that we'll take questions from the room, of course, but also from our online viewers, and there are over 350. Please do remember to precede your question with your name and the company name. The stage is nearly ready. We're going to begin in just a few seconds. I will invite now Aiman, Nive, and Fernando back on stage. Take questions. I had barely finished my sentences. Oh, sure. raised right there. You won't be surprised. Oh, go ahead. Can you please have a microphone here? Great. It's Mohammed Moawalla from Goldman Sachs. Thank you for the presentations today. Extremely helpful. I kind of had a two-part question. I was intrigued. It's starting. There's no three-part question, just to be clear. I was intrigued by the slide in Nive's presentation around how the non-AI business is expected to kind of still stay roughly flat in absolute terms, because I think the perception is obviously that there's a kind of compression from AI. Can you just help us understand maybe with some examples of the kind of portfolio, where that sort of compression may happen or how you offset that versus the incremental benefits from AI? To the extent you can contextualize also in terms of gross margin, because in the previous cycle, Capgemini has seen some pretty impressive gross margin expansion. How should we think of that gross margin going forward? Then one for Fernando. You've obviously been through multiple cycles within the industry. How does your relationships with perhaps the SaaS companies that you've had in the past change as you now talk about the domain expertise, industry expertise? Because we were at SAP Sapphire Orlando and many of the SaaS companies, including SAP, talk about that being a kind of a big moat. I'm curious to understand how those partnerships shift, particularly as you embrace OpenAI, Anthropic as new partners. Thank you. That's three questions. It's not two, huh? Let me start. First, I'll address the revenue one. Listen, there is still growth. This is what you don't realize. There is still growth in the secular stack. We say a lot of it is going to be absorbed by the AI compression. We say it's flattish projected a few years out, and here we're talking to 2028. 2026 is there, we're talking about the next two years. There's growth coming from price increase, GDP increase, additional demand. Clients will not have moved and changed everything in the next two years. We're talking about two years out. Yes, we do believe that we'll not see growth there, but we'll not see a big deflation as well because there is additional demand. There's additional work going in some of the traditional stacks that still continues. Companies will not have changed everything in the next couple of years. That's why we believe it will remain flat. On the gross margin? On the gross margin more. The expectation as has been in the past, the gross margin should continue to improve in the future. The reason that is because we continuously will focus on the biggest lever, which is the whole portfolio makes the AI as a value driver. As we do more AI, there is more of the transformation we do, the outcome base we do, the value creation we do. We expect to be able to take a bigger part of that value creation. Therefore, the expectation will be the gross margin will continue to improve over a period of time. Yeah. Just to be clear, I think you really have to realize that we're not talking about increment and improvement at clients. We're talking about significant value creation. Whenever in history, at least for me in this industry, we have seen this significant leap of value creation, this is where we can capture the most margin progression. Okay? That's why, of course, here we're talking 2028. You have to look beyond 2028, because here we have puts and takes. In the next two or three year, we are in transition. We're going to be in transition for the next three to five years to a completely new wave. That's why you have puts and takes in terms of that, and you don't see as much margin expansion in the next two or three years. I think beyond that, there is a lot of potential because you create so much value that you can capture part of that, and the client capture a lot, and we are both happy. I think that really have to have significant change in terms of frame of mind. This is not an incremental progression. Honestly, if it's for incremental value, it's not worth doing. It's too complex, too risky, too much change in organization. Unless it really helps to leapfrog, nobody's going to do it because it is extremely complex. I think he had a question. I can tell you really miss our conversation at Sapphire this year. I would tell you, if I have to put it in one word, it's very interesting, these relationships right now. I think with the dramatic entrance of the AI- native, the ones you mentioned, Anthropic, OpenAI, and Palantir, it accentuates the need for triangulation, and it double down the fact that they need us more than ever. Now, the question is, do I believe that they will disappear from the map? No. Do I believe SAP will be a formidable data and application system of record? Yes. Now, the battle is on the functional extensions and the ability to deliver them quicker, faster, and better for those clients. The ability to move from on-premise to cloud quicker, cheaper, faster. That's the battle that is about to start. The point is, we all need each other, and our value in orchestrating that is essential. Yeah, go. Thank you. Thank you, Aiman and you Fernando. Frederic Boulan from Bank of America. Question on your business model evolution. Can you discuss a little bit how you see the transformation from a kind of traditional time and material towards outcome-based model, how you think this will play out, and impact your business model in the coming years? Yeah, again, I think, we talk about outcome-based because, yes, I think the difference is that a big part of our business was around deploying technology. Okay. We got on the engineering side, the operation side, but historically, has been around deploying technology. Technology has an indirect impact on value, okay. You implement SAP hoping that finance productivity will increase, you will be able to close book faster, et cetera. The difference here, the work we are doing leads to measurable value immediately. You see the impact. When you transform a process and you identify it, you can see the result once the work is finished. It is not an indirect impact, it is a direct impact. Hence, you can measure tangible value creation in dollars and cents. It is not something that could happen because now we have deployed the technology. That makes it fundamentally different because you can measure that, you can tie part of what you get from the client to actually that value creation. It is clear some of it is going to happen over time, and some of it will be still linked to a fixed cost or fixed price for the client, plus a share of what gets created there. I think this is really where the mentality is looking. When you look at some of our other contract, there's no contract which is purely outcome-based, okay? To be frank, I would not sign on something on which I don't have the levers or I have minimal levers. That mean tying my future or my payment to something that I have no control on. The difference here, because we are transforming the process, because we are improving the outcomes, I am involved in making it to the tangible value creation to which I'm happy, and I will create more certainty over time as I build experience to say, "Yes, I can tie more of my future to that because I know it can be delivered." Okay? Over time, our confidence in terms of delivering it increases. If I take an accounts payable process or an accounts receivable process, if I've identified 10x, I know where I put the agents, I know what we expect in terms of things, I know where we're going to have the most impact, I know exactly what we'll get at the end, yes, I can tie myself a little bit more to that. The time and material is already decreasing. Compared to where it was four or five years ago, we're probably at 50%, we're down to 35%. It will not disappear. I think they'll still be resource-based, but it might drop to 25%, 20%. I don't think it will drop below that. It's a gradual reduction because we see more and more transformational deals where clients are embarking us on larger transformation, which is not just about give me resources. But there will still be a resource-based business that will continue, even if it's much smaller than what it used to be three or four years ago. We go on this side now. Laurent? Thank you. Yes, good afternoon. It's Laurent Daure from Kepler Cheuvreux. I have a question on the Intelligent Operation. The contract had been renewed, I would say the last six months. We come with some savings comparing the cost Human-P lus-A gent versus human. What is happening on the size of the deals? Because I suppose you are adding new functionality, do you manage to sign on a higher scope and higher revenues despite the savings you're giving them? Any clarity on this would be very useful. You guessed right. First, we are proactive. In this environment, you don't want to wait until the end of the contract. We are proactively approaching many clients with whom we have deals, because the best client is the clients you already have, and proposing expansion of scope with higher level of savings and transformation. In most cases, we'll end up with larger deal. We are in that process, but today we are not waiting for the contract is ending in one year, they might expect us to give them more benefits. Let's wait. Let's not do. No, no. We are proactively going to the client, say, "Listen, there's a higher level of thing can be achieved, a higher level of efficiency, transformation, and better outcome. By the way, what we propose is to expand the scope, and renew now by anticipation and start to deliver you the additional value." It tends to go pretty well in terms of discussions. Charles? Michael Briest at UBS. Can we just dig in? Sorry. That's all right. I'm flattered. Sorry. Sorry, Michael. On Intelligent Operations, some really interesting examples from Franck around the automation savings. I guess if you were delivering that on top of a legacy contract that you were replacing internally, it would be quite deflationary. I heard elsewhere, what was it? 40x increase in the number of agents out there over the next five, six years to 1 billion. Yet you're also saying Intelligent Operations, 100,000 people will grow. Can we sort of unpick that? It is not the math, but I'll explain to you with simple math. Okay? If you have a business that has 100,000 people, I'm not saying we're doing 50%, but you do 50% productivity, you double the size of the business, you still have 100,000 people business. What we're saying is that because the business is going to grow significantly, even with delivering very high level of productivity, overall, we do expect the size of our people working in operation to basically remain flat or even slightly increase depending at the pace at which we are growing it. At a group level, how much will headcount grow over the next three years? Listen, we're not putting a number. The problem, I don't want to go into numbers. As you know, it's not headcounts. Depending where the headcounts are, it's not at all the same level of impact. I do not believe that we'll see any big growth in headcounts. The growth of headcount will be much less than the revenue growth. Okay. That for sure. Bit by bit, we are identifying and able to drive that human and agent workforce. I don't say productivity. The reason why I don't use the word productivity. Productivity is a mistake in terms of concept. Okay. I just want to spend a moment on that. When you transform a claim process and you bring agents in, you're not doing human productivity. You are basically mixing a human and agent workforce on the same process. Agents consume tokens, so you're actually paying them a salary. If it was productivity, I would make an investment in an ACP system, and I have direct productivity on my process. This is not what's happening. This is I'm changing the mix of who is delivering the process. There's a cost, which can be significant if it's not properly managed for the digital agents. It's important to think about it this way as you think about the future, is that you look at the overall tech plus operation cost. I am spending so much to support my claim process from a technology perspective, application perspective. I have so many people working, and this is how much I'm spending from a people cost. That overall cost is my basket I'm starting with, and when I build my business case, is by what am I replacing it? You know what? The application will become simpler because I'm moving some of the functionalities into the agents who are now working on the process. What I look for in optimization is that overall cost, how do I optimize it? What is the business case if I was to go and drive this transformation? Okay? The simplification to say it's productivity cost is a fundamental mistake. The second one is to consider tokens, because I heard some people do it's a technology cost. Token is an operational cost. You have digital workers in your process that consume tokens. You're paying them a salary. Token is an operational cost in the process. It is not a technology cost. As a client, me in my company, when I look at how we're going to optimize the process, I consider token is not an IT cost. I say, "How much are you going to consume in tokens?" When I look at my delivery teams and we tell them, "Okay, we're going to give you whichever model, OpenAI, Claude, Gemini, whatever, to be able to improve quality, speed, et cetera." I ask them, "What am I getting in return?" The only thing you're going to see is an increase in cost. For me, the token in that case has become a delivery cost. It's part of my operating cost of delivery. It's not an IT cost. All these concepts, I think, are very important. If not, we're going to make big reasoning mistake in terms of how we look at overall this new token economy and how we look at digital agent and the consumption of token. If you may want to make it more complicated, the challenge between the digital workers and the human workers, is the digital workers, you pay them a variable salary because their token consumption is not steady. It makes it even more complex to be able to manage. Yes, go ahead. Yeah. Hi, it's Toby Ogg from JP Morgan. Bigger picture question. Clearly, there continues to be concerns in the market around the growth outlook for the whole of the IT services sector. Not just Capgemini because of AI. You've clearly laid out your view this afternoon that AI is a net tailwind for you guys, not a headwind. What do you think the market is getting wrong or missing? Is there anything very specific that you can point to that's not being properly understood? Thank you. I think what is not properly understood is basically what we just showed. What does it mean to scale an agentic enterprise? What are the different things that need to be changed? From some of the obvious one, like the tech debt. That's the first value pool, everybody got that. The technology stack, I don't think people are getting that. The complexity of building the new agentic technology stack. There's a difference between giving a developer Claude to improve his software productivity and getting agents to operate on a critical process. It's fundamentally different. If you want to be able to do that, you need to build the technology stack. The control plane. Do you want to have tens of thousands of digital workers running across your organization with unlimited access to making changes without any governance and unlimited cost? Probably you don't want to. You're going to have chaos. How do you actually get the value out of that? The transformation on the product and service side and the transformation on the enterprise process side. Well, I don't think people have been presented with that picture. This is the kind of transformation that enterprises will have to go through if they really want to become an agentic enterprise, which is different from, "I have deployed a few agents right and left, and I'm getting some benefit." Here, we're talking about fundamental transformation. You're talking about processes. I can tell you in the future, you might be talking about changes of organization because you might be looking at end-to-end processes that cross-function, and the way you organize your function might not make sense anymore in an agentic world. We're talking about significant enterprise transformation, which is not going to happen overnight. The potential in terms of value creation is so huge that basically people are going to get on there, and will start bit by bit on that learning curve, because it is a learning curve, process after process, area after area, building the Agentic Transformation. For that, they better have what I call the prerequisite, which is the agent tech stack and the control plane. If you don't have that in place and you start doing massive transformation, you might end up in uncontrollable environment. Through that, I think what fundamentally changes is that we are not tapping in the IT cost or in a transformation budget. Our play becomes the whole operating cost of the firm because that's what you're looking to optimize. It's not one specific pool, it's the whole operating cost. For enterprises, that's what's going to be at play in the Agentic Transformation. Yes. Go ahead. Thank you. Hi. Balajee Tirupati from Citi. Thank you for hosting us and taking us through in detail your view of the industry and Capgemini's evolution with agentic AI, qualitatively as well as quantitatively. Firstly, the five AI-driven value pools. I appreciate the unprecedented pace of technology evolution and enterprise willingness to adopt the technology. However, opportunity and timeline-wise, where you see these drivers evolution over coming years in terms of contribution in your 2028 growth outlook? Then in that context, in 2026, between AI incremental opportunity versus compression, are we estimating net contribution or headwind to the revenue growth? Second part of the question, if I may. Do you see the engagement from enterprises to be ready to harness the technology as well as the availability of new tools actually presenting an opportunity for a compressed modernization? For a? For a compressed modernization. What do you mean by that? Faster modernization? Faster modernization from. You want to go first? Yeah. If I look at your first question, Balajee, the expectation is that we expect that if we consider the secular demand as we talked about it, as well as if you look at the compression, we expect there or thereabouts for it to be sort of flat. Actually, if I come back to Fernando's full TAM, if you just considered those two items, then we'd be flat until 2028 or beyond. The excitement or the interest is really those other value pools beyond, which is the tech debt modernization, the Intelligent Operations, and the other value pools. Capturing that element of demand is very important for us because that will, of course, then give us that net incremental as we go on and moving forwards. You ask about 2026. We already have some of that playing on- Yeah of course. Net, we are still growing. We still have positive organic growth. Q1 was good, Q2 looks good. I think overall, we're seeing that playing across already in a number of areas because we see the agentic part growing, we see some of the deals on Intelligent Operation- Yeah that will start to kick in terms of higher level of revenues. We see, yes, there are areas of compression that are happening, of course, as we renew some of the deals. We have areas of expansion and areas of compression, and that's normal, and that's something we expect. Your second question is around. Compressed modernization. Yeah. Listen, I think so. The interesting thing is we start to see every CEO will say, "We have to accelerate AI adoption." Okay. What does that mean? I think can mean very different things. When we have discussions, and maybe I'll ask Roshan to talk about one or two example, we start discussing about things around the tech stack reset. Honestly, we have some CIOs or CTOs or even CEOs, because we talk to CEOs, in companies, "This is the first time I understand what is changing." Okay. We talk about the control plane. They get it. I have talked to chairmans, I've talked to CEOs, and I start to explain and project what does it mean to have an agentic workforce. You start to explain why you need the governance, why you need the control plane. It's good discussion. Now we're being asked by a number of client C-suite to run master classes for their top layer to explain what this new world looks like and what it implies. Roshan, maybe you want to talk about, you have to come here, talk about a couple of example of clients without, of course, saying who the client is. Yes this tech stack You hear me? reset. To complete, when you look at this new value pool, and what Fernando explained when we talk about the total addressable market, we said it's a combination of three wallets today. The first one, which is our historical wallet, which is the IT spend, around EUR 1.9 trillion, meaning the IT spend. There is a second wallet also because when we are saying what market is getting wrong about Capgemini is not just an IT company. We are the biggest engineering company in the world. It's 50,000 engineers. There is an engineering part, the IT part, and the BPO part. These three wallets, when you say now how clients are executing on that, I would talk about the Intelligent Operations today. Already we see an acceleration in this kind of deals. Here the objective is really first, well, we have seen Imperial Brands around cost takeout, more agility, more flexibility in the organization. Here agentification has concrete application, and we have a lot of ongoing deals. The second one, when you look into the Intelligent Industry space, the more complex mission-critical, today client starts applying it on specific projects first, not directly at enterprise- level because you understand the level of mission-critical operation. You can't apply it specific. They will start applying on a specific program. For instance, for an OEM today we are working, but big challenge, how do I compress the lead time in my car program development compared to Chinese? They are doing it much faster. For that, you can't completely change the stack. This will impact other projects. You start applying it on a specific program. You start thinking, okay, how will engineering, manufacturing, the R&D will reduce the lead time through agentic, removing a lot of interfaces. We start getting proof points. This is where we start getting a full tech stack, but apply on a particular program, then when it will get successful, then we can roll out because there's a cushion on adoption. When it comes to IT, Karine mentioned it, we see today applied to our delivery. Now what we are seeing with our clients, how we get through that, when we have big, large ADM contract that comes under renewal, or we have clients who are thinking about renewing, I would say, the contract with an SAP, with a Salesforce, comes a question around, this is our job as advisor first. The new tech stack looks like that. How do you plan the evolution so that you have a total cost of ownership of this tech stack which is economically viable? Where you start getting into this strategic discussion, reflecting with client, what is the architecture of a new tech stack and how they progress. You have seen the different layer. What stays as a system of record? The next three layer, do I do it with OpenAI? Do I do it with Anthropic or both of them, or Google? How do I transform the experience layer? This kind of big complex systemic transformation, we see today an appetite for our clients. Once we have big contract that are at stake, how do I reflect? Here we see it's a build that will run five, eight, 10 years. It's not a switch on, switch off. The more you anticipate, the faster you will build it because we know it's a competitive advantage for clients. The one who will move faster to that, they will leapfrog their peers in the industry. Thank you, Roshan. Just we're going to take the next question from the internet. Just to summarize part of what Roshan is saying, when we go for renewal, even like an ADM contract, we move the discussion from the renewal, just the renewal of the ADM contract, is how do we help them get ready for the next phase, which is basically getting the agentic tech stack. What is the compression on the ADM becomes a broader discussion around beyond the ADM renewal. Which, yes, there'll be some compression because we'll automate and use agentic and generative AI to be able to deliver, is actually how do we help use that time and that effort to basically get ready to have the agentic stack done. We have the next question from the internet. Allow me to jump in, yeah. You referred to a transition phase. What does that mean for financials short term and long term? No, listen, I think for the financial is, we have done transition in the past where there was some deflation, right? The deflation, I think Fernando explained it a bit, is it's at the front end. You don't get deflation over 10 years. We're taking the deflation now over the next three to five years. At the same time, we have the growth. When the deflation stops, you get accelerated growth, okay? I think the growth is tempered over the next three to five years, for the next three, for sure, five potentially, by the deflation. Once the deflation starts slowing down because it's not continuous, you will see a further growth acceleration, and that what we expect. To be frank, on the margin, the same thing. We are in transition on a number of things. As we tie more, as we deliver more value and there's more basically value being created at client, we can share more in that, and that provides the additional margin expansion that we want to go for. Yeah, go ahead. I will come to you. Yes. Thank you. Nicolas David, ODDO BHF. I have a question regarding, you mentioned several times that AI transformation is more a business transformation rather than a technology transformation. Given that, are you happy with your consulting skill right now, or do you need to scale up a bit business? Don't you think that it will open the door for business consulting players to enter more aggressively into this market and scale up also in the AI transformation? Is it why you are not factoring market share gains in your growth prospect for the next three years because your growth target is in line with the market growth? Thank you. Listen, I think it's a good question. Yes, we need to continue to reinforce some of our business consulting skills. Remember, this is not process re-engineering. It's process redesign or reinvention, and for that, you need to understand every step of the process. You need people who know how to run the process. A claim adjuster knows how to do that process. When we have claim adjusters in our operation, they can help redesign how it works. I'm sorry, a consultant cannot do that because a consultant day-to-day job is not to do claim adjustment, and that's where the big difference is. You need the consultants because they can help redesign, but you also need the people who are actually currently doing the claim adjustment, who know how to do that, and you need both, and you need the technology. It is not one thing, but it's a transformation because at the end, we are transforming the businesses. We are not deploying technology hoping for the better. We are actually, the process will have changed once we finish our work. The process change is not just about process re-engineering. It has all the other component, defining the agents, building the data products, checking for cybersecurity, and yes, there are some consulting aspect like the Human-AI interface, et cetera, that we have to work on. Okay. Of course, you heard some of the client testimonial that we have gone through reinforce some of this message. It's not as simple as a BPR. It doesn't stop there. Okay. You had a question. Go ahead. Hi, Sven Merkt from Barclays. Just one question maybe on core modernizations. I would be interested to hear what percentage of your revenue that represents today, how you think about how this might develop in the mix over the coming years. How quickly are your clients really able to work to all that tech debt, and what's the durability of this revenue stream? Franck, I'm not sure it's easy to measure because it always mixes with other things. Franck, you want to talk a bit about the core system modernization? The first thing I think important in modernization is, as you're most probably aware of it, most of the transactions today on the market are still run by mainframes, and it has been a totally untouched market. That's what we call a Blue Ocean opportunity, because you have hundreds and hundreds of mainframe to replace. The first thing we have done, we have done it in two step. We have built, roughly two years ago, an approach with generative AI that make possible the modernization of mainframes. Now we have agentified fully the process, we are seeing a huge acceleration, a huge number of opportunities on that. That's one example. When you think also about modernization, all what was presented earlier by Roshan, you have to modernize the different part of the company. That is, when you take a large company today, you have maybe 3,000, 4,000 custom applications. A large part of them have to be modernized. On that, it's also an opportunity because before, sometimes to modernize an application, you need six months. Now with the acceleration of AI, you can modernize your application landscape much faster and so on. It's the same for the data layer, it's the same for all the layers. We are seeing a strong appetite from clients to reduce their technical debt, because for them, their technical debt was really blocking their innovation capabilities. That's a huge opportunity on the market. Okay. Richard. Richard Nguyen for Bernstein. Thank you very much for the presentation. Today we talk a lot about the IT stack. My question is more about the R&D engineering part. Can you please tell us a little bit about how AI is impacting your business there regarding the pricing, regarding the business activity, things like that, and how do you see that evolve in the future? Thank you. Listen, it's no different than in IT. There are some areas, like areas of software, for example, we do see of course, an enhancement in terms of productivity. Even more on the product side, sometime on the commercial application side. At the same time, we're discussing with clients around how we're going to leverage a lot more, all the part of the discussion we're having with Siemens. How are we going to build agents who can do engineering work? It creates complexity. If you think about engineering, it tends sometime to be a bit more resource-based, right? There is a reason for that. Because if you want work to be done, you need to be able to specify the work. Most engineering don't work this way. People work together, they talk to each other, et cetera. They enable. If you want to have agent deliver something, you're going to have to learn how to specify things, okay? I can promise you, when talk to a lot of engineering client, it's extremely complicated. This is why often you see it as we bring engineers to client to work with their teams, as opposed, they specify plenty of thing, they give it to us and say, "Go do that for me." Why? Because that means they would know how to do specifications. You're going to see in many companies it's extremely difficult to do, because writing specification to define the work, which what you need to tell the agents to do, is extremely complex. On one area, on the software, it's moving faster. In some other areas, it's going to be slower because of that. You have all the transformation as well that's coming from the physical AI that are creating new opportunity. It's one of the areas we are working on. We even have our deep tech arm in Cambridge where we're already doing very advanced things in some of these areas around physical AI, including rebuilding the whole software for some humanoids, et cetera, and thing like that. There are compression areas, there are expansion areas, there's all the new areas in terms of helping to develop that Roshan talked about. Develop the new products and services of the future, embedding agents. As you embed agents, what we call the V&V, the verification and validation, becomes even more critical, because now you have to make sure that the agents' safeguards, rail guards, everything, is much better even than in some of the commercial processes you're using. There is, yes, plenty of potential, and yes, there is compression on some areas like you'd have in IT or in process, but you also have a lot of new areas where we see opportunities to be able to develop. Again, it's very interesting discussion with clients because it's also new for them. It's not like all the product exists. There's some clients with whom we had discussion in aerospace about basically co-investing to build some of the specific engineering capability, because we know there is shortage in the market of some of the skills. The question is. How do we focus building some of these engineering agents, if you want, in areas where we expect shortage of skill to be able to compensate some of that and not to slow down some of the potential development work that some of these clients need to do in the future? We'll take one more question. I'll take two questions. Okay, three questions. They told me one question, but I'll take three. One, two, three. Okay. Hi, it's George Owen from Morgan Stanley. Talking about customer behavior a little bit, and it was mentioned earlier, and we hear it more broadly as well. There's a little bit of, I guess, paralysis out there from some customers who don't know what the right technology bets are to be making at this point in time. If you look across the broader ecosystem, actually, you're outperforming organic growth at the moment, but generally speaking, at a low ebb. What do you think it's going to take for customers to get more confident at making these big decisions? Because it doesn't feel like the pace of technology clarity or transformation is going to be easier in six months' time compared to today. What might it take for customers to gain that confidence? Listen, I think there's different areas. I think on the technology debt is probably the area that's moving faster. On some of the other areas, what's going to be challenging is the scalability of skills. The limitation might not be client ambition to want to do, is basically do you have enough people who can know how to architect the future of business, reinvent the processes? It's a learning experience. We are learning. We're deploying agents. It takes time for clients even to approve, sometimes up to six months before we can deploy anything because they're afraid, because of the guardrail, because of legal, because of compliance, et cetera. What is going to slow down everything is not the technology, it's the skills and the governance to get the decisions made for clients to be ready to take the risks to make it happen. Okay? It's an experience curve. Of course, we're going to go with the experience curve. There are some areas like you say, okay, software development, everybody is going to be using some form of LLM to help become more productive. When you come to deploying enterprise critical process, it's going to be a little bit slower because what's at stake if something goes wrong is a lot more critical than if a software developer has developed some bad code. Yep. I said one, two. Yeah. Three. No, I took one of your question. You can try at the cocktail after. Thanks very much. It is Ben Castillo-Bernaus from BNP Paribas. Fernando, a question for you on that market growth slide. It has got a lot of attention, as you may have expected. What underpins your confidence that that AI compression effect moderates or is temporary in some way? Why does that not just keep accelerating as a headwind, as the technology improves and therefore clients want to extract even more efficiencies? Maybe asked a different way, when might that AI compression headwind peak, in your view? How far along in the journey are we? Thanks. It's a good question. There's a lot of intangible elements around it, unfortunately. It's based upon, to be quite honest, we created a taxonomy that defines what typically, in terms of the analysis I've done, tends to compress more than other elements of the taxonomy. It will not go away. This is a transitory process. We are basically doing our best assessment on information that we have at hand in partnership with what McKinsey has provided to us. The calculations that we did is definitely, and I think maybe express it in the context of gap, it is noticeable, it's happening right now. We believe in the elements that define that compression. They tend to be less effective or diminish post after the five-year term based upon our best estimates on the information that we have in front of us. I believe after five years, it should not be even called a compression. It is what it is. It's very difficult to play with the time frames because to be honest, it's moving very fast. Remember, it's a balancing act. Yes. Because if there's compression, that means there's acceleration on the other side. Okay? That mean we're able to deploy faster, that mean there's more acceleration of growth on the new agentic stack. Yes for them to be more compression. There are certain things which are a bit upfront trivial that we get to quickly, but suddenly you're going to be very limited in terms of compression unless you're able to accelerate on the other side. That's why it's a balancing act. If it compresses faster, that means it's moving faster on the other side as well. The growth is faster. I have the last question. Well, thanks. Nooshin Nejati from Deutsche Bank. I appreciate the strong momentum in AI demand and sizable Intelligent Operations pipeline. Could you help us bring that into financial delivery? Are you seeing faster conversion from AI pipeline into bookings? Once signed, how quickly do these projects typically convert into revenue? Listen, again, it depends on the nature of the work. When it is Transform and Operate, like some of the Intelligent Operations, it takes a bit of time, but then the ramp-up is massive because the first thing you get is a massive transfer of people, which of course boosts the revenue. It takes a bit of time to be able to make the transfer because it's more complex in a certain way. When it works on some of the transformation program. The revenue tends to come a bit faster, right? You ramp up the team once the client decide, and we're able to start the project team much faster. In certain way, the revenue tends to be a bit faster both in terms of decision-making and in terms of work. Again, it's going to be a gradual thing because no client is going to say, "Okay, come and transform my enterprise and here is 15 processes. Bring your 300 consultants and change everything." I don't think anybody's going to do that because of the risk associated. People are all going through a learning curve. We work with insurance company, we started on claim. One of the big things, for example, is underwriting. Everybody's rushing to see how they can do faster underwrite. Why? They have to answer to brokers, and if they're not fast enough, then they're not in the game anymore. If the others are all agentified their process of underwriting, and they're the one who are not agentified, they cannot submit quotes fast enough compared to their competitors, and they're out of the game. Suddenly, everybody's rushing to see how to agentify underwriting. There are cycles depending on the industries that are quite different. We're working on some very advanced things, and Karine Brunet is involved in that with telco clients to look at trying to build completely autonomous networks. Okay? That will be a massive program if it starts. Very complex to deliver, but definitely if you do the first one. Beautiful There's a whole bunch of others behind. This is the way we think. Remember, the whole discussion that we have had is that it's very industry-focused. Autonomous network is very specific to telcos. If we do it with one client, there's a whole lineup of clients that will be interested and able to do that. That's where we look a bit to pay some of the investments that we have to do. Thank you very much. Aiman, Nive, Fernando, thank you very much. Thank you very much, ladies and gentlemen. This marks the end of our Capital Markets Day. Thank you very much for your presence and your attention, your presence here and online. There were more than 350 of you, and we hope to see you soon. Thank you very much.