eDreams ODIGEO S.A. ($EDR)

Good afternoon, everyone. I'm David de Roz, Head of Investor Relations. Over the past few months, we have engaged with many of you across global road shows and conference and one very interesting topic has dominated many of our conversations, the rise of agentic AI. And so today's session is about our next phase of growth for eDO. For over a decade, we have been building a specialized AI ecosystem that powers the world's largest travel subscription platform. We have invited you here today because we are at a unique inflection point. The recent developments in generative and agentic AI don't just complement our business. They act as a force multiplier for our unique competitive advantages. Today, we will show you how we are harnessing these technologies to accelerate our mission, scale our prime membership and redefine the new era of travel. I will now pass you to our CEO, Dana Dunne, who will take you through the agenda and first part of today's presentation.

Thank you, David. At eDO, we've always been pioneers. Leadership in any industry is defined by the ability to anticipate the path. At eDO, we didn't wait for the AI hype cycle to begin. We recognized more than a decade ago that the future of travel wouldn't be found in generic search results, but in intelligent personalized curation. We spent the last 10-plus years feeding our algorithms with the behaviors of billions of travelers to create an AI-driven engine that now operates with unrivaled predictive precision, serving as the intelligent backbone that optimizes every touch point of our global business. What the world is seeing now a shift towards automated personalized digital assistance is a future we have been architecting for a long time. While many are just beginning to explore the potential of AI, we are already operating at scale. Our position is clear. eDO is the natural beneficiary of the AI revolution. We possess the three ingredients that cannot be easily replicated by newcomers or legacy players, a massive proprietary data set, a recurring subscription audience and the proven ability to turn technology into high-margin growth. As the digital landscape evolves, travel is becoming more specialized, not less. We are seeing a clear bifurcation in the market between discovery and fulfillment. While the new generation of digital assistance is making discovery more intuitive and conversational, discovery is only about 10% of the journey. The other 90% is the execution, the heavy lifting of navigating billions of flight combinations, managing real-time inventory and providing a global safety net for travelers. Yes, the magic of AI-driven travel only works if there's a specialized engine behind it that can handle 3 billion price changes a month, manage intricate multi-carrier and multiproduct itineraries and provide 24/7 global services. That's our territory. We are the travel fulfillment powerhouse because we are uniquely positioned to be the primary transaction layer for the next generation of AI interfaces. We aren't just part of the ecosystem. We are an essential infrastructure that makes it work. Today, we'll detail how we are turning our AI leadership strategy into tangible shareholder value through the 5 key chapters outlined in this presentation. The first, eDO, an AI-first company since more than a decade. In other words, our AI native maturity. We will look at why our long-standing AI expertise provides us with a proprietary competitive advantage that is clearly ahead of the industry. Two, where we are today. Our subscription advantage will explain why Prime, our unique subscription model and our ability to build complex multi-carrier itineraries create a value proposition that is difficult to commoditize and that turn the world of one-off transactions into long-term high-valued relationships. Three, looking ahead and capitalizing on the new traveler, how we are positioning ourselves to be the destination of choice for the next generation of AI-driven traffic. Four, embracing new AI opportunities, a focus on the speed revolution, how we are using agentic AI to supercharge our internal productivity, allowing us to build more and innovate faster than ever before. And then five, cover the closing remarks. We are positioned for future success. And finally, we will show you why eDO's position as the world's leading travel subscription platform ensures our continued success in an AI-first world. I'll be covering Chapters 1 and 2, followed by Carsten Bernhard, our Chief Technology Officer, who will talk about our new travel in Chapter 3 and lead the discussion of AI opportunities in Chapter 4. After that, I will return with the closing remarks before we open the floor to your questions. So with that, let's start with Chapter 1, our AI heritage. Turn to Slide 5. To understand the trajectory of eDO, you must understand that we aren't just using AI. We are defined by it. While many in the travel industry are currently reacting and scrambling to pivot to AI, at EDO, there is no pivot. We have been an AI-first company for over a decade. For us, AI is at our very core, fully integrated into every micro macro decision we make. Our journey began in 2014. While the rest of the travel industry was still focused on basic digital distribution, we established our first in-house data science team. Our first mission was critical, building a spoke AI fraud prevention system. The results were more than just a success. We were transformative, improving our margins and security so significantly that it sent a shock to our entire organization. By proving that data was our most potent competitive lever, we achieved a level of margin security and operational precision that was previously thought impossible. By 2015, we made the strategic decision to go AI first. This wasn't just a tagline. We decentralized AI, making it available to every team across EDU. We didn't wait for off-the-shelf solutions. We built our own proprietary reinforcement learning algorithms to power our pricing. To put that in perspective, the reinforced learning tech we were using a decade ago is the same fundamental logic at the core of modern agentic AI today. By 2016, we had scaled personalization to the individual customer level. While competitors were still using broad segments, we were treating millions of customers as individuals. In 2017, we deployed what remains one of the great moats, a proprietary reinforcement learning framework for pricing. This combined genetic algorithms with multi-objective optimization, effectively building a pricing engine that learns the market in real time. In 2019, we were leading pioneers in generative AI. This is long before the world knew what a large language model was. We were deploying generative models, not to predict words, but to predict and create unique itineraries. We used our unique advantage, others did not have this capability of having a world-class flight booking business and AI to mimic the world's flight data to build travel routes that do not exist on any single airline website. This is how we became a creator of proprietary and unique travel products. During the pandemic years, we didn't slow down. We moved AI into the front lines of the customer service and performance marketing. Please turn to Slide 6. By 2023, we were recognized by global giants like Google as a world leader in applied AI. At this point, AI wasn't just in our tech. It was in our HR, our finance, our operations, et cetera. 2024 was about developing in-house AI agents, expanding copilot tools and strengthening our technical collaboration with AI industry giants. But 2025 and 2026 represent the most significant leap in our history, our agentic era. We have moved from chatbots to voice-based servicing agents and an agentic replanner that uses RAG, which is retrieval augmented generation and MCP, modal context protocol. We currently have over 100 MCPs deployed, allowing our AI to actually act on our proprietary data. As we look towards the end of 2026, we are pioneering agentic distribution. We aren't just waiting for traffic. We are integrating directly into the ecosystem of ChatGPT, Claude and Gemini. Internally, we have moved into agentic development. Using Claude and Cursor, we've moved from writing code to writing intent. The takeaway for you today is simple. We are architecting 2026 on a foundation that started in 2014. You can buy software, but you cannot buy a 10-year head start. Please turn to Slide 7. A decade-long head start in AI isn't just a technical achievement. It is a powerful driver of customer trust and a fundamental competitive advantage. As you see on this slide, AI has been instrumental in transforming the customer experience within our Prime program. In terms of Trustpilot evolution, when AI was relatively nascent in the industry, in August 2021, eDO already led the category. We were already an AI-first company. Since then, we haven't just improved, we have accelerated. Our Trustpilot scores have surged by 33%, reaching a record score of 4.4 now in May 2026. We are outperforming the competition. To put this into perspective, while other [ OTAs ] have seen a 19% increase, they are still struggling with an average score of just 1.9. Airlines have barely moved, inching up 6% to a mere 1.7. We aren't just ahead. We operate in a completely different league. Our satisfaction levels are now multiple higher than the industry average, proving that our AI-led model is solving the friction points that our competitors haven't addressed yet. Beyond customer service, this improvement isn't just about automation and support. We have infused AI across the entire customer journey. We've used it to hyper-optimize the booking funnel and deliver personalized propositions that resonate with individual customers' needs in real time. We don't just sell travel. We anticipate it. All in all, we have a leading advantage. This brings us to a critical strategic moat. In the emerging era of agentic search, trust is critical currency. AI platforms, the Googles, OpenAIs of the world want to integrate with trusted, high-quality vertical providers that users actually love. Our industry-leading NPS and satisfaction scores prove that eDO is a partner of choice. We are the travel platform with the data sophistication and the brand trust required to win in an AI intermediated world. Let's now move to our competitive advantage and strategy, what exists today and what to expect from AI platforms in the future. Please turn to Slide 9. History tends to repeat itself. Every time a transformative technology emerges, a familiar debate resurfaces, the fear of disintermediation. Today, we've seen the same discussion happening again with the rise of agentic AI. The reality, however, is fundamentally different. While customers are increasingly turning to AI for travel inspiration and discovery, they're not using it for the retail experience. General AI interfaces, whether it's ChatGPT or Google Gemini, lack the deep retail and fulfillment capabilities within travel that travelers require. We see these platforms as partners and new acquisition channels, not competitors. They are masters of intentive and discovery, but they are structurally disincentivized from taking the full accountability for the journey. They want to avoid the massive technical, regulatory, financial and operational complexities of travel fulfillment, areas where eDO has spent decades building a dominant competitive advantage. And this is a strategic advantage. This is where our model becomes truly ideal. We hold the trust of consumers and the proprietary data required to turn AI-driven inspiration into reality. In short, AI platforms may provide the traffic, and eDO will provide the travel. We are the engine that will power its transition from a search tool to booking reality. And what does the future hold? We are moving towards a world of seamless automated execution with exciting emerging opportunities. Please turn to Slide 10. In travel, discovery and fulfillment optimize for completely different economic objectives. General AI platforms optimize for information retrieval and user engagement. We optimize for the complex, highly precision world of travel logistics. Our unique proposition cannot easily be commoditized. Unlike a standard transaction on a competitor site, a prime subscription is not a commodity. It is a long-term, high-value relationship built on a proprietary pricing engine and exclusive inventory that a general AI agent cannot simply replicate. On the contrary, as highlighted just now, AI platforms are structurally disincentivized from pursuing disintermediation. Let's look at specific reality on Slide 11. We've broken this down into our strategic advantages versus the structural barriers facing general AI. First, in Box A, we have our direct-to-consumer engine. Our proprietary products and eDO Prime subscription model create a powerful pull factor. Because we offer exclusive value that cannot be found elsewhere, our members have a clear incentive to overcome -- sorry, to come to us directly. Now if we look at Box B, we can see why AI platforms are focused on a different path. These platforms face two massive structural barriers to entry into travel retailing. The fulfillment barrier. AI platforms will not own the booking experience because they're unwilling to make a merchant of record liability. This means assuming the financial risk for [ chargebacks, ] refunds, insolvency, and this just names a small few. There's the operational barrier. Travel is full of edge cases. General AI is not designed for post-booking servicing or complex disruption handling. The high friction work that happens after the flight that is part of a package is canceled at 2:00 a.m. in the morning. Instead of fighting these barriers, their actual strategic objective is advertising monetization. Travel is one of the largest digital ad categories globally and AI platforms benefit from the massive ad budgets of established [ OTAs. ] They want to be the world's most sophisticated search engine. We are the engine that actually creates the real trip. Please turn to Slide 13. To understand why we are structurally resilient to disintermediation, you have to look at what we actually sell. eDO has evolved far beyond being a simple intermediary of plain vanilla travel content. We have spent years building a portfolio of differentiated products that are extremely difficult to commoditize. Our competitive advantage is built on a powerful combination of four key pillars: One, Prime subscription model. Prime transforms EDO from a transactional shop into a subscription-led ecosystem with structurally higher LTV and retention. Prime provides an exclusive unmatched advantage that incentivizes our almost 8 million members to come to us directly rather than starting their journey on generic search. Two, unique proprietary propositions. We create unique itineraries, combining, for example, rail and flight that are not available anywhere else. We offer eDO exclusive products. Three, AI-first agility. As an AI-first company for over a decade, we are significantly more nimble and transformative than many other legacy OTA competitors. Four, global content access. We manage billions of monthly searches, providing a clean high-performance signal to our customers through world-class AI caching. Within all of this, what is critical to understand is that transportation is the ultimate complexity play. The flight and rail services sectors offer endless possibilities for us to create superior, complex and unique solutions. Let's now turn to Slide 14. Most OTAs are simply windows into plain vanilla content, the same flight and hotel that you find everywhere else. We have evolved far beyond that. We use AI to orchestrate billions of flight and rail combinations daily to build unique itineraries that simply don't exist with any single provider. In other words, we create travel. Take the example of a trip combining two different airlines. In this case, a combination of British Airways and Lufthansa. No single airline can sell that. No general AI agent can fulfill that with the necessary financial and operational protections. However, we can manufacture, engineer this connection, offering the customer one single point of contact for servicing and disruption management with unique proprietary features. We sell engineered value, not just seat numbers. We are providing a proprietary solution that makes eDO the essential non-commoditized destination for the modern traveler. Now let's take this one step further. Here, we can see another example, a return trip from Madrid to Barcelona, combining a train and an airline departing with AV, which stands for Spanish high-speed train. And that's with a return flight with Air Europa. This is a complex multimodal journey, a high-speed rail connection feeding into a flight. No single airline can sell that, no train provider can sell that. No general AI agent, no matter how clever can fulfill that with the necessary safeguards I mentioned before. We are the platform capable of engineering this connection. Let's not stop there. In fact, here's another extremely interesting example, which perfectly demonstrates what we mean when we say we engineer value where others only see fragmentation. We are looking at the journey from Toledo to Berlin. Toledo is a major cultural hub in Central Spain, but it has no airport. Traditionally, this traveler would have managed three separate interfaces, one for the high-speed rail to Madrid, another for the flight to Berlin and a third to coordinate the logistics and risk of the connection. eDO eliminates that friction, that hassle and all of the associated risks. Our AI native platform orchestrates this entire journey into a single protected transaction. We bridge the gap between rail providers and flag carriers, in this case, Iberia, to create a itinerary that simply does not exist anywhere else in the market. An AI chatbot can tell you a train exists from Toledo to Madrid, but it cannot bond that train to a flight, issue a single ticket and provide the connection guarantee. If that train is delayed and the traveler misses their flight, eDO manages the disruption and re-accommodation. And even at the purchase stage, it is one purchase, not separate ones that open the customer to additional risks to price increases and/or availability issues and having in the end a ticket for only one part of their travel journey. This is the heavy lifting of travel and of eDO. The three examples we've just seen are not just clever combinations. They represent the creation of smart inventory. By orchestrating fragmented carriers and multimodal providers into a single itinerary, we are providing options that simply do not exist. Even as we move into an era of agentic AI, we represent the technical back-end integration that makes the trip a reality. A general AI agent can talk about a trip, but it cannot build it. It lacks the depth, the real-time API connections to link a high-speed train to a low-cost carrier while simultaneously layering on the connection guarantees and financial protections that the travelers demand. We are providing the connective tissue for the entire ecosystem. Without our back end, an AI agent is a travel agent with no tickets to sell. By mastering this complexity, we ensure that eDO is not just a destination for consumers, but an essential execution engine that future AI platforms can rely on to make travel a reality. Please turn to Slide 19. Beyond the itinerary itself, we have engineered a layer of proprietary ancillaries that redefine product flexibility. We aren't just reselling standard airline add-ons. We are creating our own. Features like price freeze, cancel for any reason, et cetera, are either innovations that solve the primary pain point of the modern traveler, uncertainty. These products are high-precision AI-driven solutions. Our ability to offer these features depends upon our capacity to predict price volatility and cancellation probabilities across billions of data points. This is an unmatched level of flexibility that a plain vanilla OTA simply cannot price and a general AI agent cannot fulfill. Also, these ancillaries are a massive driver for Prime conversion retention. When a travel knows they can freeze a price or walk away from a booking with one click, eDO becomes their indispensable partner. We aren't just selling travel. We are selling peace of mind, freedom and flexibility, all powered by a decade of AI-first risk modeling. In this first example, the price freeze allows customers to hold a price for up to 7 days for as little as EUR 1 or EUR 0 for Prime repeat members. We all know that freight prices on average increase over time and leisure customers want to take time to check with friends and family, consider other options, et cetera. Here's another example. Traditionally in travel, flexibility has been a luxury for leisure travelers. Carriers often charge hundreds of euros, sometimes exceeding the original fare to provide the basic ability to cancel or refund a ticket. This is a legacy profit center built on consumer friction. We have completely disrupted this dynamic. By leveraging our massive data scale and AI-driven risk modeling, we can offer the same or superior flexibility for a fraction of the price, creating a powerful competitive advantage. This isn't just a feature. It is a customer acquisition and retention machine. It reinforces our position as the essential non-commoditized destination for travelers who demand both value, peace of mind, and flexibility. And yet another example, in traditional travel, flexibility is siloed. If you book a flight with carrier A, your ticket will be locked into carrier A's rigid rules. But travel is rarely that simple. Our Flare Plus product breaks these silos by offering universal flexibility like name and date cancellations that can be applied across different airlines within a single itinerary. This adds massive value to a trip and our customers because we provide a level of freedom that makes eDO the logical choice for travel. And we don't just stop there. We continue to develop additional proprietary products. If you could please turn to Slide 24. Now let's talk about flight plus hotel package protection. If a traveler books a flight and a hotel separately via a general AI agent or a traditional intermediary, they're holding two independent unprotected contracts. If the flight is canceled, the hotel remains nonrefundable for the consumer. Through our package protection by [ Atel ] in the U.K. and the EU package Directive in the EU, we provide an additional level of security. We take full responsibility for the entire itinerary. If one element fails, we orchestrate the solution for the whole. Again, general AI agents are structurally designed to avoid this level of financial and operational accountability. They want the click and we want the responsibility. By offering this protection, we aren't just selling a holiday. We are providing insurance for the experience. And that's really the ultimate driver of trust, ensures that our trips are structurally superior and safer. Here on Slide 25, we see the two biggest reasons why customers choose to book package with eDO instead of booking parts separately. When our AI combines a flight and a hotel, it unlocks exclusive package rates. Our AI finds hidden deals creating a final price that is difficult to match if you try to book the flight and hotel individually. It's a win-win. The customer saves money, and we drive higher conversion. But as I mentioned before, the real advantage is protection. When you book a package with EU, you are legally protected by the EU package, Travel Directive or ATO in the U.K. Now imagine a customer booking a flight today and within the next 24 hours, an AI agent has a hotel on top of it. In the eyes of EU law, that's no longer two separate bookings. It is formally legally a travel package. And because the AI agent put the two pieces together, it legally becomes the package organizer. Now that sounds like a small label, but it actually has massive implications. The organizer has to register with the regulator in every European country it operates. It has to put real money aside, be it through bank guarantees, insurance or trust accounts so that if any supplier on the trip goes bankrupt or there is a major disruption, customers get their money back or be provided with alternative travel plans. It's also legally on the hook for every part of the trip. If the hotel overbooks or the airline canceled, the agent owns the problem. On top of that, the organizer has to give customers specific information upfront, help them on when things go wrong and stay reachable when the trip gets disrupted, which in practice means running a 24/7 multilingual customer service team. And to make this more complex, this isn't one rulebook. It's 27 national versions across the EU, plus the U.K., plus the EA and Switzerland, each with its own registration body, financial requirements and enforcement regime. So for an AI agent to generally act as a travel intermediary in Europe, it doesn't just need to be smart. It needs to take on the whole compliance, capital, customer service and be the traditional tour operator. The barrier here isn't technology, it's structural, and it puts a real ceiling on how independently these agents can actually transact and travel. And we've built all of this over decades. Let's turn to Slide 27. One of the most critical technical competitive advantages is how we handle the scale of global travel data through search volume management and caching. We've evolved far beyond basic distribution to become a highly specialized technical intermediary. Airlines and hotels simply do not have the bandwidth to answer millions of search queries every day. We process over 100 million searches every day, a volume that would overwhelm most carrier systems. By protecting our airline and hotel partners from this massive amount of search queries, we ensure the long-term sustainability of our content acquisition and act as a protection layer for them. Our powerful caching technology stores accurate prices so we don't overwhelm airline systems. As a truly multimodal platform, including air, rail, cars and hotels, our technical infrastructure positions us as the leading one-stop shop for customers and AI agents alike, providing a level of reliability and depth that generic platforms simply cannot replicate. Please turn to Slide 28. To address the risk of disintermediation, we have to look at the moat provided by our Prime subscription model. Generalist AI platforms cannot easily replicate the highly specialized vertical-specific features we have spent over a decade perfecting and that we continue to develop. Think of the difference between a general purpose smartphone and a specialized garment device for professional cycling. While a phone has an app for everything, cyclists pay for the specialized device because it is built for a single complex task. The same is true at the core of the tech industry. You have general purpose CPUs that run basic PC tasks, and then you have specialized NVIDIA chips engineered specifically for the massive demands of AI. That is the relationship between a general LLM and EDU. We aren't a generalist tool. We are a highly precision instrument and the specialist silicon for travel. Our specialization is defined by the following layers. Guaranteed value and prioritization. We don't just offer a price. We offer the best price guarantee and 24/7 VIP support. We have unparalleled flexibility. Price fees, cancel for any reason products, just to name a few, apply across all providers. To get this level of flexibility from the airline, a customer would typically have to buy the most expensive full fare business ticket. We democratize elite level flexibility for every traveler. Navigating the messy core. There is an immense amount of back-end heavy lifting here. For example, our automated check-in must navigate the disparate laws and proprietary systems of every country we operate in. Generalist AI platforms are built to be broad. They're not built to handle the legal and technical friction of over 200 different jurisdictions. The instant wallet, our advanced refund lands in a customer's eDO wallet instantly. Crucially, these funds can be used across any airline or travel product, not just the one originally booked. We have created a universal travel currency. Generalist AI is focused on being everything to anyone. We are focused on being the specialized chip for travel. We handle the messy, the complex core of travel commerce. Let's move to Slide 29. Here, you see how we have structurally decommoditized our business. To become the preferred partner for both travelers and AI platforms, we have built a pyramid of value. At the base are standard return flights with a single carrier. This is a commodity. Any airline or basic intermediary can do this. One level up, we use our technology to manufacture return journeys using two different airlines. This immediately gives our customers more choice and better pricing than any single carrier could ever offer. Climbing to the next level, we layer on our advanced proprietary ancillaries, products like price fees and cancel for any reason provide high-value flexibility. Very few competitors have the data sophistication to price this risk, let alone offer it. At the Summit, we have Prime. This is our unique crown jewel. It combines top-tier service, cross-sell integration across rail, car, hotels and deep functionalities found nowhere else. By building this multiproduct platform, we have created a proprietary stack. While a generalist AI agent might be able to find the base of the pyramid, it will find it nearly impossible to replicate the peak autonomously. At eDO, we are a vertically integrated platform that moves further away from commoditization with every layer we add. We provide the one-stop shop reality. If you please could turn to Slide 30. Let's look at the strategic North Star of us versus AI platforms. Their objectives are fundamentally different from ours. These platforms are optimized for general intelligence and monetization. They want to be the horizontal layer of the entire internet. Their business models rely on being asset-light and liability-free. When you look at their monetization strategies, they are focused on subscription fees for their models or high-margin advertising. They're building to provide the answer to a user's query because the scales infinitely with zero incremental cost. They're not building to provide the fulfillment, which requires 24/7 support, merchant accountability, regulatory bonding and so, so, so many more things. Furthermore, AI platforms need to capture ad budgets, and travel is one of the world's largest digital ad categories. Their primary incentive is to remain the top of the funnel gateway, not the bottom of the funnel travel agent. As an OTA already operating a massive scale and paid acquisition, we are their large and necessary customer. We provide the transactional reality that makes their search engine AI systems valuable. This isn't just theory. The industry leaders are telling all of us this directly. Look at Slide 31, the quotes from Google, OpenAI, important financial institutions, all confirm discovery versus the fulfillment divide. When Google discusses their travel options or when OpenAI talks about agents, they focus on reasoning and personalization. They talk about helping a user decide where to go. They notably stay silent when it comes to travel on the plumbing, the details, the millions of little things in edge cases that need to happen in order for it to be fulfilled for the traveler. As these quotes illustrate, their goal is to be the orchestrater of intent. They want to be the brilliant concierge, which they do extremely well, but a concierge still needs a hotel to check the guest into and a car service to actually drive the car. By their own admission, they are building the brain, while eDO has spent two decades building the body, the essential transaction backbone that turns their AI intent into a booked reality for the consumer. Thus, we aren't competing for the same dollar. We're providing the fulfillment. Please turn to Slide 32. To understand the current environment, it is helpful to look at the history of vertical travel integrations. A significant example is Book on Google, which was launched in 2015 to reduce mobile friction. The time line for this project highlights the immense operational complexity inherent in our industry, and it demonstrates that travel is not a simple digital good. It is a layered commercial system full of edge cases. For a horizontal player, the strategic value lies in providing high-quality leads, while the vertical-specific fulfillment and servicing is the focus of specialized retailers like eDO who have spent decades mastering that operational complexity. Turn to Slide 33, please. As we have underlined previously, global travel fulfillment is defined by profound operational complexity. Let's look, for example, at IATA licensing, which is a massive barrier for any AI entrant. To issue airline tickets directly and OTA needs IATA accreditation and must join the local billing and settlement plan, BSP or ARC in the U.S. in every market it operates in. This is not a single global license. It is granted country by country, and each market sets its own rules. Each accreditation usually requires a local legal entity, financial guarantees, typically bank guarantees or default insurance, audited financial criteria and ongoing compliance obligations. Accreditation in new country can take months and frequently human communication and negotiation between IATA and the distributor. There's only a few companies in the world with the level of IATA integration that eDO has. To match eDO's global content, an agent would need a vast network of multi-country IATA licensing. Even with accreditation in place, accessing competitive content requires a second layer of bilateral commercial agreements that cannot be automated into existence. For example, GDS contracts with Amadeus, Sabre, Travelport, multiyear volume-tiered agreements that determine both content access and unit economics. There's NDC agreements negotiated airline by airline, each with its own technical certification, commercial teams, revenue sharing models. There are over 100 [ NDC ] capable carriers and content quality and pricing depends on the depth of each individual relationship. Direct hotel chain contracts and connections to bed banks, wholesalers, where preferred rates and inventory are a function of volume already shipped, a chicken and egg problem for any new entrants. Also low-cost carrier integration, which sits outside the [ GDS NDC ] entirely and requires direct technical and commercial deals. And there are so many more localized, specialized financial and legal complexities that we simply don't have the time to discuss today. Please turn to Slide 34. Let's look at the financial reality facing AI agents. Despite the hype, AI companies are currently absorbing massive losses and the pressure to monetize is reaching a breaking point. Even a leader like Anthropic has reportedly had to throttle performance drop from 86% to 65% in mid-April to manage soaring compute costs. This proves that free intelligence is not a sustainable business model. Eventually, these platforms must follow the money. Open AI, for example, is reportedly targeting $100 billion in ad revenue by 2030. To hit those numbers, they don't need to be the retailer. They need the click-outs to retailers. They need a high-volume, high-spending merchant to pay for the traffic they generate. This plays directly into our core strength, performance marketing. Because these AI platforms won't be paid by the consumer for travel bookings, they must be paid by the merchant via advertising. Let me remind you why travel can be their North Star. It is one of the most lucrative ad categories on earth. In 2025 alone, the top four OTAs spent over $20 billion on marketing. AI platforms are competing to be the next gateway for our travel budgets. We aren't their target for full replacement. We are one of their most prized customers. If you could please turn to Slide 35. Let's wrap up this section and distill everything we've discussed into three takeaways. One, AI does not equal disintermediation. We are dismantling in the midst of disruption. eDO possesses proprietary differentiated products like our unique itineraries and advanced flexibility options. And we, of course, continue to build many more that simply cannot be commoditized away. Two, AI represents a new traffic channel. Far from a threat, we view agentic AI as an emerging acquisition funnel. The major AI platforms have started strategic reasons -- sorry, have stated strategic reasons as well as massive technical and legal hurdles that prevent them from wanting to own the actual booking. Instead, they are turning to advertising for monetization, which plays directly into our strength as a performance marketing leader. And three, our moat is built on fulfillment plus prime plus scale. Our resilience is anchored in three structural advantages; a, fulfillment. We handle the messy operational core of travel that platforms want to avoid; b, prime. Our subscription model provides a relationship-based LTV that allows us to compete in these new advertising-based environments; and c, scale, our technical pipework handles the immense volume and accuracy requirements that these AI platforms demand from their vertical partners. In summary, as AI brings more complexity to the surface of commerce and a specialized tech-forward intermediary becomes more valuable than ever. With this, let me hand it over to Carsten to talk to you about our vision looking forward and how we are embracing the new AI opportunities with our technological capabilities.

Thank you, Dana. In this next section, I will address how leisure travel is going to change in the context of AI and also how in an advertising-based agentic AI travel environment, the user journey of some customer segments may change with additional complementary future channels. Please turn to Slide 38. As you look at the landscape of the global travel market, it's clear that channel fragmentation is a permanent feature of this industry, not a temporary one. Today's market remains deeply split. Nearly 40% of the travelers still book direct and a surprising 36% remain even with offline agents. The OTA segment holds a strong 25% share, but what's most telling is the sentiment behind these choices. Recent peer surveys confirm a massive cross gap. 66% of the customers explicitly prefer a specialized travel company over an AI platform and 68% simply do not trust AI to handle the financial and logistical act of booking on their behalf. However, we are not complacent. While we are currently thriving on the high-intent traffic through various channels, we do recognize that agentic AI is introducing new, highly sophisticated entry points into this ecosystem. Leisure travel will change. The search phase will become more conversational and more agent-driven. But as the complexity of the interface increases, the value of the trusted fulfillment partner on the other end of that agent becomes even more critical. If you could please turn to Slide 39, let's discuss exactly how eDO is positioned to be the primary beneficiary of this evolution. We see a future where the existing travel funnel is being augmented by new high velocity channels. As we transition into an advertising-based agentic environment, we see two additional possible future pathways. The first one is new AI native ad formats. We expect the rise of AI native clickouts. Imagine a world where a user describes their dream trip to Gemini or to ChatGPT and the AI doesn't just give them a link, it presents a fully orchestrated eDO itinerary. With one tap, the user is passed directly into our high conversion environment to finalize the deal. This is performance marketing 2.0, and we are perfectly positioned to lead in it. In fact, we are already experimenting on this. The second is the seamless agentic fulfillment. It's a more radical and less likely vision of the future because it translates into what we call an embedded agentic booking. This is a world where the AI would complete the entire transaction on behalf of the customer, leveraging our deep APIs. The user would never leave the AI interface, but eDO would always remain the merchant of record. We would continue to handle the complex plumbing, the pricing, the ticketing and the fulfillment entirely behind the scenes. In both of these scenarios, eDO's role in the ecosystem doesn't just remain, it truly solidifies. We are becoming the essential vertical engine that powers global travel. Whether the customer sees our brands or uses our pipes, eDO is the infrastructure that makes gigentic travel possible. We are the intel inside of the AI travel area. Please turn to Slide 40. Now let's look at how this evolution I have just described translates into clear scalable monetization. The first and most evident path is one we've already mastered. It's a bidding and CPC model. In this scenario, AI interfaces like Gemini or ChatGPT act as the new Google. They suggest the trip, and we bid for the click out that redirects the customer to eDO. We are already world-class at performance marketing at scale. So for us, this is simply a transition to a more efficient AI native ad channel. The second path has a very low probability. It's what we call the ageentic booking. Here, the AI platform would act as a digital subagent. They would complete the entire transaction via our APIs and potentially charge us a commission on the back end. This is still a highly efficient model for eDO because it essentially turns the AI giants into a global automated sales force for our proprietary inventory. The critical takeaway here is that while technology moves at light speed, human behavior does not and remains sticky. Just as off-line travel still commands 36% of the market decades after the internet arrived, different customer segments will embrace these AI journeys at different speeds. But on top, we have an added advantage. We are not betting on a single outcome. We are building the infrastructure to thrive in both. Whether it's an app-driven click or an API-driven booking, eDO remains the essential engine behind the transaction. We are ready for the transition, however fast the consumer chooses to move. To sum it up, by doubling down on our technical core, we have built a platform-agnostic powerhouse. We are positioned to win in every possible user journey, whether the customer chooses to click through to our interfaces or to settle the transition directly with an AI chat. We are essential engine for both. The transaction finishes with us. If you could please turn to Slide 42. Because the question is not just whether the funnel will change, but it's also about why will we be the winner in this new world. And our confidence here isn't just based on optimism. It's based on three structural advantages. First, our Prime LTV advantage. Our subscription model fundamentally changes the math of customer acquisition. While traditional OTAs must fight and pay for every individual transaction, we are focused on acquiring long-term subscribers. And because Prime members have a higher and more predictable LTV than a one-off traveler, we can approach the bid for high-intent AI traffic with much greater efficiency and precision. We have over 20 years of experience excelling in the world's most sophisticated performance marketing environment, and we know exactly how to turn a click into recurring relationship. This isn't about outspending competition. It's also about outmaneuvering them. We are building a high-value member base, while others are stuck chasing the next single transaction. The second point is around our proprietary products. From our unmatched global content, our excellent customer servicing, our unique multimodal itineraries to proprietary ancillaries, we provide a level of value that generic LLMs cannot replicate. Any AI agent is only as good as the option it can offer, but we manufacture and own travel products that don't exist elsewhere. And lastly, our high-performing platform and technical readiness. AI platforms like Gemini and OpenAI have zero tolerance for providers that do not meet the highest standards, be it on bookability, accuracy and latency. We share this, and we are ready. We are already meeting the strict industrial standards that these AI platforms demand. All in all, we are structurally set up to succeed in these new channels, and we have been leaders in performance marketing in hypercompetitive environments for the past two decades. When I look at AI, I actually see opportunities, first and foremost, opportunities to deliver new value propositions to create new travel features and to help our customers discover the world through Prime in more intelligent ways. Let me walk you through this in the next following slides. And if you can please turn to Slide 45. Because indeed, AI is not just about efficiencies, but it's rather about opening up for these new opportunities for us and for the company. And we are more than ready to capitalize on these opportunities through three core levers. The first is capabilities. We are leveraging a mature extensible platform that allows us to innovate with a level of agility that is really outstanding. All of that platform is in place, and it's ready to run advanced AI use cases. We can leverage years of investment and launch rapid innovation on top of that super quickly. Second, it's about results culture and mindset and execution. We are extremely results driven. Our delivery-first mindset has already shifted us from theory to practice, producing a massive new base of AI native features that actually add value. And I will show you examples of how AI is delivering value to our customers and us in the next moments. And lastly, it's about new opportunities. AI-powered distribution is opening up a new frontier for eDO. And our industry positioning, our partnerships, our product offering plus our infrastructure are all creating this huge boost to rapidly capture these emerging opportunities. If you can please go to Slide 46. Let's deep dive in our capabilities in our eDO technical platform. Because our technical foundation is a scalable cloud-native base, which is designed for rapid innovation. So what does it mean? First of all, we are cloud native. All of our platform have been running on the cloud for many years already. Second, we are standards-based. No need to reinvent the wheel. We leverage best-in-class technology and a technical architecture based upon leading standards. Our ecosystem manages over 500 microservices and more than 100 model context particle or MCP servers. And for those unfamiliar, MCP is an open standard that enables AI models, specifically large language models to securely and consistently connect to data and software. This robust infrastructure is what powers over 6 billion daily AI predictions across 247 apps and websites in 44 different countries. And lastly, we are AI and data first. Data is the oil of AI. We utilize a data mesh architecture that handles over 100 terabytes of daily data ingestion and powers rapid innovation on the AI side. This scale and sophistication has been highly recognized in the industry, for example, by leaders like Google. To show you how this technical maturity translates into customer experience, I want to share a video of some of our AI customer-driven features. And to set the stage for this demonstration, you will see real-world examples of how eDO has moved from theoretical AI to a live end-to-end customer journey, leveraging this high-performance engine to refine travel. And as mentioned before, we have been working on AI for over a decade. So this isn't just a prototype. You will see some of the over 100 AI features we have deployed across our business to improve conversion and customer satisfaction. So please roll the video. [Presentation]

Wow, I hope it gives you a good flavor of our AI-first approach across everything that we do. But let's please turn to Slide 48, and let me show you how we're using AI within eDO. First of all, our innovation capacity is currently skyrocketing. We have fundamentally redefined the role of the engineer by turning AI into a massive force multiplier for our technical talent. Our engineering productivity increased by more than 47% year-over-year. This isn't just a marginal improvement. It's a structural step change in how we operate. In terms of engineering velocity, in our most advanced development teams, 100% of all the new code is now AI generated. Of course, always under human command and design and always human verified. So how does this translate into business impact? We are seeing a fivefold increase in the number of business features delivered through agentic development. We have unlocked the ability to bring 5x more concepts to market. This is a remarkable achievement that allows us to scale our innovation without the traditional constraints of headcount. Let's move to Slide 49. And let me show you some examples on how these productivity gains translate into a massive tangible impact across the business. And an easily understandable example is in our customer service transformation as we move from traditional human support to a sophisticated AI orchestrated model. Here are three core examples of why we believe we are leading in this area. First, we have a high-impact AI-powered voice assistant. It's a system that resolves customer queries in 5 languages without any human interaction. Second, we built a proprietary agentic AI orchestration layer. This allows us to automate complex end-to-end journeys like flight changes and cancellations by letting the AI interact with our back-end systems and no hallucinations. And lastly, we have rolled out 24/7 chat resolution features. We deliver instant around-the-clock resolution for a wide array of customer needs. As you can see, the impact on our results is already very significant. We've achieved a 13% reduction in total service costs compared to the previous year. Nearly 1/3 of all the support calls are now solved entirely by AI with no human intervention required to resolve the issue. And most importantly, the AI [ CSAT ] levels are now comparable to human handled calls, proving that our customers value the instant results this technology provides. So AI is not a future promise at eDO. It's a live P&L positive reality, which has been underpinning all of what we can do. To show you this in action, we want to share with you a customer experience with a behind-the-scenes view of how AI agents are handling these calls. And please roll the video. [Presentation]

You just saw an agentic AI voice agent listen and talk to customer and perform action at the back-end to price end-to-end resolution of incoming support calls. There are many parts of this moving puzzle, and we are extremely proud of what we have delivered already, and we're excited about what's ahead. So if you could please now turn to Slide 51, let's take a look at another example in pricing. Moving beyond service, our productivity gains are delivering a massive tangible impact on our pricing strategy. We have engineered a proprietary in-house AI-driven revenue management ecosystem that is truly industry-leading. Our pricing engine executes 4 billion daily pricing predictions. We leverage deep neural networks and real-time competitor data to optimize for long-term customer LTV rather than just single one-off transactions. And this is delivering a competitive edge. This ecosystem increased value creation from AI-based pricing by 24% between fiscal year '25 and fiscal year '26. Crucially, we achieved this while simultaneously improving our prime price competitiveness, ensuring our members always get the best value. It's a clear example of how our decade-long head start in AI allows us to outcompete the market through superior data intelligence and intelligent tooling that we have built on top. If you can please turn to Slide 52. Because to conclude on this section on our tangible AI results, I want to highlight how we are fundamentally transforming our marketing operations. By transforming our creative workflows, we are achieving levels of scale and efficiency that were previously impossible. We have fully transitioned our in-house creative agency to an AI-centered workflow. By implementing modular design systems, we've enabled high-quality AI-powered creative production at an industrial scale. We've achieved a 3x increase in [ Selic ] creators and a 30x increase in reader creative production. This enables us to fill the digital shelf with fresh relevant content across all of our global markets. And we have delivered this massive output while keeping our headcount stable and even slashing external production costs by 75%. This shift demonstrates exactly how AI allows eDO to produce more content at a higher quality and significantly lower cost, creating a sustainable competitive advantage in customer acquisition. Please turn to Slide 53. Because we do not view AI-powered distribution as a threat, rather, it's a large opening for eDO to expand our global reach. We are uniquely positioned to capture these opportunities through three strategic levers: Through our expanded reach, we can now expose our differentiated value-add products outside of our core funnel to a much wider audience than ever before. By challenging market participants, we are prepared to challenge established leaders. By surfacing our Prime offering, which consistently beats the competition, we can drive growth on these new AI-driven services. And through co-innovation, our strong partnerships allow us to jointly pioneer new frontiers such as our ongoing experiments with large partners on agentic commerce. eDO is perfectly positioned to rapidly experiment, launch and seize these emerging opportunities in travel landscape. And I will now hand it back to Dana for a recap of today's session and some closing remarks.

Thank you, Carsten. Everyone, please turn to Slide 55 for our closing remarks. To wrap up, let me revisit our unique moat and the immense opportunity ahead of us. As a specialized retailer, we own the deep operational fulfillment layer that AI platforms recognize as essential and structurally difficult to replicate. These platforms understand that the technical and regulatory complexities of travel retail requires a very specialized skill set. We have a clear strategic edge. We are leveraging this technological shift by integrating Prime and our proprietary products directly into these emerging AI channels. This brings huge new opportunities for eDO. Agentic AI as a new channel. We are transforming conversational AI into a high-performance engine for customer acquisition. Also enhanced experience. We are utilizing these models to fundamentally improve the customer journey and sharpen our unique product propositions. Also innovation velocity. Our internal capacity is skyrocketing, allowing us to develop and deploy business features faster than ever before. We are more excited about the future than ever before. We have put a strategy in place to successfully leverage this opportunity and deliver accelerated growth to the benefit of our shareholders and customers alike. Turn now to Slide 56. To conclude, by leading in the AI era, we are developing and delivering a fundamentally better and more resilient business. Our internal operational leverage is the engine behind this transformation. We've touched on the skyrocketing productivity, where we've achieved a greater than 47% productivity increase year-on-year. Also AI-generated code in our most advanced engineering teams, 100% of our code is now AI generated and always human verified. And accelerated innovation through our agentic development, we are delivering 5x more business features to the market. These efficiencies directly translate into superior financial and strategic outcomes as we look forward to FY '30. And we have said we will deliver higher growth in which we're targeting 15% to 20% CAGR for Prime members between FY '27 and FY '30. That's very strong growth. We're also generating higher customer LTV. We see a greater than 13% increase in lifetime value for Prime annual members through our new monthly and quarterly payment models. With stronger loyalty, we are driving a greater than 10% increase in customer loyalty and the increased diversification. We have said that by FY '30, 66% of eDO volume will be driven by non-flight products and flights outside of the top 5 European markets, and that's up from 43% in the first half of FY '26. To sum, we aren't just reacting to the AI era. We are leading it. The infrastructure is built, the results are tangible, and we are perfectly positioned for a new phase of profitable growth. With that, we would now like to take your questions in relation to this session. Please refrain from asking questions about the financial results and outlook because we will, of course, cover those during next week's FY '26 results presentation. We'll now answer the questions sent to us in writing in the webcast. We'll take them on a first come first serve basis while grouping similar topics together to cover as much ground as possible. If we do run out of time before we get to yours, our Investor Relations team will follow up with you directly after this session.

With that, let's turn to our first questions. The first one is from Guilherme Sampaio from Caixa Bank. Would the subscription model be communicated in this channel? Yes, absolutely. We would make the subscription model available and communicate the benefits of our subscriptions. And the LLMs would want to provide users the most complete amount of information. You can imagine also with the technology, the way in which things are going is that it allows for much more granular answers to customers' queries than if I can call it traditional search would do in the past. So absolutely. The second question from Guilherme is, could eDream's customer acquisition cost change? I'll take that again. Well, let me first put customer acquisition costs in the context of, look, you can't just look at customer acquisition costs in isolation. Really, customer acquisition costs are nothing more than the function of LTV to CAC because you have to run a profitable business. And that's really what drives acquisition cost then. So we've proven in most competitive environments, be it in SAM, in metasearch, et cetera, that our superior LTV, the one that we get from Prime and with our neat products, et cetera, and the proposition, that really does allow us to outperform most of our competitors. So when I think about in terms of the future about a native AI environment, this isn't any different in it. We would expect an LTV to CAC in the new agentic channels to be very similar to what we currently have in the highly, highly competitive channels such as Google and metasearch today. Let's see. Guilherme has a third question. Could economics differ in an agentic versus a click-out channel? Let me take that. I think the answer again is very similar to the previous one where we -- it really is a function of LTV to CAC, and we would expect an LTV to CAC and new agentic channels and AI-based click-out channels to be similar to what we currently have in the very high competitive channels that we see today and that we operate today, such as in Google and in metasearch. Guilherme has a fourth question. And so it is -- let's see, likely is a more structural agentic bias towards OTA versus supplier direct content considering the mentioned orchestration needs? So I would say we do believe there is many reasons why agentic AI would have a bias versus OTA versus going direct to one or a combination of travel providers as there are content advantages that we try to demonstrate in the presentation, right, where we facilitate things to a customer and make it far easier for the customer to do. They have a much better experience. We also showed how it's paying off in terms of our customer satisfaction, customer experience levels, and they're far superior. So again, on average, agentic AIs would likely bias towards those, the one-stop shopping as well. Also because of our Prime, we have -- prices will be lower in the context of that. So again, this plays very well to us, particularly as we think about in the new world with agentic AI being able to surface far more information than what it did in the past. That absolutely does play towards us. Those are the questions from Guillerme. Next set of questions are from Carlos Trevino from Santander. The first one is, could agentic AI new channels have a lower conversion rate into Prime members than traditional channels? At the end, consumers are looking for a booking, not for a subscription. I would say we don't see that. Well, but it is, to be fair, on low scales. But I don't think it would be hypothesized about that. Consumers do subscribe to Prime based on how much they would benefit from Prime in their travel. And so while they are looking for booking, they are, at the end of the day, making in a sense, a calculation on either benefits on a huge portion of our customers are through actually word of mouth signing up to it and also the price savings that they would see right on the spot for it as well. And so in an agentic world, the expectations would be very similar. In fact, it might be slightly biased towards us to be seen because agentic AI does allow for a much easier surfacing of additional information than under a traditional search model point of view. Carlos has a second question. So how much are you reducing the time to market of new functionalities at using AI in the development? As an example, how can you reduce the time to market of all the necessities to launch Prime in a new country using AI? So Carsten, why don't you take that one?

Yes, sure. First of all, the first answer I have to give is it kind of depends because the time savings we're seeing really depend also on the nature of the task. If it's very heavy on engineering, we see very, very meaningful reductions. And if it's more heavy on contracting and external dependencies, there's different time lines on this. So we are at the beginning of this journey on this. I believe there's immense opportunity, but it's going to be a very meaningful reduction in timings that we are expecting, and we're seeing in the initial results.

Very good. With that -- I'm just looking and I think that is the last question that we have. So with that, then let me just thank everyone for joining this AI update today. Hopefully, this was informative for you. But before we leave, I do want to remind you that on Thursday, the 28th of May, we'll be hosting our conference call in which we'll discuss in detail our FY '26 results. Obviously, in the meantime, if there's any other queries that you do have upon reflection of this, of course, our IR team is here, and you can reach them through the e-mail address, [email protected]. With that, I want to thank you for your attention.