Banco Bilbao Vizcaya Argentaria, S.A. (BBVA) Earnings Call Transcript & Summary

Good morning, everyone, and welcome to the second edition of BBVA Strategic Talks. Apologies, we got started with some delay due to the traffic. It's been particularly bad today in Madrid. But in any case, it's a pleasure to have you all with us today here, both in Madrid and virtually. Thank you very much for joining us. As you can see in the screen, we have a great agenda for today. We will start with one of our key strategic priorities. It is artificial intelligence, data and technology. For this session, I will be joined by Antonio Bravo, Global Head of AA Transformation and by Carlos Casas, Global Head of Engineering. Then we will turn to Spain, one of our main markets where Peio Belausteguigoitia, Country Head of BBVA Spain will share with us how we are bringing this strategy to life in one of our core franchises. The session will have 2 parts. First, a presentation from our speakers and then a live Q&A. We very much encourage you to participate and make the most of this session. With that, it is my pleasure to hand over to our Chair, Carlos Torres Vila, for our initial remarks. Thank you very much, and Carlos, the floor is yours.

Thank you, Patricia. And welcome to our second edition of BBVA Strategic Talks. Welcome, and thank you very much for joining. And apologies again for the slight delay due to traffic. The sessions as Patricia just said, are designed to give you a deeper view of BBVA's strategic priorities, which, as a way of reminder, you have here on the screen to embed into everything we do, a radical client perspective, to leverage sustainability for growth, scaling up the enterprise segment, promoting a value and capital creation mindset throughout the entire organization, unlocking the potential of AI and technology and then finally, a team embedded with empathy to succeed. In the first edition of the BBVA Strategic Talks, we cover the Enterprise segment. We also cover the Mexico business. Today, we will, as you know, talk about AI and technology as well as the Spanish business. Now let me start with the simple conviction which is that AI gives us an extraordinary opportunity to create value for our clients, for our customers and through that, for our shareholders. Capabilities, as we are all seeing are advancing extremely fast at breakneck speed, AI agents can increasingly reason, understand context, execute complex tasks. And with its improvement, we see how this expands what we can do. It's how we serve customers, how we operate, how we manage risks, how we build code. So across the bank, it expands what we can do. Because banking is particularly well suited to this transformation. Financial decisions depend heavily on context. So your circumstances, your goals, your needs, the needs of your business. And then finally a team [indiscernible] Managing your finances often takes done with less time and less effort. So therefore, our ambition is a bank that understands each customer, anticipates their needs and helps them act. Our teams complete processes with less administrative work, including fully automated workflows managed by AI agents. This all means more capacity, more capacity to serve customers to develop new services and to grow with a more productive operating model. Of course, we should also be clear about the competitive implications that this will have. AI will make it easier, no doubt for customers to compare alternatives. It will raise expectations. It will intensify competition. It will put pressure on margins. So part of the value of the increased productivity will go to customers through better service through more competitive offers. And increasingly, the better players like BBVA will gain disproportionate share and economics. So we intend to lead that process, use AI to deliver greater customer value, deepen relationships and win more business while improving our cost to serve. That is how we aim to turn this transformation into sustainable growth and attractive returns. Now the critical question is execution. There's already a substantial gap between what the technology can do, which grows ever faster and what most organizations have deployed. So a successful pilot is one thing, but integrating AI into real processes, connecting it to data, to systems in a secure, reliable way and operating it, all of it reliably at scale is much, much harder. And this is where we believe BBVA is particularly well positioned. First, we have strong foundations. We have lots of data. We have engineering talent, and we have millions of digital customers who trust us and with whom we have developed deep customer relationships, strong foundations. Second, we have demonstrated that we can do this, that we can transform successfully, leading the industry. We moved early and decisively in digital, as you well know, we changed dramatically how we work internally. We turned that transformation into a competitive advantage, which lasts as we can see in our performance metrics. And third, we're making AI a priority across the whole organization with the same conviction that drove our digital transformation. That conviction is not just some theoretical abstract concept. It's something that drives and it translates -- sorry, into concrete choices, how we organize, where we invest, the talent we dedicate, the partnerships we build. It means being willing to change priorities and redesign how the bank works. We're obviously focusing on the areas with the greatest potential first, there is potential impact, driving adoption across our teams and building capabilities to scale. You will see some examples today, including early results from our priority initiatives. Antonio will cover them. He will also explain the frame which is the common foundation we're building to create, deploy and manage AI agents across BBVA with security and control and accountability built in. So this means defining what each agent can access, what each agent can do, when human approval is required and how we monitor performance, actually defining what performance is, what good is and then monitoring each and every execution, not only to ensure that it's within the bounds we want, but to keep improving. This matters, this frame, this scaling, this common ground matters because our ambition extends across the whole bank. We need to rethink processes, priorities, ways of working and to make successful solutions reusable across countries. So scaling is the name of the word. So we have use cases, but scaling through this common platform is the critical point. Our starting position is strong. We have now to earn the advantage through execution. After Antonio, Carlos will explain how our technology is evolving to support that ambition of delivering change faster with the resilience and security our customers expect. And finally, Peo will take you through Spain, which is well connected. Spain illustrates really well how sustained transformation can strengthen the franchise. It also shows the opportunity ahead, deeper customer relationships, more effective bankers and greater productivity, building on what we have already achieved. By the end of today, I hope you will have a clear view of the opportunity we see, the capabilities we're building and how we intend to turn them into value. We have done this before. We understand what it takes, and we're determined to lead again. Thank you, and welcome, and I turn it over to Antonio.

Thank you very much, Carlos. We will now start the first session focused on AI transformation and technology. Joining me today are Antonio Bravo, Carlos Casas and our CFO, Gonzalo Rodriguez. So without further delay, Antonio, the floor is yours.

Thank you very much, Patricia, and good morning, everyone. Today, I'm going to walk you through 3 sections of content. I will start by setting the context of these very transformational ways that we are living to then go deep into our strategy and the progress that we've made so far, and I will conclude with some final remarks. Going into the first section and talking about the opportunity that we see ahead of us. The first reflection is that if you look at what happens in an industry like ours, the fundamentals are very much reliant on data and analytical model. across the value chain in all customer journeys in different segments, everything has data and analytical models underneath customer onboarding, cross-selling, credit underwriting. The foundations of our business are very analytical driven. And that's why we believe that this new wave of transformation that comes with AI presents a huge opportunity for a business like ours. Because it can be very transformative in the way we manage and relate with clients. It can make our experiences more personalized, more tailored. It can also bring greater transparency for the clients. in terms of processes, it enables faster execution and processes, which will also [indiscernible] into a better experience of our clients. And it will also have a dramatic way in the way our employees work and in the productivity. We believe that AI brings abundance to all these 3 dimensions. It makes easy what typically was complex. It makes fast to what typically used to take a lot of time. It can parallelize efforts that used to be sequential, everything in banking can be rethink to AI for the better of our clients for the better of our shareholders for the veto employees. With this, what is it that we are doing and how is it that we are planning to win this transformation. Let me start with saying that we've already seen before they witnessed how technology can transform a business like ours. Through the digital transformation, BBVA has been able to combine the change in our grade trajectory with an increase of our profitability while also most importantly, doing it for the better of our clients as we see customer experience improving across all our markets. And as we step into this new transformation era of AI, we believe that our strengths and foundations are very solid. First, we have data ready for all the countries in a global way in the cloud, which is one of the core foundations of every AI-related efforts. We've also been able to establish very deep partnerships with the main ecosystem players which, in a context where technology evolves so fast, we believe is very relevant to make sure that BBVA always operates in the state-of-the-art of technology. And lastly, we created a data unit 10 years ago. Throughout that period of time, we've been building capabilities of core disciplines such as data scientists, data analysts, [indiscernible] engineers across all our countries. And those capabilities and those teams are now at the core of the execution of our strategy. So we have the data ready for all the countries. We have also 4 partnerships and resources from the ecosystem. And we have the teams and the organization ready to be pioneering in the execution of our strain. As already said by Carlos, AI is one of our core strategic priorities, obviously plan that we outlined last year. But more than that is present across all the others because it's very transversal. It's going to help us be more radical about the way we relate them we serve clients. it will serve us as we will see later on, to better advise clients also in the enterprise segments in sustainability and of course, to create value and transform the way our teams work. But not only we have AI as one of our core strategic priorities, taking all the others. But we've also recently rewired our organization on an operating model to make sure that we have our AI transformation unit well equipped and self-contained, so as to execute on the strategy that we will now go deep into. And what is it that we've been working on. We outlined in the strategic plan last year around our AI strategy, which we [indiscernible] . VA is basically a top-down agenda that is bandwidth touches some of the main pieces of our value chain and it has been the first strategic agenda on agents that we've been working on, as I was saying, for the last year. It's made of 6 robots. It's robot is a set of agents and [indiscernible] billers. The robots are structured in 3 blocks. One of them, which is -- has the goal to transform the way we serve clients both through a personal adviser on digital channels, that is multimodal and we serve retail clients and enterprise clients in a multinode way. It's what we call do and another one that is geared towards helping our clients spend more time with clients in advisory roads. On the processes front, we have a couple of robots, one of them with full focus on our risk processes, in particular, for the wholesale business, which again is one of our core strategic priorities and then also another robot focus in processes in the back office operations. And lastly, for our employees, we've been focusing specifically on software development for which AI brings a huge opportunity to transform our productivity as well as on providing AI for every other employee at BBVA regardless of the role of the unit that they are part of. And these 6 robots have been relying on 2 core pillars, one of them, which I've already referred, making sure that data and information, which is a core differentiation is ready for agents to be consumed and also that we have the technical capabilities not only to build, but to operate this set of agents. And what I wanted to do now is to bring this to life to all of you. And in that sense, we have prepared some demos, not of all the robots of 3 of them, the first 3 of them. to show and demonstrate how these experiences that are already either into production or will be into production in the next few weeks look like. And with this, I will start with the demo of robot #1. Before we go into it, what you are going to see is our personalized assistant for our retail clients through a mobile experience, and you're going to see 3 cases that we've selected, one of them in which blew advises clients and helps clients navigate their movements and enables them to track to their data and also get some advice. In the second one, we'll see how Blue allows our clients to be advised in the sales process for a consumer loan. And lastly, we will see a third example on how Blue can also support clients in the process of putting a claim of an insurance. So with this said, let's go with the first time and later on I introduced it to others. [Presentation]

Let's switch to English. Can you please help me check my latest expenses Available in any language and comes back with an answer across accounts and cards and any other financial products. What about almost came spending on grocery and now group then by merchants trying is having literally a conversation with the data -- how does this compare with my spending in July questions. only, and it will provide a comparison of 1 month versus the other? But to all my subtended transactions, what ones do you think that should double check and why? You can hear it -- it's a very open question. Blue will go. Look, into the spend patterns of these clients and analyze some high amount, duplicates or unusual merchants and we'll provide the 3 bullet answer according to the client request. Okay. That's into thinking about buying a new car and need some help. Second case biting and consumer loans. I'm talking at the Volvo spec, I went to the dealer the other day and was close to an estimate price of EUR 36,000 SP-9 So we provide a couple of options. We approved loan and also an EV loan I have EUR 8,000 in savings. With the preapproved loan enough SP-10 Blue be a good option. And the client ask to go ahead with the full amount 30,000 who provides the options and the pricing. What would the monthly payments look like over 8 years instead of 10 years? Now the client wants to see a different term. And it's pricing offer that and also offers a comparison. Do you want to compare the 8 years to 10 years lines go ahead with it. And with this, it drives you basically to the page where the client can just confirm and get the loan. So you've seen very conversational. It would be easier to speak in English. And now we go with the third [indiscernible] Great. Let me say what happens and see if you can help today after doing my grocery shopping. I got home and I broke the gas on my induction cooktop while putting the bank. The big thing that could happen to one of our clients. And first, we ask on what is it that happened? Is it very crack? What type of [indiscernible] has happened. As I mentioned before, recalls the answer of the previous question. is not going on. semantical correctly. Some questions about it and offers a couple of questions. We can repair it or either receive the compensation. I prefer to receive the money and by new ceramic come. Yes. are correct. And with this, the plan is already goes into processes and it's all the demand for the clients. So you've seen this is some examples on how -- the new experience is that we will enable clients to be served with through our mobile app, our web, and eventually through any other device that might come to the market in the future. will enable our clients to have a conversation with the data, with the services that we provide in a very dynamic way, same way to use generative AI tools. And that will apply to both information about your day-to-day banking, also advice to get products such as consumer loans or to solve complex things like claims that typically also create a lot of burden in different cases. So this is a product that, by the way, we are going to be putting a new version of into production over the next few weeks for all our employees and then progressively through the remaining of the year and starting of the next year, we will roll it out to clients in Spain, Mexico and also in countries in South America and in Latin America. So that was the first demo on Blue. I'll later on touch a little bit on the metrics. Let's go now with the second demo, which is the robot #2 A banker. We are going to be focusing in particular in AI Banker for corporate and investment banking, again, is one of our highest growing segments, but we are also developing such type of agents to support our commercial SMEs, bankers and of course, our retail bankers. However, in the benefit of time, today, we decided to show you the demo of the AA banker for our corporate and investment banking business. So let's go with this second demo, please. [Presentation]

So that was an example on how we are enabling our bankers to talk also to different information that we have about clients. It was an example of internal information. We also are working on a set of agents that will couple that internal information with external information. Also will allow bankers to create features and generate ideas for those features. In essence, as you see, and as we will discuss, we have a road map to continue to bring these type of capabilities to make easier the day-to-day of our bankers, again, not only in corporate and investment banking, as you've seen, but also across the rest of the segments. And so we've seen a demo of the robot #1, digital experiences with Blue in the robot #2 for our AA banker capabilities for [indiscernible] and we'll go now with the third demo on the processes front, in particular, for the credit and the program -- financial program generation on the risk front. And let's go with the demo in this case, I'll be explaining what happens in this screen. Everything is integrated into the workflow analysts can select the client, the language that they want the financial program to be built in and then they can select the sector in which the client operates and the context information that wants to be -- the analysts want to be part of the financial program. With this, the agent starts putting together a financial program. It's a very long process, so we will typically can take up to 30 minutes. Just to give you a reference, this typically will take a human 25 hours in different days. Here, the ADM takes on time and produces this output in which, first, it provides a confidence score on each one of the fields based on the ground [indiscernible] that we've defined so that the banker can review all the different sections of the financial program. It also enables -- sorry, the analyst, it also enables the analyst, of course, to edit any of the fields that came out of the automatic PF. And also, it enables the analysts to check the sources. Typically, there's many sources that have been involved, public and private. And you can go always route to source of each one of the contents that were brought to the draft of the financial program and again, do [indiscernible] and adjustments accordingly, one shop there. And also we have the ability to establish others that might require a specific review by our analysts. And in that sense, those are flagged into the system or the risk analysts to review and [indiscernible] Again, they can be added. And with that, the financial program can be safe as a drop and the end into the workflows where it will be feeding our approval systems and limits for the specific client, which, in this case, might have been developed for. So this was the third demo. Again, in this case, showing how we can bring agents through our risk analysts to make their life easier and also shorten times compressed times in the workflows, which will also be producing nice improving our customer experience due to greater ability to execute faster. So with this, let me show you what are the first early results that we've gotten out of rolling out the products that you've seen. Across the client front, we've seen already that some of the intents that we are prioritizing in countries like Peru and Mexico and in this case, into production. [indiscernible] are able to absorb 90% of the calls that we get into the contact center end to end. We've also seen that the initial versions of up, not even the 1 that you've seen today, have been able to filter 50% of the conversations or the chat conversations that used to be managed by our relationship managers in this case in Spain. On the processing front, we've been able to reduce 30% in total, the time that we take to produce a financial program and also a rating of a client, and we are going to go up to 80% in this case and also on the processes front we've been able to shorten 80% of the time in the way we manage claims. Lastly, on the employee front, we have 75% of BBVA employees using AI recurrently with [indiscernible] savings of 2.4 hours. But beyond that, we believe that this is a very strong foundation as we envision a future in which our employees will be working alongside AI agents. The fact that we are being able to push adoption to these limits clearly prepares us well for what the future looks like. And also on the employee front, as Carlos will discuss and will present later today, we are already seeing increases of productivity across certain pieces of the software development life cycle of close to 30%. So what you've seen is already real. It's into production in some environments in different countries, and we now also see that there's a huge opportunity to scale all these efforts that you've seen. We have more than 15 million monthly calls in our contact centers. Almost 40,000 bankers at BBVA serving all segments that only, let's say, spend 50% of the time advising clients. We have more than 15 million operations managed annually at BBVA across all our businesses, too, and an installed capacity of 12,000 solar developers for which we have a huge opportunity to do a step change in the productivity. So we believe that scaling what you've seen beyond the early impacts that we've already reduced, will allow us to tap into a very nice opportunity as you see here. And precisely for that, we have a plan to continue to scale the agents that you've seen across countries and continue to give them also greater functionality. On the client front, as I already mentioned, we have a road map to progressively roll out Blue to all the countries. And by the way, not only for retail but also for enterprises. On the contact center, we are already into production in countries like Peru, and will soon be in production in Mexico. AA Banker for retail is already in production in Spain, in Mexico, in Peru, in Colombia and will soon be in the rest of the countries also with increased functionality for our bankers. And in the case of wholesale, the demo that you've seen, we are already going -- we've already developed this into enterprises segment and Corporate and Investment Banking, and we'll continue to enhance the experience as I was mentioning before, to pitch building, sectoral notes and visit presentations. On the processes front, our goal is to have 100% of our risk analysts in wholesale using tools such as financial program generation and also the rating by the end of the year. And from there, in 2027, we will expand an equivalent experience for our commercial banking bank analyst. And on the processing front, we are working in some core areas such as, as I was mentioning, claim management, recoveries, insurance claims, which you have also seen a demo in Blue and also in some core processes in our finance function such as accounting, which we believe presents also a great opportunity. And last, on the employee front, again, Carlos will elaborate on this one later to everything related to software development. But we're already developing and deploying agents for some core pieces of our value chain, such as design and testing. And for AI for all the employees, we are defining a set of agents that every other employee of BBVA, regardless of their role will be able to use and that are already also into production. But through the process of building this progress in which, again, we've been for the last year building agents under this framework that we call the putting them into production, registering and seeing some early signs of the impact that AI can have for our business, now called client processes and employees, as valuable as that has been the learning that came through the way in how to build an agent. Just it was something that we hadn't done before. And in that sense, we have touted the challenges of doing these first agents in a very [ artisanal ] way. We've been literally exploring uncharted territory for us. And again, as valuable as the progress that we've made and that we've shown today has been the learnings that we've had in basically identifying that there are common primitives that their patterns that every time you build [indiscernible] there's a process that you can systematize that you can industrialize. And that's what has inspired the next wave of our transformation, which is what we call the frame, [indiscernible] is to industrialize agent development so as to allow an enabled to capture the abundance of agents beyond this initial framework that we've used and take it to the rest of the bank. And this, again, is an effort that we've called the frame whose goal is industrializing aging development to bring agents to everywhere in the bank. And with 3 goals: we want to build agents faster. This is very relevant because again, we believe that there is a huge competitive advantage for being first movers. We also want to build agents not only faster, but of course, in a controlled and cogent way in a homogeneous way and also preserving our sovereignty and optionality. We think that our sovereignty will rely on our data, our context our connectors, our systems, and we will be leveraging on the best-of-breed capabilities out there in the market at any point in time. So again, this is the next wave of our strategy which will serve us as a booster to bring the abundance of AI to all the organization well beyond the efforts that we've shown within the [indiscernible] and to bring to life what is it that we want to industrialize, so that you can also understand who it entails to build an agent. We've highlighted here some of the core technological components that we are in the process of industrializing. First one, is the tech environment in the same way that every time we onboard an employee, we provide that employee with the chair and the table. We have to provide agents with a technological environment to leave in Also, we have to provide them with access to data and data has to be ready, as we've seen during the demos with context and semantics so that agents can understand that data. Not only data, we have to give them access to, but also tools and systems. If we want agents to execute on certain actions, they need to have access to those tools. Of course, we need to provide agents with guardrails [indiscernible] are very relevant, be very clear and deterministic in things that we don't want to happen at any point in time during the execution of an action. Evaluation and monitoring is very relevant, showing the agent what good looks like, what is the ground truth, what's the standard of quality that will want the agent to convert to in every other execution and then observe how the execution converts or not to then adjust the agent. Governed assets and security is very relevant. We will talk later about how we believe that trust is going to be increasingly relevant in the agentic world. Resilience also, especially for those agents that are going to be serving clients live and value management, which is also very relevant, I guess, we have all seen the news coming on token [indiscernible] and token consumption. Here, we want to manage this agent by agent and making sure that token spend comes with value in every other development. So this is -- these are some of the components that we are in the process of industrializing to build agents. And again, we discovered that these components are necessary and that there's a process that you can see during the process of building the first agent within the framework of the [indiscernible]. Now we are going to scale this in a very better mine way across the entire organization. So wrapping it up, we started with the execution program of the [indiscernible], produce some initial promising results that we are in the process of scaling and also [indiscernible] get some very relevant learnings about what it entails to build agents that has inspired our -- the next piece of our strategy, which is the frame with this goal of industrializing agent development to bring it to all the processes in the bank and by doing that, being able to capture the abundance of AI for the better of our clients, our employees, our shareholders and also in society. We are going to be operating in a context where we are also conscious of the risk that we are going to be managing. It's already been mentioned that we will face tougher competition from new competitors that in the last few weeks many of them came in into the market, new dots by open AI are meant to be disintermediating client relationships, bringing different experiences to the market. We believe, again, that in that sense, the trust that we've built over the years will continue to be increasingly relevant to retain the relationship with the clients. Asset quality I've mentioned before that this is going to be a very big transformation in the financial sector. But actually, we believe that all the productive sectors are going to be facing similar challenges and also opportunities. And in that sense, we are going to carefully manage our portfolios, especially in wholesale, Same way we've been observing and managing transition risk to other technologies such as decarbonization. And lastly, cybersecurity and resilience. It's also been a very trendy topic over the past few months. Carlos will touch on this also later today, but we are preparing well to be able to respond to this new environment where AI and agents are clearly raising the bar. With this, I'll go to the final remarks. Again, we believe that this AI transformation that we are now living in presents a huge opportunity for us to transform our business and our operating model. And as we step into it, we believe that we have the trust that we've built over years of relationship with our clients that will continue to be better relevant in this context of agentic relationships. We also have the track record in having been successful in demonstrating that we are an execution machine of our priorities that was true in digital, and it's already been true today in our agentive transformation, as you've seen today. And also, we have the strategy to go big to capture the impact at scale through the frame and to make sure that we embrace the amendments of AI in a very structured and determine way. With this said Patricia, I will pass the over to Carlos.

Yes. Perfect. Thank you very much, Antonio.

Thank you, Antonio. And after the presentation on AI. Now we are going to be covering a broader perspective on how we are working with technology to ensure that we deliver on the strategic priorities that we have. The way we are going to be doing this is, first of all, we are going to be reflecting on what we are, what we've done, what are the sources of our core strategic assets that we have today and how this is a base and a foundation for the future. Then we will reflect on what it is coming next, as Antonio was saying, the world it is entering a new era, a very different scenario, and therefore, we need to get ready and to prepare our technology to tackle the challenges that we are going to be finding. And finally, we will close with some final insights and remarks and some conclusions. So starting with what we are today. Well, a very important foundation of BBVA is technology. And this is a core source of competitive advantage. We really believe that it is one of the engines that we need to deliver on our strategy. And I think that we have a very good track record of delivering on that. And we have several recognitions as a world prepared bank. In this particular dimension, some of them very recent. But I believe that the best way to show this is by looking to who we are today, you could argue that we are one of the largest fintechs. You can find in the world, our business is digital. And as you see today, a vast majority of the millions of customers that we acquired a single year are coming through digital channels. But this is not only client acquisition, most importantly, the way that we are developing our relationship with them most of our digital sales are going through digital channels today. So as I said today, we are very digital and obviously, this is as a consequence of everything that we have built in terms of our technology. Now the question is what do we find behind this, behind this strategy. And we believe that we have very clear core assets, core technological assets that are behind our competitive advantage. First of all, we have the best channels and experiences in the industry, number one. Number two, this is based on tech stack on a platform that we have been evolving over time, and it is a timely modern, and that is helping us to deliver what we need to do. Number three, we are a global bank, and that means that we have an advantage as we can create once and deliver in many parts of the world, and this is providing an advantage in terms of speed and in terms of efficiency. And finally, we put a lot of effort on resilience. As you can imagine, trust of our clients and customers is the most important asset that we have and therefore, technology has to be supporting that. Starting with the first and initial core asset that we have, which is our digital channels, we believe that at the end of the day, what technology has to do is to ensure that we have the capability to better serve our clients and customers. At the end of the day, technology for the sake of technology, doesn't make any sense. It has to be focused on anticipating the client needs and making sure that the experience that we deliver are the best. And that is true in every single segment, in retail, in commercial, in large corporates. If we take the example of our retail segment. And we take the example of our mobile app, you could argue that it is one of the best that you can find in the industry. And this is basically because we are trying to cover 2 dimensions. First of all, it has to be convenient. It has to make things easy for our clients and customers. And there are several features, several characteristics that we have to a solar, for example, the capacity that we have to enable very fast payments with our mobile application. But it's not only that, let us say, easier part, also very important to make sure that we are able to offer a very good advice. Meaning that with the usage of data, with the usage of our technological capabilities, we can offer the best financial advice, and we can make sure that the value added that we are providing to our customers is of note. And that can be done very easily with our mobile application. Antonio was talking before about how we are evolving that with our capabilities Saas. And this, as you can see in this page, there are some very strong numbers in terms of the rating that our mobile application has and also the strong client engagement that we are able to develop. Secondly, as I was mentioning before, it is not only the -- our capability to serve our customer well. It is also the technology stack that we have behind. As you know, we are a bank with [indiscernible] yet of history. That means that our technology has been evolving over time. And obviously, we have had to modernize our technological stack so we can offer a platform, but it is modern and that can serve the different needs that we have. In particular, when thinking about our [indiscernible] we have put a lot of effort in the last few years. To build a technological stack that it is next gen, but it is global, but it is model and that it is based on the core principles of mobile technological architectures. For example, making sure that it is cost based that it is layered a modular, therefore, that is composable. And with that, we can offer good and flexible exigencies for our clients and customers. We have a facts about how the development of this particular technology. It has been helping us to transform our bank. As you can see, we have grown through a pattern of modernization of our technology, we have been migrating a very significant path of our transactionality. And today, we see that 2 sets, [indiscernible] meaning 66% approximately of our online transactionality is going through this NexGen platform, which is very significant. A very significant majority of our transactionality has already been moved. And that has a very clear impact. On the one hand, we think that the cost base that we have today when running our technology is EUR 200 million per year lower than what it would have been in the case, we have not gone through this particular modernization. And that has a contribution as well. to overall cost-to-income ratio. It is one important component, not the only one of our efficiency, but we see that all this effort is bringing us to a cost to income, which is best-in-class. Obviously, this is an effort that it is ongoing. In the next section, when we think about the future, I will further comment on how we are making sure that we are continuing modernizing and transforming our technological platform. As I said before, a third core asset that we have, first one being the channels, the digital channels that are top in class our ability to offer the best solutions to clients and customers. Second, the next gen tech stack. Number 3 is how we leverage globality because when you see this. The fact that we are a global bank and the fact that we operate in different countries in the world, enables that we can have global technologies where we build once and we deploy everywhere. The tech stack I was referring to before, the next-gen test that it is global by definition by nature. And then there are elements that contribute to this. First of all, our global software development practice, our creation practices. We have a global pipeline. We have global source code repositories and therefore, whatever we develop is accessible and available for the different parts of the world where we operate. So we have one single [indiscernible] development practice and single software development tools, which is a great advantage to make sure that what we develop is available for everybody. And then there is another very, very good example, and Antonio was mentioning that before, which is what we call ADA, which is our beta platform which at the end of the day is our analytical data and AI global backbone. Basically, this is where we manage information. We manage information globally from one single platform. And this is very important. I mean, when you think about one global strategic asset and when we think about data, which is at the core of what we are doing today, and at the core of what is going to be needed in the future, in the role of [indiscernible] that we have such a platform is critical. And it is available everywhere with the only exception of those countries where we cannot use for regulatory reasons -- cloud technologies. And most importantly, we were able, and I think it is a very good example of speed and the usage of globality to be able to deliver faster. We develop the type of platform and the product in every single country in less than 1 year. So I think this is a good sign on how our global practices are allowing us to be faster and more effective when developing our technology and our capabilities. And the fourth core asset that we have is resilience. As I was mentioning in my introduction, when we think about what is the core strategic assets of a bank regarding this, the customer and client base that we have, and therefore, the trust that they have on us. So it is something that we need to protect, and we need to reinforce every single day. And at the core of our practices in technology, is resilience by design, meaning that it's not something that basically we put at the end of our process to make sure that something is seem to be stable or it's going to be available. or it's going to be protected. It is part of the initial design of everything that we do. And this is something we have been reinforcing lately to make sure that this is something that we cover about. And a couple these of the very strong results and metrics that we have that clearly show that we are -- that we are very resilient and getting better and better. When you see the number of incidents that take place in our technology, in our installation, you're seeing that it is half of what it used to be just a few years ago. And when we have an incident will recover also in half of the time we used to do. So this is the result of how we are implementing those resilience by design practices in the way that we operate technology. And then I would also highlight a clear fact when you see the availability of service of our channel tech mobile retail challenges. Probably the most important one that we have on the more visible one either in Spain, in Turkey, Mexico, we have availability levels that are top-notch in the ballpark of 99.9%, which is a very good number. So basically, with that, we have covered how we see technology. So as you see, #1, how we offer the best experiences with the [indiscernible] channels. Number two, how we run a technological aspect that we need to modernize to be able to run our services in the best way and to develop new value propositions that enable the future, #3, how we leverage globality to do things faster and to build one and in a very efficient way, a number for resilience, right? But when we look to the future, we see that Antonio was mentioning that in his presentation, we are entering in a very different world, and there are things that are going to be different going forward. And with regards to technology, you could argue that what we are seeing is a computing revolution. So just a few years ago, is sort of productive computing that we had was the classical computing based on deterministic algorithms, bids and allergic operations, as we know. Now we have generative AI, which is a different way of competing based on not numbers, but on language, language models. It is not based on complete certainty, but rather on patterns. It's a different way of understanding computing and opening new possibilities such as generating new content and replicating how a personal human would think. And then very soon, probably before the end of the decade, we are going to have a productive quantum computing, which is a different sort of competing based on a different way of thinking about the problem, right? It is exponential variables, and it is going to enable competing alternatives and probabilities that today with classical computing, we cannot do. So that's going to be a different paradigm. So this is what we are going to be encountering. And we reflect on 3 topics. First of all, is that technology is going to be even more important, number one. Number two, that this is going to come faster. And number three, that this is going to be complex and that the context of risk where we are is going to be different. We are going to be encountering bigger risks and more complex risks. So that is why in the 4 assets that we have, we need to think how we are going to be evolving those. When you think about best-in-class channels and experiences, Antonio was showing that before that what we are going to be relating to our trends and customer is going to be different. First of all, it's going to be agentic, meaning that we, as a bank, our channels are going to become agents somehow. And that the sort of conversation that we are going to have with our clients and customers, you saw that in Blue, [indiscernible] #1 is going to be very different. And not only we are going to be interacting with people. We will interact with other systems, other agents of other clients and customers or even for some partners. Number two, we were talking about our tech platform also in how we are modernizing that to enable for the future, but this is coming really fast. So we will have to accelerate on our modernization in different ways. And one important point is that we are going to face a huge increase in transactionality and that will imply that we will have to scale up our technology to be able to face the future that we have in front of us. Number three, we were showing how our scale and our creation practices offer a competitive advantage in terms of speed and efficiency, but the world that it is coming something really fast and we will have to be even faster compared to where we are today. And that's going to mean that we have to drastically change the way we do technology, the way we create solutions where we adapt our technology for the future, and that's going to be a significant shift . And finally, and I was mentioning that in the previous page, the context of risk is going to be very different. Therefore, we need to reinforce and secure our technology installation because we are going to be facing a different context. Let me briefly go by one by one. The third one, as I said, is that we will have to enable new ways of interacting with clients and customers. Antonio was showing how we want to be doing that in terms of how the experience will be in the future for our clients and customers. Hyper personalized, it will be very event-driven. So it will happen in a much faster and convenient way compared to what it is today. We will need to enable a lot of context, a lot of information. So what we do and the solutions we offer for our clients and customers are more complete, more relevant, more proactive more conversational, so somehow more personalized, right? Antonio was also saying that because of that, we are building the infrastructure that we need which is the frame. And let me just focus not to repeat what Antonio said before, on the technical capabilities that are behind and that will have to be part of the agented technological architecture that we will have to embed in our tech stack. So we will have to connect new channels, agents. We will have to make data even more available with a unified data foundation that we will need properly connected, properly contextualized. For example, with [indiscernible] layer. Antonio was mentioning that in [indiscernible] in order to have that capacity to be able to contextualize well the information that we are going to be using to our clients and customers, we are going to have technical components such as semantic lasers that are going to be helping us doing that. We will have to develop different run times, obviously. So the environment in which we compete when thinking about a -- and from the point of view of security and resilience elements such as [indiscernible] jobs, bars or the different security components that we will need. So basically, in order to be able to serve our clients and customers in a different way, our tech stack needs to be complemented, adopted, and we need to introduce new architectures, agent architectures to be able to do that. Then let me move to the modernization of our platform, as I was saying in this new context, the modernization of our technological platforms seem to be even more important. As I said before, we have gone through an effort of modernizing our platform in the last few years, which is why today, as I said before, we have 2 sets of our online transactionality, already happening in cloud, already happening in next-gen architectures. But we want to accelerate that. And one important message here is that the acceleration that we propose is not just accelerating for the sake of having a next-gen technology. It is basically practical and focused on ensuring that we develop the components that we need going forward in terms of what we want to build and also that we can scale what we have in a world of more transactionality that we can do this in an efficient way because technology, as you know, is expensive. And finally, that this is resilient. So with those principles, it's how we are evolving our technological stack, and we follow our modernization pattern. So as you can see, thanks to what we have done in the past. We've seen that with increase of transactionality in an annual basis, close to 30%. Our transactionality or processing costs are growing only 5%. So that means that with the use of the right technology and the right modernization, we can be pretty efficient in the way that we scale up, and we pretend to continue doing that in the future. And as a matter of fact, in the same fashion as I said before, that our -- that today we have of our transactionality in NextGen. By the end of this strategic cycle by the end of 2029, a very, very vast majority, 90% approximately is going to be next gen. We have a plan for that. We have a clear path for modernization. This is going to leave us in a situation in which our business rules, business logic, are going to be in modern technologies that is going to help us be much more flexible and adaptable. Therefore, that is going to mean that our legacy technologies, particularly the core banking that we have is going to be reduced to a deterministic system of record, which is very, very transactional. And then the -- an important part is that we can accelerate the transformation basically because AI is providing an advantage in this. It used to be multiyear, very complex projects or programs. With the use of AI, and we are partnering with top vendors such as Anthropic on this particular point, we are going to be able to accelerate this very fast because AI is very good, as you will see in a second improvement. And this is basically about coding and making the right modernization and transferring more roads from legacy to next year. [indiscernible] we were referring to have to do with speed, and we had to do with how we make sure that in a world where things are moving very fast, we are able to create faster. Antonio was mentioning that before, if you remember when he was talking about the 8 and the different mission-critical robots that we have, there is one about our creation model, robot #5 which is about software development. And I would argue that when you subdevelopment attack [indiscernible] , it is broader because it is the entire creation process. So basically, we are drastically transforming the way we are letting some more, creating solutions to make sure that this is fully agentic, number one. that it is very interactive as opposed to previous processes that were much more sequential. We are able to create prototypes and to try things much earlier in the process. And we are able to do this in a fashion with the different people, the different laws that need to cooperate and work together can do it in a different way, and they can do it all together from the very beginning, which means better quality, much faster, much more productive, better outcomes. So the way we're thinking about this with some domains, big domains that we are covering within our software development operation process. The third one is obviously about thinking about a unit, which is the design part where basically you take an idea and you finalize with the technical specifications that you really need and that part is very iterative with business representatives and technological representatives. And most importantly, with the agents behind as [indiscernible] are helping us do the process in a very different way. Then we have the construction itself which used to be complex because there are many different architectural solutions, integration with our systems and so on. We are identifying that part as well. So we can put that into production much faster. And then the final part is related to the operation of the solution itself. With this, we are, as I was saying, changing drastically the way that we are creating software and creating solutions. This is already in production. So what it means is that we have several programs, [indiscernible] programs in the group where we are already implementing this, and we have a clear path of scaling this up to all domains by the end of this year and to all projects by 2027. And as you will see, we will have the capacity to deliver more things because we will make a better use of the resources that we have. And we will go from projects just to take months, if not years, to weeks as we are seeing with the current output that we see from the projects that we are doing. What we are going to be doing next is to have a look to in particular demo. It's a bit technical because it is based on, as I was saying before, on the real project that we are doing right now using this new way of thinking, and we are going to be focusing on the first part, which is about behind, so meaning the idea generation and how we finalize with the use of agency in the entire process with technical specifications. So we can see the video. [Presentation]

So as you have seen, this is a very different way of building solutions and our expectations, given the very positive impact that we are seeing already in the project that we are conducting is that we are going to be improving a lot in terms of speed capacity to build more things faster, as I was saying. And therefore, this will continue to be a source of competitive advantage in a world where speed is going to be paramount. And the fourth element that -- for assets that we were referring to in this presentation has to do with resilience, I was saying before that it is the most important thing that we have relating our clients and customers trust, right? And as we are entering a new world new we need to prepare for that, right? Because the context of risk is going to be quite different to what it used to be. And there are many dimensions here, let me mention 3. The first one is that we are going to be running a new technology. We are going to be identifying our processes. So that means that we are going to be complementing our current systems with new ones that behave in a different way. So we need to rethink all the core principles of resilience that we have in our traditional technological stack, how that has to be or can be replicated in the new systems that we are going to have. So that is why building safe agentic systems, ensuring that we have the right components to manage real ability in terms of availability of service in terms of recovery, in terms of data protection and so on and so forth. -- is a core part. Again, Antonio was saying before, within our framework, the frame of the structural pieces that we need to scale up our agentic situation. This is part of that, number one. Number two, in the same fashion that AI is bringing superpowers to us so we can better serve our clients and customers. Unfortunately, those super power can be used the wrong people against us. And in that dimension, we can think about flat. We can think about the money laundering and of course, several parts, right? So that's a core part of what we do. As you know, this is a source of concern for the entire industry that we take extremely seriously. As you know, we have been working the different banks in the European Central Bank to make sure that these are the right plans, and we do. So we are fully committed to reinforce what we need to reinforce. So we are -- we'll be well-equipped and we are able to defend ourselves if necessary. And we have a plan and we are putting all the resources that are needed to be able to do that in a short period of time, talking about the following weeks following months for some pieces. And the third element is preparing for an explosion of transactionality and growing volume. So when we think about resilience and availability of service from very relevant point is that our infrastructure is ready to be able to do that. So we can scale exponential if needed to be able to run the compute that we need. So that's a core part. Again, I'm referring back to our modernization plans of our technology. All our journey cloud is obviously putting us in a very good position. And that's going to be part of our lines of work with regards to technology, and in particular, to our resilience efforts. To finalize, so some of the points that we have been discussing or debating today, the first one is that technology is a core asset for us. It's a core differentiating factor, had been in the past, as you've seen. Thanks to the use of technology, we believe that we are in a very good position today. We're an extremely digital bank. As I said before, you [indiscernible] one of the largest impacts in the world as our business basically is digital today, a very big part of that. Second, this is the consequence of a very good track record on managing technology as recognized in the industry. We have quite a practical approach in terms of not being pragmatic with technology because at the end of the day, what we need to have is proactive technology that it is able to serve what we need, but that in place, of course, a lot of modernization elements as needed. And what we have today is a tech stack that it is ready for the future, although we need to continue, obviously investing in that as a world continued changing. And as in the past, it has been very good and a core asset to be able to develop a competitive advantage. It will continue to be in the future. Number three, what is going to be different the pace accelerating. This is a very technological transformation, what we see with AI and some of the sorts of computing that are coming such as such as quantum computing. That is creating a lot of opportunities, but obviously, we will have to face a more complex world and continue holding what we have, and we are very conscious of that, but we are extremely committed as we have a clear strategy, clear commitments and we are ready to continue strengthening our technology to be able to support the growth and the success of the bank as needed in the future. And with that, I finalize this presentation.

Thank you very much, Antonio, and Carlos for your insightful bus. We will move now to the Q&A session. We will start with the questions in the room and then move to the questions from our online audience. [Operator Instructions] So let's start with the first question, Alvaro, please.

Alvaro de Tejada from Morgan Stanley. I have 2 questions, which I guess for both of you. And when we think about sort of AI sort of agent, we immediately think about margins and lately, especially after me about deposit sweeping technology and what can that do to the bank's business model. When you see -- when you're deploying the tools today, can you maybe talk us through your thoughts on sort of any obstacles from automatic sweeping beyond customer trust, how do you see that working out in the future? I mean from a technology perspective, is there any regulatory sort of impediments to automatic sleeping -- how do you see that affecting the broader deposit franchise of BBVA in the industry? And maybe the second question is with that in mind, how can you defend against this? And maybe on the cost leadership side, presumably, this technology will be available to everyone and it's about sort of bringing down the costs. Maybe you can give us some color. You mentioned the 2 [indiscernible] transactionality now to the cloud, you're going 90%. How does that change the cost income picture? I mean maybe not an exact number, but a flavor of the order of magnitude of some of these cost benefits why you changed the cost platform?

Yes, I'll take the first one. And Carlos, you take the second one, okay? So thank you, Alvaro. I've alluded to it, the fact that we are operating in a new competitive context and news, which we've also referenced to is a good example on how clients might be shifting the relationship channel to agents of third parties such as me. I guess it's something quite new that has been rolled out in the U.S., not yet into Europe. What we clearly see is that to your question, we will continue to raise the bar for trust, which is going to be basically the competition is going to be a competition for Trust -- on the client who is the client going to trust. It's going to be a third-party agent or a BBVA agent. So we believe that, yes, there will be competition. There will be margin pressure that will be coming through that new type of competition. And our plan is to continue to build trust and enhance our value proposition so that we not only retain the Trust, but that we hopefully increase it down the road. But clearly, the rise of a new type of competition in terms of agents for general purpose that will serve clients across many different things across industries is something that we need to tackle on that type of competition. And again, enhancing our efforts on Trust, the client experience is going to be one of the core areas of focus.

Is there any regulatory hurdles today, for example, from the AI agent go with the parts into VA accounts can shop around deposits.

As of today, there hasn't been any regulation. Yet again, these products have not -- in the case of us has not come yet to Europe. We'll see down the road, how regulation might evolve and eventually might be easier or not for agents to do such type of execution or money movement actions yet to be seen, as I said, early days still.

In terms of the productivity and efficiency, you're right. When we look to the future, and we see our modernization of our platform, one of the reasons why, not the only one, but one of the reasons why we are doing that is because they are more scalable and they are more productive. And as I said before, with the 2 sets of online transactionality that we are already running in cloud. Our cost base is clearly in the ballpark of EUR 200 million lower than what it would have been. We have not gone through that transformation. When we look into the future, and when we see going from, as you said, from 2 beds to vast majority, the ballpark of 90%, that is going to have an impact. And again, our projected cost base in the future is going to be in the work part of additional EUR 100 million lower than what it would be if we don't go through that particular path. Thing is that this is going to help us contain rising costs basically because our expectation is that transactionality is going to be much higher in the future. So therefore, as we are seeing, computing is something that is increasing. So the by unit, if I may say, the transaction cost is going to be lower. However, depending on how this plots and it's something that we will have to see in the future. We see that those costs are going to be rising as well. Having said all that, the fact that more and more, we are more efficient and productive when running our technology, that is one of the factors that it is helping in the projected cost to income that we have in our strategic fund that has been communicated to the market. Obviously, it is part of that as well.

[indiscernible] from Goldman Sachs. You mentioned quite a few times that you're expecting quite a big increase in transactionality and volumes. Could you just focus through what will drive that? Is it expansion in the new markets? Or do you expect existing customers to do more transactions? Or what's the gate driver?

Well, I would say that it's mainly the customers increasing the transactionality with us as they will have a new way of interacting with the bank with all this data, and also we will be more proactive and provided insights they will drive that increase in transactionality. It's not regarding expansion in our business outside our footprint currently.

And also Patricia is growing organically. As you know, we are focused on customer acquisition, and this will drive obviously more transactions.

Absolutely

And if I may complement, there are patterns that we have seen in the past, which is obviously every single year, we have an increase in transactionality per client, basically because we are creating better experiences for them. our digital channels are more useful, if I may say, and that is driving more usage. So that is happening, thanks to the fact that we have better things to offer, and therefore, more consumption from each new client customer. And when you think about the future, with AI, this is also do, right? At the end of the day, transactionality is not going to be only classical computing. It Is going to be as well agentic transactionality. And all the robots and all the agents that we are creating, of course, they are going to be generating technical transactionality as well in our infrastructure. So all of that compounded is what we see in the future that our transactionality seems to be higher also from usage of our clients on the technical design that we have.

Just wondering do you see like the [indiscernible] growing or do you see [indiscernible] getting a bigger share of supplies?

In going as was saying, if I will grow, we grow because we'll have more transactions with more agents making accessions. Our asset would be larger because we want to -- our commitment is to lead this transformation, so we'll have increasing market share. So it's a bit about growing the customer base. It's about building deeper relationships and having more interactions with our customers.

Yes. This is Luis Pena from [indiscernible] Asset Management. So thank you very much. It's quite impressive when the bank is doing. I have 2 questions that are related. So first one is how much is the AI budget for the group? How that compares to the overall debt budget? And then the second one is related because there's been comments from U.S. banks and also from European banks. Seeing that the cost of the AI is going to be significantly higher of what initially the institutions were budgeting. So the question is, how are you valuing the cost of AI going forward? It seems that the price of the token is going to come down somehow, but the models are going to consume a larger number of tokens. And in this regard, actually the largest bank in the world is growing the cost base as we speak at more than double digits and is trading at more than 3x tangible book value. So it seems that the market is paying for the institutions that spend massively and wisely. So I just want to sort of understand and take the view of the bank on how this AI cost is going to impact the bank in the short term because it seems that in the very short term, it's going to be an additional cost. -- it's not going to bring an overall cost reduction for the group. -- and how management, we have the Chairman and the CEO here, see basically sort of these costs going forward. It really the management is ready to make this a significant investment and spend more. Thank you.

Yes. I take and then I'll pass it to you detail figures on cost and the weight. So I think the first idea I want to mention is that every talk in consumption that we will have will come hand in hand with an assessment of the value that the corresponding agent will bring which we believe it's a very structured way to address our genetic transformation. So every time we put agents such as the one that you've seen today into production, it is true, they have a cost. Some processes such as the risk agents that we saw actually also required 1 of the best -- some of the best models that are more talking heavy. So every time we build agents, we are very conscious on the fact that they have a cost and that we put it -- we have to assess and we assess it hand-in-hand with the value that they will bring. So in that sense, you rate of taking consumption that we are going to happen that we are already having comes hand in hand with the impact that we want each 1 of the agents that we are developing to bring. And my point is that as we scale the spend in tokens, we are going to be doing that in a way that we ensure that we also bring the value alongside the spend.

Thank you very much, Antonio. We have total, you can provide some insights on our person investment and cost going forward?

Sure. So definitely, we will continue investment on technology. If you look total IT cash out. That year was EUR 4.2 billion, and this year going to be higher. You look at also at the weight of IT cost of overall cost base. 2019 was 21%, 25% was 27%. So definitely, we're investing in and we are committed to continue investing in [indiscernible] and this, in a way, is what has really helped us to achieve a leading position in cost to income. Second quarter achieved 37.8% cost/income ratio. And we are committed to a strategic retail plan of reaching 35%. So we believe it's compatible to invest in technology as we're investing and banking is becoming more and more and ecology business. So we will continue definitely investing in technology while keeping our cost ratio at the levels that we are committed on.

And on [indiscernible] within the figures that you've mentioned, the weight that of AI is obvious increase increasing

Do you have -- sorry do you have long-term contracts with your AI providers with a set price. So I mean the market right now basically is trying to understand how this cost is going to evolve. And I don't think that anyone really has a clear view on this. So we just want to know really, I mean, how are you setting your contracts with your providers? If there's certain cap above which you are not going to go in terms of total cost? I mean, how do you structure those contracts?

So yes, we definitely -- and I was talking before about the partnerships that we've established with some key players in the ecosystem. And as you've already mentioned, so we typically allocate some compute capacity. And that's what typically comes as part of the agreement. It doesn't make sense to do the contract at the talking cost because as he was saying, it continues to drop every now and then. So we save the capacity and with that capacity, then we are able consume tokens or credits of different models depending on the pricing that again drops and continues to evolve over time. Yes, we're sourcing contracts, some of them for terms of 2 to 3 years with some of our partners. And typically, we say compute for our agentic organization to run rather than cost per tokens that [indiscernible]

Actually, the answer from sorry, maybe [indiscernible] . I have a follow-up on Luis' question. as long as the cost may change within time. And do you want to have resilience on your processes? At some point, what do you want to keep human how -- I'm not going to tell you to ask how many are you going to get rid of. But at some point, you mentioned that in 5 years' time, whatever happens, energy costs are pretty high. taken cost goes up by 2x, and you've maybe substitute 1 person that at some point was kind of productive but that new token cost it isn't. So how are you managing this process of taking this process out from people in order to make this resilience for the long term. And then another thing that might be a bit sorry about that, is that on the -- on the cloud development model. You've taken quite a different approach than another [indiscernible] you started before that won't have done that. But you rely more on partnerships instead of the in-house development. So which one will provide in the long term better cost, we have more resilience against these attacks because at the end of the day, if you have that building all you might be certainly better productive, but probably you have less ability to react those rats. And how -- what's the one that could adapt more rapidly to the changes that we have here. And one -- first, sorry about that. That's for Antonio, because coming from when we can carbon neutrality of this process because BBVA has been a leader in terms of becoming carbon neutral. But the heavy expenses in terms of energy. So how can this be managed in the future?

Perhaps the first one for you...

I did 1 and 3 and Carlos take the second one. We love big questions, by the way. Thank you. So going into the first one, Well, actually, it's reflecting about what you were saying, our growth and our plans are typically constrained by the capacity that we have to execute on them. take any business in any country, typically have plans that will beyond the capacity that we have. The view that we are taking on this, as I was explaining at the beginning, is to embrace the abundance of resource to be able to do -- to clear the huge backlog that we have in all the countries to do also that in a faster way. And we believe that doing more with the teams and resources that we have. Yes, along the way, there might be roles that will be eventually changing like 10 years ago, we didn't happen in designer at BBB. Now we -- that's one of the core disciplines or capabilities. So yes, there might be some jobs that eventually might change. We'll onboard new profile, so we will reskill some of them. But also if we look backwards throughout the digital transformation, which, again, has yielded great results in terms of profitability and growth. We haven't seen a dramatic drop in the number of employees at BBVA. And I think that can be used as a reference. Going to your third question related to carbon neutrality, yes. we're seeing clearly how AI takes a lot of energy and also, in some cases, water consumption, though that's still being more recently well managed by all the hyperscalers and players operate in data centers. We are committed to our net CRO targets, and we will factor in the token consumption that we do as an organization as part of them. So same way we do with other carbon emissions related to our third parties providers for value chain. Carlos, you take the big one?

Yes, of course. Good question. Different dimensions to your question, let me try to cover those. First of all, when compared to others, I cannot comment on this strategy, some others. But what is in the market is that there are some places, particularly when you think about traditional banks that need to modernize a stack. Some are taking away that at the end of the day is basically becoming more efficient. So you see some technical optimization that what they look for is that they are going to be running in a cheaper infrastructure, which is good in itself. Our approach is much broader than that because we are not doing just [indiscernible] Basically, what we are doing is to rethink what our technology should be. We think our logic, we think our technological architecture, so it is ready for the future. And that implies that our modernization process goes through a different pattern. That is why when I was mentioning in the initial part of my presentation that our modernization looks not only to cloud, of course, it looks to cloud, but it looks to modularity leisure and composable architecture data being available and exposed as needed and so on and so forth. So that is part of the core modernization plan that we have, which is pretty much based on rethinking the technological architecture that we need going forward. And of course, we look for productivity and efficiency, and I was providing significant numbers in terms of cost avoidance in the past and what we expect for the future. But we also believe that our core objective is to make sure that we offer the best solutions to our core clients and services, the best experiences that we do it quickly and fast time to market and that which is resilient as well. So that is the approach that we have -- you were having another comment regarding how we interact with our partners, and that's a different -- it is true that we have very good strategic partnerships with those that can offer the best technology for us and for the best technology for our clients and customers. And we are proud of doing that because we believe that this is the way to make sure that we offer to our clients and customers the best the best available. But we always keep control and we keep ownership of the core strategic assets that we have. Our tech stack is ours. Our technological architecture is our what we do basically is to make sure that, that can work with best-in-breed solutions of some partners when needed. And then one final, you were having a comment also connected to what Antonio responded before. Regarding about Brazilians and in our war agents are going to be doing things how we did control and so on. I would like to think that at the end of the day, we are going to be improving a lot in the way that we are running the bank with agents. But we are accountable for that. So the way that we think in terms of how we build resilient, human in the loop and so on in with the core principle that we people. We are accountable for everything that is done and we are not going to be losing that.

[indiscernible] from Citi. I have 2 questions, please. And firstly, you have highlighted the importance of scale and investing in AI to successfully deploy the AI. In your view, does the AI increased the importance of scale in banking and also, if you could -- I guess maybe the larger banks are maybe relatively in a more favorable position than the smaller banks in this regard. And then my second question would be tech is evolving very fast. I know you provided some useful data points on 29. But if I look towards the next 5 years, for example, how do you see the competitive landscape? And where is the biggest source of competitive advantage? Is it the customers' trust? Is it the data? Is it the scale for the capabilities.

So I'll take the 2 and then you chip in. So the first one is, of course, you think that scale matters in this transformation as alluded to the fact that most of the agent efforts that we presented in demo today are being deployed globally, in our franchise in Spain, and [indiscernible] will also comment later on Mexico, South America. Clearly, scale matters in AI, big time, and we believe that is going to be also increasingly relevant. Then into the second question on competitive advantages. I think clearly, we've been talking about trust in the context where there's going to be new competitors also claiming or competing to get the trust of the client for general total things. We believe that we have a strength -- existing strength, which is the trust of our clients to manage and serve them with financial services that we can step up with the tools and the agents that we are going to be deploying, both in our enterprise segments and our retail segments. That's going to be, we believe, very, very relevant, increasingly relevant in the agentic revolution, the trust and the retention of the trust. And then as we are today, as where we are today, we believe that one of the core competitive advantages will be, as we were saying, the ability of an organization to structurally change their operating model so that agents become a new workforce managed by our employees. That requires some structural change to onboard agents into the organization as we've been putting some examples and it's very deep. But if done right, we believe will present a huge competitive advantage so it will allow us to transform our productive model, combine our existing workforce with agents that will allow us to raise our execution capabilities. And to do that at scale, you need a very determined and structured approach which is what we are going to be doing through the frame to really raise the challenge and capture the entire opportunity. So I think trust and then the industrialization capacity and determination and execution to make it happen at scale will be differentiating factors. Because then with that, by the way, to your first question, we'll be able to tap on the opportunity that size might bring to each one of us.

Just to add one comment, Antonio, on your comments. Definitely on the first 1 is matters, both on AI and technology given that you have a larger customer base, you can dilute those costs on a larger customer base. And this is why we think that we are in most -- in a more favorable position, given that we are -- most of our franchises, we are either #1 or #2 in retail banking. And this is obviously, to your second question, creating a competitor vantage together with trust, the skill that we already have in our retail franchises provide us with a very good competitive advantage.

Media for [indiscernible] from S&P Global. You've been talking a lot about this industrialization of agents now the agent economy, where we see agents transacting not only internally but also externally. So I'm curious to see in terms of like the governance of that infrastructure, whether do you think there is a business case for those reasons to be running on chain to be underpinned by smart contracts and [indiscernible] or other type of heparan to transact and to make that fully transparent and accountable engine? Or whether do you have another business case where that's a function in the future?

No, thank you for the question, me I mean, very emerging trends on the agent disintermediation side, the products that have been launched literally over the last few weeks. And yes, I agree with you that eventually we could have this new general agent services, they could have an intersection, as you were saying, with ledgers. And eventually, that might be an eval infrastructure underlying for agents to operate still have to say how things are going to be evolving. Again, we are in the early days of the rise of these new product offerings.

Miruna from Jefferies. So there is clearly a lot of opportunity from deploying AI into our day-to-day processes. But I also wanted to understand a bit better what are the risks and how we are striking a balance between deploying new opportunities, but not deploying them too fast as in to create risk in the business. So I was wondering if you could share with us maybe some of the more negative experiences that you had in your early AI deployment process. The problems that arise at how you address those? And then secondly, I was wondering to what extent you are already using AI into your credit decisions today and how you see the opportunity there?

Thank you, Miruna. So we've made, I would say, will lend a lot along the way. We've done a lot of mistakes in the development of the agents. Of course, we to make sure that before putting an agent into production for an employee for a client, those mistakes are removed. But certainly, we've learned a lot. And we've been precisely encapsulating all that learning into some of the core components that are very relevant within our deframe effort. I was mentioning before, to security guard rates evaluation turned out to be some of the most relevant assets for us to make sure that agent executions comply or converge to the quality standards to the golden truth that we define for each one of them. And interestingly enough, another one of the learnings that came through the way is that our agentic organization has greater observability than we thought, we can monitor and see how agents are executing and performing versus the standard that we've set for them. And based on that observation, we can adjust or fine-tune if there's any deviation on performance. But to your point, we've had a lot of learnings throughout this year. We've learned a lot. We've made mistakes and precisely those learnings are what we are going to encapsulate what we are encapsulating into the frame to make sure that as we scale our agentic organization, we're also managing the risk in a very structured way. And to go into the second part of your question related to risk, actually, risk underwriting for retail clients is one of the use cases of AI within the AI Act, defined as high risk alongside insurance and recruiting. And in that sense, we haven't done yet any efforts on that front. What you've seen basically is AI supporting analysts always with a human in the loop for our wholesale underwriting processes. And that's a little bit where we are today. And we will continue to explore the boundaries of the AI act to ensure that we make the most of technology while also complying with the existing regulation frameworks.

If I may comment quickly on that one. Something that's interesting also to highlight is that agents can be extremely resilient and secure even more than what we had before. The reason being that by design, you can embed into the agent, the security requirements that you need. In terms of resilience, for example, when we think about equity requirements, we are moving from processes in which you would have advance to make sure that our systems are secure to something that by design, by definition because it is part of the context, part of the instructions that you are giving to the agent is part of it. I think it's important because although we need to continue learning and make sure that, that is working absolutely well. I think that the promise of having a more secure environment is also by design because the way it is.

Alvaro from UBS. Just kind of for a follow-up question on Alvaro's first question. Do you think AI-driven pressure on margins might be higher in developed markets than in developing markets. And in this context, BBVA being a bit more projected than other entities?

I don't know if you want to add some...

No, I think the only thing I would add is, I mean, it goes back to the trust question. So trust is an issue developed markets and emerging markets. And definitely, the pace about this will evolve. It depends a lot on how customers trust about getting an agent moving in a actions on your behalf accountability or would happen if something happens to the customer or there is fraud or the contract of 1 product and also regulation and this both impact emerge and developed markets. What I would add also is the fact that, in any case, -- the most important thing, and this is part of our core strategy, both in Mexico, LatAm, South America and Spain is to become the primary bank for our customers. So the best way to defend ourselves against this threat. -- is when you are in the flow, the payment flow of our customers, you have the payroll, you have direct debits, if you have the bit card, credit card, the risk of the intermediation is much lower. Whereas if you compete just on pricing or deposit pricing, this risk is much higher.

[indiscernible] So 2 very quick questions. The first one is you mentioned there's an advantage of being a first mover, but there movers sometimes do mistakes and competitors from that, right? So how do you plan? Or how is it possible to track actually what other competitors are doing right, better than you guys and implement those things into your process. And the second question would be how is this all the tools that you are developing, how it is to implement those tools or those -- yes, those tools. into potentially inorganic growth. So when you develop all these tools, do you also keep a focus on how we see is those to implement? Or you just develop it, just to focus on organic and forget about potential M&A and how it is to implement those tools, or is it a case by case? So it will depend for some potential targets it would work for some you want? How do you think from this?

On the first move about banking. Perfect. Yes. No, you're totally right. And there was also a question before on the fact that when you're a pioneer and you move first, you can also make mistakes. We see it as learning process. And we believe that being an organization that is able to learn fast enough in this very fast-paced changing environment. It's also a core competitive advantage. Yes, sometimes it requires flexibility because, again, the context is very dynamic. But the learning that we get through the process of being pioneer, we believe is a huge competitive advantage because it actually turns back into a greater execution capability. And on the second question.

On the second question, I don't know if you want Carlos to give some insights on how we build our stack, but I would say that it's primarily focusing our current operations.