Customer Ignite Talk: Antonio Bravo Acin (Global Head of AI Transformation, BBVA) & OpenAI

So next I want to share a fantastic example of a company that's going allin across all three of these pillars that Katie was just telling us about. BBVA are one of the largest banks in the world. Their goal wasn't to make AI an add-on. It was to infuse it throughout the entire business. Not just answering questions but actively supporting critical business functions. They started by rolling out JBT Enterprise to over 120,000 employees worldwide and they've really turned AI into an enterprise operating layer. By doing this, they expanded that vision into six specialized robots and two supporting pillars, each built to support a core part of the business. But rather than me tell you about the details of BBVA's ambitious AI strategy, I'd love for you to hear from the author himself. So, please join me in welcoming Antonio Bravo, chief data and AI officer from BBVA to the stage. Thank you very much Matt uh for the introduction and thanks everyone for being here today. I'm going to use uh just three slides to share a little bit of our journey so far. So bear with me. So first thing Matt already introduced that uh at BBBA we have defined a very top- down agenda to guide our AI strategy and interestingly enough is something that we've done alongside all the executive team and the different business areas and countries to make sure that it actually addresses the main opportunities that AI has for a that they brings sorry for a business like us for a business uh like a retail universal banking uh machine that operates in many different countries. So that's why we came up with an agenda that I'll elaborate in a minute that is basically wiring AI to the main pieces of impact across our value chain. You will see that this is an agenda that is not technological uh technologydriven or that is actually in a way uh only steered by the data and AI team. So what we have here is as Matt was saying six robots. Robot number one is the retail robot, retail business robot. This is same way in the past 10 to 15 years we've been investing a lot in bringing a new distribution uh model to our clients through basically mobile. We now want to create a new experience for our clients to interact with our products and services through digital channels, smartphones or any devices that might come into the future. And that's our robot one to my earlier comment before is co-sponsored by our head of retail and the different heads of retail in the businesses. Robot number two is how we transform the relationship with clients but in this case through the bankers in the segments where typically advisory plays a more relevant role. Which are those segments? Corporate and investment banking, enterprises, private banking in those segments just to give you an average of the time that a banker spends with the client is roughly 20 to 20%. So can we use AI to actually spend more time advising our clients? Can we make that 20 25 30 35 the impact that we can get is probably huge. That's robot number two. Again to my earlier comment on this being a bankwide agenda, robot number two is co- le by the head of corporate and investment banking and private banking and enterprises. Robot number three is in uh mainly anchored in the core engine of analytical capabilities in the bank which is risk. How do we bring AI capabilities to our risk risk analyst same way we do to the bankers. So as for them to do better analysis and also do it faster so that we can also streamline the process of risk underwriting to also make it more customer friendly. Robot number four is processes and back office operations document extraction classification of information in the back of offices where we manage all the paperwork for mortgages, consumer finance, insurance and all that. Number four, an area where uh AI is expected to have a huge impact for a big software development shop like us, software development. How do we bring codex to our software developers so that they can become more productive so that they can clean the backlog that we have? We have a huge backlog uh and uh and we always manage a scarcity and have to do prioritization of things. Can we get rid of that backlog? And lastly uh robot number six is what we call connected robot which is basically enabling AI to be used by everyone. Matt referred before we deployed 120,000 CHBD licenses to every other employee regardless of their role. And our goal also is to help them build agents that are general purpose to help them manage things like calendar uh email our HR tools and different tools that you you use at BBBA regardless of you being a risk officer or a banker or whatever. So that's robot number six and then two pillars. First we need our data to be ready to be uh consumed by agents. That's what we call becoming a data rich company. And the other pillar is managing across all these network of agents that we are starting to build. That also requires technology and orchestration across the company to be able to to also do the continuous improvement at the operation of the new organization of agents that we are deploying. Quick ideas also uh we we on top of having this top down agenda we we also want to be very iterative and make sure that we deliver value short term. So that's why we came up with this notion of aha moments where we've been progressively every two three months trying to come up with impact that is very tangible to the business or to the sponsors and that are realization moments that show the way and prove the impact and uh here we're also creating a little bit of a snowball effect uh in terms of adoption because everyone wants to chip in operating model I'll elaborate in a minute we basically are putting all our data u um teams and our AI factories to work with the businesses uh and with the teams that are actually living each one of the robots. We're putting a lot of focus also in experimentation and here partnerships I'll elaborate in a minute are very relevant and uh partnering with the ecosystem and also making sure that we have an stagegated investment approach in which we are progressively based on a moment funding teams with tokens and also funding teams with investment capabilities so as to create this uh rolling effect over time. I was mentioning before principles of the operation operator model. We have a very top-down agenda. This is relevant. But also through the wide adoption of AI, we also get a very nice bottom up impact that we then plug into these same priorities. So I think uh combining this very structural top-down approach with also the impact and the benefits of distributed creation that we get through enabling everyone to use AI is a great thing. governance and coordination again we are a large organization 120,000 employees I guess many for many of you is the same thing so wiring everyone in especially so that it doesn't become an AI agenda only is very relevant and that's why I was uh referring before to the fact that I try to have everyone all the sponsors in the robots engage into what we are doing and that's very relevant for the uh governance then multi-disiplinary teams uh we basically deploy all the capabilities is data scientists, machine learning engineers, our software developers to the teams in retail, commercial or risk so as to work together in building these agents kind of might sound soft but this collocation thing is very very relevant and actually uh flips a little bit uh the way we've been working previously in some of these efforts and then change management and cultural um and cultural change is very relevant in a way I mean if we build agents and then our employees don't use it because they scared uh or fear about what uh what they might uh imply we're not going to be successful. So we're also putting a lot of focus in change management training and also having a positive narrative that actually is true and we demonstrate to the employees so as to make sure that while we deliver the agents and while we deliver the tooling to the impact uh to the employees sorry they really embrace it and are not scared of them. Lastly, uh I was mentioning before connecting with the ecosystem. We've been uh very conscious throughout this journey that we operate in a super dynamic environment. As you've heard already from the team and as you will hear probably during today's session, things change almost every other two or three weeks. Uh and paradigms become obsolete very very fast. So that's why we've been very conscious on how relevant it is to be surrounded by the right partners so as to make sure that we operate in the state-of-the-art of technology and we don't uh work with absolute paradigms that then uh will will yield some stranded assets to the organization. So connecting to the ecosystem is very relevant because we have to do things the old way and in a such we don't have to do things in the old way sorry and in such a dynamic environment we cannot cope with the speed of being up to date. So that's why we have a very strong partnership with the open AI team in which we've been getting the support across the entire agenda of the aid across the entire agenda of robots and it's helping us a lot. Uh the amount of learning that we're getting is huge. We've done 180 degrees flips to some of the work that we were doing. So we actually made mistakes and the open AI team has been very helpful in helping us realize that uh we were not heading into the right direction before and we had to flip and I would say that my advice to any of you is that uh in this journey that is going to be so dynamic and so fast is essential to partner with the ecosystem and our experience in partnering with open AAI has been fantastic because they've been helping us drive the agenda of our robots in a very very impactful way And that's uh that's basically it. Uh those are the three slides uh that I brought and I'll be happy to take some questions. Yeah, thank you Antonius. Um so it's really fascinating to hear uh about the journey that you've been on. I'd love to understand like what are the key lessons that you've learned on this journey and how to drive AI adoption like what's been working well or what mistakes did you make for AI adoption? So uh meaning like the deployment of uh licenses to the to the teams, right? So I think the the first one I would say is that uh as as as it happens with everything in life, if you want uh things to happen to happen fast and to happen structurally, you need to put focus and uh a little bit of resources. And uh basically our AI adoption efforts which we now framed into robot 6 uh have been successful we believe so far because we have a team that is solely devoted in doing that. Uh we have a team that manages AI adoption across the organization both globally and in the countries and uh and basically they do a lot of training uh they do not only trainings that are online but they go literally it's a team that goes literally to every other team to the finance team in Mexico to the risk team in Peru to the uh um retail business team in Spain and they do trainings to show them what they can do with uh chip what they can do with AI and I think that's been very success successful because uh in a way um we've also experienced in the bank in previous occasions that uh when we went for instance to the cloud with collaborative tools there were pockets of the organization that were not uh working with them and they used to work in uh offline and that was uh tricky and in many cases when uh when you introduce new tools and this is a super powerful uh tool and context you need to accompany people you you you don't think it's going to work uh by only deploying in the product you need to accompany, you need to do the training, you need to track with metrics, which is also that something that we do, you need to fine-tune plans for this area of this unit that is lagging behind. And I think that's very su successful. And and another thing also I think uh when we when we've done this massive deployment of licenses to everyone uh there were also some fears of uh some of the colleagues saying what is people going to be doing with this? What are we going to get out of this? Maybe they start building automations that they doesn't make sense. So I think that's fine. Uh the our observation is that letting people uh use the tool and uh and and find out what they do with it is also very informative for your top down strategy. In a way it's like this story of planting grass before building roads. You plant the grass, see where the flows where people uh chooses to go and then you build the road in that uh in that in that path. No. And that's also been the case because uh as you know we've had some very impactful uh cases in which people has been building automations and GPS that are literally now used by thousands of employees and saving times uh by 70 to 80% in many cases we have more than 100 GPS saving 80% of the time to more than a thousand employees in many different countries and that has come thanks to enablement and wide adoption that again requires a team and focus that to help the organization move into that prediction and one of uh the most insightful things for me when I look at BBVA's adoption dashboards is that the global leadership team are some of the the biggest power users of the tool. Can you talk a little bit about how did that happen at the executive education? Yes, that that's it was a controversial at the beginning but uh but what we do is uh we send weekly uh monthly sorry monthly now monthly adoption dashboards to every leader in the organization and we will tell you uh this is your adoption level your personal adoption level this is the adoption level of your peers so that you can also benchmark yourselves uh and then the adoption level of uh your different units underneath so I think that that has been very useful because to my earlier comment it has helped us identify an area in Colombia or in Peru or in Mexico that was lagging behind. Why is that? Then maybe we put more focus on training and it also creates some nice pre pressure. Yeah. Uh we do that at the executive level. So the CEO and the chairman get the dashboard in which I am part of and my colleagues are part of and it's also important because uh if we as with anything in life no if uh if you do it and you use it and you're pioneer and you use codeex or you use chat GPT then your teams see you use it then they use it and then that scales down exactly and it's been great you know I've often get some really advanced questions from uh from some of the executive team at BBVA about how to use the tools. Um, so coming back to the robot strategy, Genai is so general purpose. It feels like you could solve almost any problem with it. Uh, but one thing that stands out to me from your strategy is there's some good focus here on the key areas of the business. So, how did you decide to build this strategy? How did you decide which areas of the business should deserve a robot uh to have that focus? Yeah, that's very relevant uh because um it's been a collaborative process across the executive team and uh I cannot stress that enough. I think that's very relevant because uh for a large organization like BBBA uh historically um tech uh tech things have been a little bit scary for the executives like uh you guys in tech you spend a lot and uh it's it's always increasing you know so that's why I think uh this is so different uh and this is basically technology that can enable a huge value and lock for what we do in BBBA and that's why we had this process in which we came out with this agenda that is uh very uh transversal as I was saying before and in which I basically was putting out the framework but then uh it was a collective decision on okay does it make sense to to use AI to invest in the transformation of our digital channels to make them more conversational multimodel and advisor focused and it was it was okay yeah we went together in the executive team to take that decision and robot by robot That was the agenda. We had to take trade-offs. Uh we don't have a robot for our finance team which is a great team and our CFO and we are bank come on. But we say okay let's stay focused. That's why the beauty of our agenda is that we call it the eight because it's only eight things, eight big things. But uh but to your question Matt, it was um it was a process in which I made sure I involved everyone and every on the executive committee on the executive committee of course the chairman and the CEO so that it's everyone's agenda. It's not my agenda. Of course, I'm the the the executive that manages and wires the agenda across and I spend 90% of my time with the teams and uh that's my main focus but everyone knows that is the bank's agenda. It's not my agenda and as you as you have seen it's not techy things that we are doing of course there's uh software development robot and some others but is anchored to the main areas of impact for uh for BBBA. It makes total sense. This is all moving really fast. We've been working for several months now and our teams have been working really well together to deliver the robots agenda. Uh but what's happening next? What are you excited for? How are you going to evolve your strategy as the technology continues to evolve? Yeah. Yeah. That's um I mean uh again evolving so fast. No, but I think what uh what I'm excited now about is uh as we've been building our first agents under under each one of these robots, it's also being clear that there's a pattern in a way. There's a way of of building agents. there's uh there's common primitives, the evaluations, the access to the data, the connectors to the APIs. So um as as we figured out that that there's a way there's there's a playbook to be to build agents that we have been finding through the building of our first agents underneath each one of the robots. We have the opportunity to systematize and industrialize the way we do it because honestly our organization is not fit uh or is not designed to onboard agents which is what we've been doing and there's complexities on the work that we've been doing with you guys. We've encounter complexities in terms of governance, control, security that of course we have to manage but we have to manage in a different way that we typically done uh because our process has never been operated by by agents so far. No. So what I'm more excited about is about how do we come up with this uh notion of industrialization of uh our agentic uh building capabilities so as to scale beyond what we are doing here to the rest of the processes and teams in the bank in a very structural and controllable way and that's a little bit how the future uh look like looks like I think and that's how we will be able to capture all the abundance that uh this AI era is bringing. Yeah, it's been amazing learning together. Uh, and we're excited to kind of collaborate with you on that next step. Thank you so much, Antonio.