 Sam from the audio team said Phil Collins was my walk-on music, and I did not believe him. But here we are. It is a great honor to be here. It's very exciting. And I'd like to just start by thanking the organizers. We had a lot to deal with this week, so can we get a round of applause for all the volunteers? So my name is Brendan Reynolds, and I lead a couple of our design teams at JPMorgan Chase. I believe I'm supposed to have speaker notes, but I can wing it, if not. So I'm going to start by introducing JPMorgan Chase. I do work for a bank, so I'm going to do what we do and throw a whole bunch of numbers at you. But some of the things that I'd like to talk about are kind of a macro context of what we're seeing with AI in certain strategies and the effect that that's having on JPMorgan Chase as a global entity. And then from that, we can get into the role of design and designers, and we'll get into some aspects of our design system, which we call the Manhattan Design System, which has been shortened to MDS. I know that's creative. So for starters, JPMorgan Chase is the world's largest bank by market capitalization and the fifth largest bank by assets under management. So our roughly 300,000 global employees manage just over $3.5 trillion in assets. We have our technology platforms on any given day move about $10 trillion, and our technology spend or our technology investments in 2023 to manage all of that activity are a little north of $15 billion and new this year we've started publicly disclosing or talking about an objective for the financial implications of our AI strategies on the bank. So our current goal for 2023 is to realize $1.5 billion of net benefit that is measurable from about 300 different AI strategies throughout the bank. So happy to say at this point we're on track to meet that as we stand at the cusp of the fourth quarter. And then within the bank, our division is Chase, the consumer and community banking group. So this is your retail bank, your high street bank. This is the bank, part of the bank that's dealing with individuals, families, small businesses. This is the part of our bank that's most connected to kind of the day-to-day lives of people. And this is the part of our bank where design plays a very big role. So we are the largest deposit taker in the United States. We have roughly 80 million customers predominantly in North America in the United States. I always have to temper that number when I'm here in India. 80 million people may not seem like much here, but trust me, we're quite proud of that. It's about one in five U.S. households. And those customers are active. We see over 60 million active monthly users, so our monthly usership is high, and we see that across both the digital channels as well as our branch networks. So we have 4,700 physical locations that our customers also interact with, along with call centers. So a lot of the journey thinking that we do as a design organization is omnichannel in nature. And on an annual basis, those customers move about $5.6 trillion, which is a 2022 number. So that's everything from wire transfers for home purchases down to Zell payments or peer-to-peer payments for dog walkers. And our tech investment in CCB, or the Chase brand, for 2023 is $2.7 billion. So $2.7 billion is not our tech cost. That's not an all-in cost. That's just the net investment in new technologies and modernization, some of which I'll talk about in a bit. So PXT is basically, if you think about it, as the digital arm of the bank. So that is where our product strategists, designers, researchers, and engineers that are building digital experiences for our customers. There's about 15,000 of us. 850 of those are designers, researchers, and content folks. So that number is expected to grow to about 1,000 people by the end of the year. And those are collectively the teams that are focused on the way our customers experience over 100 financial products. As of now, only 25 of those people are here in India. So we have four sites, Pune, Mumbai, Hyderabad, and Bengaluru. And we're growing that team pretty significantly, because from a talent strategy, as we evolve towards full-stack teams, because of our engineering footprint in southern India, it's a foregone conclusion that those CX organizations are going to grow along with it. So this is a recruiting plug. We'll be doing a lot of hiring. We have a booth outside. There'll be more shameless plugs later. Okay, so why am I up here talking about the scale of our bank to a room full of designers? The reason is all of that activity and all of that customer behavior results in data. First-party, valid data that we can use to fuel better AI strategies. The important part about data is that even in the context of public LLMs, we still need first-person proprietary data to help the tuning and training processes to get those models to be more relevant to our customers' experiences. So our mission as a team is to help people make the most of their money so that they can make the most of their lives. So a lot of what we strive for is experiences that are simple, intuitive, and transparent, almost invisible in a way. So I'm going to talk a little bit about not only the role of AI in the bank, but I want to talk about where I personally see AI as a little bit of a moment. So especially to those of you who are earlier in your careers or those who want to pivot careers to position yourself in a way, in a company, in an organization that is poised to take better advantage of AI. So I do believe that AI is a rare opportunity. When the commercial internet was largely introduced, say, 25 years ago, since then we've seen all kinds of technologies emerge that are really exciting, very novel, and are things that capture a lot of intention. They're things that capture a lot of headlines. They're things that warrant a lot of venture capital. We see these technologies spinning up a startup ecosystem around them. We see these technologies driving mergers and acquisitions activity. Ooh, speaker's notes. Sorry, I was freestyling this whole time. Now they're gone. Never mind. Back to freestyling. So if you're anything like me, as these technologies emerge, you kind of geek out on them, you learn a lot, you read a lot, and again, if you're anything like me, you start to wonder, okay, great, but what do we do with this? And for some of these, I think they're more difficult or challenging to envision real-world applications or widely applicable applications. So we look at technologies like virtual reality, and we see that there are clear applications for things like entertainment. We look at augmented reality, and we see that there's probably some applications for things like industrial training or surgical support. If we look at something like blockchain and beyond crypto, you kind of wonder, it still feels like a problem in search of a solution or a solution in search of a problem, rather. Not that it won't eventually find one. Things like self-driving cars, like for those of you who are caught in massive traffic jams earlier this week, you have to wonder, will self-driving or autonomous vehicles make those problems better and how fast and how much? This is not to question the validity of any one of these technologies as a technology. All I'm saying is that it does take a little bit of vision to understand how these things might be applicable in a real-world scenario on a wide basis. Whereas AI... Sorry, wrong. AI is different in that it is a... Here we go. AI is different in that everybody in this room, everybody who has any grasp of technology can imagine all kinds of implications. The technology is transformative in that way. We don't require a particularly visionary leader to help us understand how AI is going to impact our work, our consumption behaviors, entertainment, pick an industry, you name it, it's pervasive. So you see this clear energy around how quickly can we start to implement these types of technologies in all kinds of different ways. And those ways are not limited to any one set of use cases, any one industry, or any one part of the world. They're pervasive. And then finally, these things are rapidly emerging. We have not seen technologies that have been capturing the attention of the general public this quickly. So with that capturing of the attention, they're also very well capitalized. So the investments in AI as a category of technology are not typical in what we see in other types of technologies. And the reason that's important is because that investment is ultimately going to accelerate the development of these technologies. And I think over the last 25 years, what we've seen is a more predictable or dependable or mature ecosystem of funding to start up to leadership. We have better product management skill sets. We have better engineering skill sets. So we're better positioned to take advantage of this type of technology more quickly. So for those of you who are looking at AI and wondering, hey, is this a substantial milestone in the development of the technology ecosystem overall, I believe that it is. And if you're a UX professional of any kind, if you're a designer who is looking to position yourself in a company who is poised to take advantage of this, I'm going to go through what I believe are four things to kind of look for. Also, shameless self-serving plug, these are things that are core to the JPMorgan Chase technology strategy overall. So first of all, we have data modernization. So any company that is a large incumbent with lots of legacy systems is going to have a challenge to modernize data to make it accessible for AI strategies. This is a challenging, fraught, expensive problem, but if you're at a company that is not putting in the effort and prioritizing this work, you're going to have challenges getting to full realization of AI. And this is a slide from our last investor day earlier this year in May where our Chief Information Officer sort of outlined the investment we are making in data in particular as a skill set, as a talent strategy within the bank. And you'll see 900 data scientists, 600 ML engineers, 200 AI researchers. So it's over 1,000 people that we have focused on what are the frameworks and processes that we need to implement to modernize our data, to position it to be consumed by AI strategies in the future. The second thing is public cloud migration. So without the compute power and the ubiquitous access to this data, any company is again going to struggle to take full advantage of AI strategies. So right now we're sitting at about 30% of the total firm-wide data is cloud-based. The 2023 target is 50%, which we're on track to meet. So at this point in a couple of years, we will be approaching 100% cloud-based data that data will be modernized. The third thing, and I think this is often overlooked because we tend to be enamored by the technology. We tend to focus on the technology. But we do require, both at a government level and at a corporate level, a set of privacy and protection policies that help guide us in decision-making towards the safe and responsible implementation of these technologies. So at Chase, or I should say JP Morgan Chase, we take this very seriously. And we have an interdisciplinary team, including ethicists, ethicists, data scientists, engineers, AI researchers, and risk and controls professionals who are assessing the risk of these technologies as they get implemented throughout the bank and helping us to define a better policy framework to prevent unintended misuse, to comply with regulations, and to promote trust within our customers and with our communities. So the last thing, and Andrew just touched on this quite a bit, is design systems. And Andrew stood up here and said, I'm not going to talk about AI, I'm going to talk about design systems. But I actually think he was talking about AI because mature design systems, tightly codified rules, guidelines, and patterns that govern customer experience is a base set of fundamental criteria that, again, can power AI-based strategies to drive better customer experiences. And this is one of our biggest tactical challenges at the bank with 1,000 designers. How do we get them to make consistent design decisions, both big and small? And how do we make sure that those designers are focused on the most strategic decisions? So, again, these types of activities represent a substantial investment on our part. So we're looking at north of $2.5 billion on an annual basis to drive specifically these types of outcomes and to lay the groundwork that will position us for better optimization of both AI strategies and other types of technology strategies. So how are we using AI today internally to make more consistent design decisions, to improve testing, and to accelerate delivery? Those are our three goals when we talk about using design systems at the bank. So, right now, this is a long list, but what we're starting to see is if you are a designer or an engineer or a content strategist or a researcher, you're starting to see what is largely ML-based, not generative AI tools, kind of come into play in your day-to-day work. So we're starting to see, and I don't think this is any particularly unique to us, but you're starting to see natural language processing, fuel things like chatbots, PII redaction, machine translations for foreign languages. We're starting to see data classification. We've seen sentiment analysis for years. We're starting to see ML used for things like entity recognition, and then with code generation, more and more co-pilot-like experiences where we have engineers and, frankly, designers, thanks to Figma plugins, who are starting to be able to get a sense of what that next best line of code is based on what other developers have already created. A lot of focus on documentation, which is the bane of existence of most engineers, so starting to automate a lot of documentation, debugging, testing, but automated testing is a huge focus, and then, finally, content creation, which I think is the one that is possibly the most misunderstood, where we get that sense of, what do designers do if they're not actually generating icons? Well, I don't think anybody got into this field because they want to stay up all night designing 25 gear icons to see which one tests better for a settings menu. So creating multivariants using technologies like Firefly or others to just help generate more options for multivariant testing within confined scenarios is a big part of what we expect to see and are starting to see things like different call to actions, different copy lines, different iconography, things like that. So our system, again, the Manhattan design system, is a catalog of over 300 coded components, and what we're looking at is, because it's been deployed for years, because it's so mature, because it has 300 components, how might we create a scenario where our 1,000-plus designers are making more consistent design decisions by knowing things like when you're laying out an address field, what are the typical ordering of components, how elsewhere in the bank, across seven lines of business, have other designers pulled this together? Is there a best practice? Are you using a component that hasn't been used in two years? Are you using components in ways that are not typical? So, again, not making design decisions for designers, but helping them understand more about the context in which they're designing. So, with Figma plugins and with the full suite of the Manhattan design system, which is not just the library of coded components, it is also a set of guidelines, it is a dock site that's fully supported, there is a team dedicated to nothing but training. Every designer that comes into the firm is trained on MDS. We have detailed release notes that are published to the organization. The system supports 900 designers, 1,000 product strategists, 3,000 UI engineers, and over 100 different product teams. So, we think about MDS not as just a toolkit for designers, but as something that requires organizational communication for consistent implementation. So, the investments in MDS are, again, laying the groundwork just like cloud migration, just like data modernization to allow us to create better or realize better AI strategies as those things come together. And one of the core principles, like many design systems, is because we have so many different lines of business and so many different content strategies and so many different types of customer experience. Modularity is a core design principle, so the whole system is really designed around consistency where it matters, but autonomy where you need it. So, if designers are working on travel-related content that has very lifestyle-oriented imagery versus service-oriented content that has very instructional-oriented content, those things can be autonomously defined by individual designers, by their teams, as needed. So, with this comes limits of errors, design performance. These things are more familiar. They're more widely shared. When designers do make changes or innovate on these modules, there's a requirement for that to come back into the system. There's a process for that. But I do think that design systems are a core part of an AI strategy going forward. And I'm going to wrap up and leave a couple minutes for questions, but I did just want to point out it is a pleasure to come to India. The firm has had a long commitment. We're going on about 25 years where we started with 75 people in Mumbai and we're now about 50,000 people across, again, Pune, Mumbai, Hyderabad, and Bangalore. Bangalore is, in fact, our largest employee community globally and we have a new state of the art campus that we're incredibly proud of. So, just saying. It's a great place to work. One more thing. My team assures me that that says thank you. But they're a little mischievous. They're hard to trust. So, if it doesn't say thank you, somebody has to let me know. Any questions? Yes. I think it takes a... Oh, you're in a minute. Hi, Brendan. Hi. My name is Manas. I'm a product designer at Cache Free Payments. You spoke about something which is personally very interesting to me, that is design systems. Okay. And in fact, I was having a chat with Andrew a couple of days back and we were talking about two approaches to building a design system. One is you design first and then you extract components, build a system. Other one is you start with building a design system and then you start to design. While it may be a completely different ball game for large organizations, what is your direct recommendations to startups completely, you know, building something from scratch? What is your direct recommendation to startups that are literally going from zero to one? Should they take the approach one or the second approach? Well, the answer is, I guess it depends, right? Because, you know, if design systems work well with scale, so if you're a startup that is probably not going to have more than a couple of designers, you sort of have less of this issue. But if you're planning to have hundreds of people in multiple locations in different product areas, then these are investments you probably want to start making. You know, I think part of the counterintuitive reality of design systems is, you know, they're in a way designed to get designers out of the job of designing. And what I mean by that is that even at a very, very large and complicated organization, you really only want a couple of people making decisions about corner radiuses, color choice, font sizes. Like, you don't want that democratized across thousands of people, right? So if you can kind of isolate those core design, UI design decisions and codify them into a system to be published out, chances are those investments are going to pay off. A couple more. Hello, my name is Danish. Something interesting that you mentioned towards the end of your slides was the process involved in pushing changes back into the design system. If you could explain a little bit more about the challenges there and somewhat what process JPMorgan Chase follows, that would help a lot for people building design systems here and the longevity of the system existing and evolving over time as well. Yeah, I mean, it is a challenge. It's a challenge for us as well. And some of the things that we do to help address that are we're very clear or we endeavor to create as much clarity as possible about how to bring those changes back. So whether that's specific forums, we have an ambassador program where there's a centralized MDS team that has an embedded ambassador in every product group. So the teams that are working through different strategies have a way to contribute back to the model overall. And they're incentivized to do that. They're recognized for doing that. We definitely have issues like any company does where sometimes it's just faster and cheaper to copy-paste the last component versus go through the process of contribution. So we push against that organizationally, but sometimes that happens. And sometimes, rarely, there are scenarios where what a designer or a team is working on is so incredibly unique that it really isn't going to be reused, so we don't have to make that investment. So there's a judgment call that needs to be made there. Designers tend to have a very optimistic notion of what is unique, so a lot of times we need to pull those things back in. Yes. We have questions all over. Hello, one question. I'm Nilesh Bharaskar. I work for DBS Banks in Singapore. What? Enterprise design team, basically. So my question is, like, when we have design system thought through, right, into a live documentation journey across the product development involved, the similar challenges what we also face, like, when we rely on the technologies like Angular, React, and so on, like, same system goes on and mobile experiences all across the platforms. So how you work with those version control, which comes around in the front-end technology, because some product are maybe in the legacy, Angular 14 or 13, I don't know the current version, but how you can adapt to that so quickly and what are the strategies around it, how you guys... Yeah, I mean, I think... I don't think I'm overstating it to say it is a best practice to avoid any technology dependency, but we have to have an abstracted headless system, right? You have multiple divisions of the banks, some of them were acquired, recently we have all kinds of... We have a very heterogeneous technology environment in general. So any UI systems we build have to exist on top of all kinds of different technology contexts. So where we have migrations from older systems that were rooted in specific technologies like Angular, moving to more modern technologies, often react, those things can be, you know, can be a manual process that has to get backlogged by the backlog of each individual product team and prioritized accordingly and that's up to that, right? We don't control that. So one of the things that we do is every once in a while, landscape mode, dark mode, any of these things that are going to kind of shine a light on who's not complying with things like the semantic tokens for colors. That's where we can really start to pinpoint who is out of compliance or not. We're starting to do better reporting on general compliance for this system but NetNet, we are trying to avoid any specific technology dependence because they're going to change. They're changing now.