#433: Marco Argenti of Goldman Sachs

So first of all, thank you for having me here. It's been an interesting journey. I've been in technology pretty much all my life. Believe it or not, I started to write code when I was 13, which is now over 40 years ago, and things have definitely changed. So I spent the last, before coming to Goldman, I spent the last six, six and a half years at AWS, where I was really driving some of the innovative areas. Launched, for example, serverless Lambda, the messaging, a lot of the internal things, product. And at the end, I was really engaged in what we used to call digital transformation, especially in the context of how internet of things and how some of the new emerging technologies like edge compute will transform the way some of the industry will do their business. And so when I started to realize that, you know, my primary focus was shifting from talking to the CIOs to actually talking to the CEOs of those companies, for example, automotive companies, I started to realize that technology really was starting to have a seat at the strategic table, right? Was starting to be top of mind for CEOs, was starting to be top of mind for boards. And so it's kind of when I decided that I wanted to be part of that transformation rather than from a vendor standpoint, to actually be within a company that is going through that transformation. And so that's kind of what led me to, you know, I was looking at, let's say, what would be an industry that will kind of drive that sort of transformation. And you know, financial services is a fully digital industry. It's fast, it's not constrained by, for example, physics. You know, you don't have to bend metal, you don't have to build, you know, airplanes or things that are large and complex from a physical standpoint. And so I wasn't anticipating what was coming, which is kind of this AI revolution that we're all living right now, but I felt that there was something there. And so that's what kind of made me do the move.

Well, right now you cannot escape the topic of AI no matter where you are, so that kind of is a little bit dominating the conversation at the moment. But in general, especially David Solomon was always an advocate of putting technology at the center of transformation and of the strategic agenda. And in particular, the idea of empowering developers and empowering people that are into technology, not only to improve the way we use technology internally, but also how could we offer technology externally to other developers, which is kind of what led to what we used to call the Financial Cloud, the externalization of our services, and then which culminated with the creation of our Platform Solutions unit. And so today, if I look at, you know, you have these two kind of opposing forces that are top of mind pretty much for every CEO, especially in our industry, which is on one side you have an increasing regulatory activity and regulatory pressure, which really is looking to put safeguards in place. And the other one you have the opportunity of AI that is kind of pushing to really, really rapid pace of innovation. And how you balance the two, and how you do it safely, how you actually navigate that line between innovation, safety, compliance, is actually one of the biggest challenges that every CEO and every CIO and every board needs to think about today.

Yeah, I mean, everybody's coming up with their own, you know, hyperbolic metaphor in a way. So you hear people saying it's like the internet, and no, it's not. It's like fire, it's like the wheel. So I picked the printing press, but not randomly. I mean, so there are two ways to think about AI today, and they're not mutually exclusive. You can think of it as a sort of a sustaining technology, which essentially makes your business more productive. And you can think of that as a disruptive technology, especially on the area of knowledge, and that will actually make your business more competitive. Now, productivity is great, but it's not enough. I always say, you know, you can get fit as a human body, but that doesn't necessarily make you a champion, okay? Right now it's a necessary but not sufficient condition, and that's like efficiency. So you can be very efficient, you can still not win in the market. And so I think it's the revolution of knowledge what AI brings that I think is transformative. And let me actually explain that a little bit more and why the printing press. So the printing press created the conditions for scalability of knowledge. So removed the barrier of physical access to knowledge. Before, in order to know, you know, like if you wanted to know math, maybe you needed to know a mathematician and have access to his manuscripts or hear his words. The printing press eliminated the constraint of physical access to knowledge and led to the creation of libraries, universities, and, you know, obviously schools and education as we know it today. Still, a very important barrier exists, which is the accessibility of content from an understanding standpoint, right? So if you have a very complex book that is kind of, you know, written for a mathematician but maybe still contains a lot of concepts that you as a business person, for example, want to access, or a technology book, you would have to even ask someone to translate it for you in simpler terms or in different terms, or you would have to study a lot. So there is a barrier there. What we see with GPT and with AI is that it's almost like a book that explains itself. It's a book that actually explains itself based on how you are actually interacting with the book itself. And for the first time, the reader and the writer are at the same level.

So a lot of the knowledge in a company, and I think people that are listening to us today might relate to that, if you think about your own company, where is knowledge stored? And definitely the answer is not databases, is not documents, right? A lot of knowledge is tribal and it's kind of in the heads of people. And then when you join a new company, the very first thing you need to do is finding someone that knows about a certain subject. So you create this network and it sometimes takes years. Imagine a new employee joining Goldman or joining any other, you know, company large and small, the time that it takes to master knowledge, the time that it takes to have full productivity is generally very long. So what if you could codify the knowledge of a company into a model that you could query and they will give you relevant answers the same way as the most or the biggest expert in that company would give you? And so I think one of the things that I see coming, I mean, is that every company at one point is going to actually want to create those models that are highly personalized, that are really distilling and really like codifying the knowledge there is within the company itself that now is not written anywhere in a way that is interactive to people. And I think that could be the biggest productivity boost that I've probably seen in my lifetime.

And even this superhuman kind of idea, it might sound theoretical, but then you can think about the following: what would a, and then you can put your own percentage, what would a boost of productivity of X% in a particular person with regards to his knowledge would yield? And so let's start with developers. Developers is kind of the area where we started also. We started to, you know, do proof of concepts and experimentation on products that will automate or actually like suggest code that then developers would review and then put in their code, okay? So a boost of developer productivity, you can easily see at least a 10 to 30% boost of productivity. So a superhuman developer could be 30, 40% more productive. If you map it to the typical IT cost of an organization, especially in our field, that, you know, very quickly can add to hundreds of millions of dollars a year, which then you can choose whether you want to realize it as a velocity increase, so you do more and faster, or you can have a cost saving. But that's kind of the parameter you can think about. Superhumanizing your top people, what would that happen if they could be, you know, 10, 20% more efficient in terms of the clients, the companies that they cover, the clients that they cover, the strategies that they come up with? And then you can kind of price their amplification, and that will give you a little bit of an idea of the return of investment. And that in turn will allow you to prioritize where to invest. And that's kind of a type of exercise that we're going through right now. Because I tell you, one of the things, in a moment like, I've been through a few of these revolutions. So, you know, I've seen the internet revolution, I've seen the app revolution, I've seen the cloud revolution, the mobile revolution. And in all cases, the two kind of factors that are so important for anybody to make decisions are, A, enabling people to experiment, because you really cannot plan what the success is going to be. There's just too much variable right now. But the second one is also to actually make bold decisions, you need to have some form of intuition to say, okay, I cannot do everything, but I will choose this and maybe something else, and then you really focus on those. And so that kind of intuition also comes from experimentation. And I think this is a moment where, you know, every company and every CEO and every CIO needs to go through that mental model. And I think this idea of who to superhumanize to get the highest yield, I think is a question that I think will be interesting, you know, for everybody to reflect on.

Well, so broadly, I think, you know, we're looking at three categories. One is obviously on the impact on your developer productivity, like I mentioned, on your IT spend in general, on the fact that today actually there are products that are off the shelves that will immediately make your developers being able to really shorten the time that it takes to develop code, but also to test code, and also to create, you know, the test cases for code, etc. So that part is extremely interesting and definitely something that I think everybody should focus on, especially because developers today are kind of ahead of the curve and they almost demand that. So it will be a question of talent at some point. You attract talent if you give them the opportunity to work with the latest tools. Then we're looking at the broad area of knowledge digitization. And it starts from, for example, document classification. Every one of us receives, you know, hundreds of thousands, if not millions of documents, which are in the form of, for example, contracts. You know, for example, think of derivative contracts or think about loan documents, etc., etc. Those need to be classified and then you need to do what's called the entity extraction. So you need to actually extract, for example, covenants and terms and conditions and make them readable by a machine. And it turns out that AI is extremely good at that, and generative AI can actually take it to the next level. And so the whole aspect of going from document management and document classification, entity extraction, and then knowledge extraction, you're looking at what are the most valuable sources of data within your company, where are some areas where you could train an AI to start reasoning interactively about that data. So that part I think is definitely a very important one. And then lastly, you know, we're looking at also automation. One of the things, one of the emerging characteristics of large models is that they are really good at figuring out step A after step B and actually being very creative at creating workflows. And I think that is also a huge area of impact in a lot of companies like ours, where, you know, we have extremely complex front-to-back workflows. And thinking of an orchestrator, next-generation solution for workflows front-to-back, I think is something that could be extremely disruptive.

So when I joined the company, actually the first few days, literally, Goldman did an investor day where we talked, in fact, it was Stephanie Cohen and I on stage, we talked about the idea of externalizing technology for the first time. We started to talk about developers as our clients. It was never done before. Developers were never clients, especially of a bank. And that meant taking some of the technology that we had inside, that we've been using sometimes for years and sometimes was completely natively created, to serve other developers in other financial institutions or other corporates that wanted to offer financial services. And then being able to really heavily simplify that process and allowing things like an incredibly highly personalized credit card experience, like, you know, you guys are very well aware of that, or an incredibly efficient corporate sort of a checking account in the form of transaction banking. So we took a very sort of, you know, kind of bold approach of creating cloud-native products to be extremely developer-focused. We created developer.gs.com, which is our developer portal, where developers could find well-documented APIs, where they could find getting started guides, etc., kind of things that are generally not associated with the way a bank operates, are more like thinking about a technology company, especially at our size. And so we started in this journey that led to us actually starting to offer solutions that have obviously high finance content, but they're also characterized by extreme customizability and extreme developer friendliness, which led to products like the Apple Card savings, TxB, like I mentioned, and also Marquee, which is really our digital storefront for institutions. We recently launched a product called Visual Structuring, which is a fully mobile product to do essentially structuring of derivative products. So I'm quite excited about that because in a way it serves a dual purpose. It pushes our developers to actually use a certain approach, which is you externalize, but also you treat your internal developers as clients, and that changes the game internally to your organization. Whenever, you know, it's something that I kind of learn from Amazon, that if you build something with the externalization in mind, even if it's internal, most likely you're going to make your internal developers much happier. And so it's interesting how this shift of philosophy within Goldman on how we actually operate our own technology translated into, you know, the benefit of being able to offer those products externally. I think that's a really good synergy.

Thank you so much. I appreciate it.