By coinnews 
0:05 spk_0
Welcome to a new episode of Opening Bid. I’m Yahoo Finance executive editor Brian Sai. Like I always say, it’s the podcast that will make you a smarter investor.Period. And of course, open bids sponsored by our friends at Vanguard. Real cool episode here. We’re gonna dive into all things AI. We’re gonna take a break from all that tariff and economic stuff we’ve been talking about here in the podcast and go back to drilling into AI because it remains very much the year of AI, new innovations, new companies doing cool things, or companies that have been around for a long time doing absolutely amazing things in AI. And to that end, I want to bring in uh Marco Argenti, Goldman Sachs, Chief information officer.Good to see you, Marco. Appreciate you coming down here for me. Of course. So let’s, before we get into what Goldman Sachs is doing on AI, you came from what? Amazon, right? I did.
0:50 spk_1
What did web services?
0:51 spk_0
What, what did you do? What did you do there?
0:53 spk_1
I was running, uh, a number of areas like, uh, serverless compute. We started that with Lambda. We, I was running uh internet of things. We actually started that. I was running all the mobile services. So at the end I was running maybe 2025 services overall. Yeah.
1:07 spk_0
What was that like?
1:08 spk_1
It was pretty amazing because, uh, I came in, uh, in 2013, which was, uh, almost like, uh, at the beginning of a mega cycle of cloud adoption. And then I left in 2019 and the company had grown, uh, like, you know, like 50 X or something like that. And, uh,And when I joined, we had something like 13 services. When I left, we had like 230, so it was one of those times where I saw a lot of uh a lot of transformation for sure.
1:35 spk_0
Before weget into what you’re working on at Goldman Sachs, what were those earliest days? I mean, we’re not too far away from them, but as someone that created technology that really didn’t even exist or put more rocket fuel on technology that a company like Amazon was working on. What what was it like those first couple of years you were there?
1:52 spk_1
It was uh uh exciting because we were uh uh uh really like learning at the same time, uh.How to overcome certain technology barriers, but also most importantly, really trying to understand what clients really wanted. And the one thing that we used to say was always to work backwards from the from the client. And so a lot of the time was really spent to try to understand the why and then kind of adapt the how to that. Uh, for example, is it about reducing cost, uh, or is it about, uh, inc increasing speed, or is it about increasing agility?And uh at the end I think uh what we kind of concluded was that uh uh yes, cost is a very important thing, but acquiring agility and speed and velocity was really what companies were goingfor.
2:42 spk_0
Can I give you any credit for getting my packages to my house even quicker? Anything, even a little bit, just a little bit,
2:47 spk_1
little bit? Well, I mean like AWS obviously powers a lot of the, of the Amazon, so maybe a little bit.
2:52 spk_0
OK, fair enough. All right, give this guy some credit, guys. Go hit me with those comments out there on social media. So now you’re at Goldman Sachs. What?What made you make that transition?
3:02 spk_1
So, um,Towards the end, uh, like in the last couple of years at, uh, at, at AWS, uh, one of the things that I was, uh, uh, running was the so-called Internet of things, which was about, uh, wiring up, uh, uh, like, uh, industries and companies with sensors and then having machine learning in order to optimize process. So I started to engage uh a lot uh with companies, uh, more of the strategic level, doing digital transformation. And the real question there was, how do you really transform, uh,How a company run and operates, uh, thanks to technology as an enabler. And, and, and, and so I started to develop an appetite, uh, of actually being, uh, on the other side of the company being disrupted rather than, uh, someone sort of a, you know, helping them, uh, by selling tools and services to disrupt. And so, when, when a call came, uh, and also like, you know, our CEO David Solomon was like, hey,We think that uhTechnology and finance needs to really step up to a strategic level because we will not be able to be goldman unless we are like leaders in technology, like 5 years down the road or, uh, or, or keep our leadership in technology. So, that was, to me, like, uh, almost like, uh, uh, irresistible because it was a, uh, different, uh, uh, field, and I really love to learn. Big challenge.And also I was feeling that there was a big big transformation I had, and so that’s kind of what what drove me there. So
4:35 spk_0
when I think Goldman Sachs, I storied financial institution, been around for for many years. How is AI changing the business? and maybe we could start in a few different ways. How is AI changing what an investment banker does and what initiatives are you putting into place? Yeah.
4:52 spk_1
So, um, right now, um, we’ve been in this journey for about, uh, 2 years, I would say. We’ve been in the journey of machine learning for, you know, over a decade, but specifically generalative AI. And we have people from all over, uh, the firm actually using, uh, yeah, I would say, I would estimate that at this moment, at least, uh, uh, you know, maybe close to 2/3 of the, of the organization in a way or another is kind of exposed to a, to an AI tool.And I think a good way to think about it is that uh you have uh uh sort of a three waves of adoption. One is more when you are using a uh AI mostly for uh making an existing process more efficient. An example beingDeveloping software or, you know, running call center, etc. those are like the low hanging fruit use cases. And of course, uh we kind of have them at scale, and I can tell you a little bit more, but we’re already seeing, uh, you know, uh, a pretty significant impact on that. Say, for example, a 20% increase in productivity for software developer, etc.Then you have kind of the wave too and you’re starting to integrate AI more deeply into the fabric of the organization, which for us means a.And AI now not only optimizes an existing pro process, but it gives you an opportunity to actually change the way you do things, right? So an analyst, for example, uh, uh, can fundamentally change, uh, uh, the way, for example, they analyze earning reports uh instead of, you know, spending maybe 80, 90% of their time going through pages and, and trying to extract signals.They outsource the extraction of signals from uh to an AI.
6:35 spk_0
So is it analyst AI analyst clones, is that what thisis?
6:40 spk_1
Well, I mean, that’s uh I, I, I, I think cloning is probably overrated.
6:44 spk_0
We tried
6:45 spk_1
it
6:45 spk_0
withsheep, right?
6:46 spk_1
Yeah, exactly. We don’t see too many clothes out there, but, um, uh, but I think it’s more like, you know, like having this concept of an AI coworker.Which is slightly different than the concept of co-pilot. A copilot is someone that essentially does the same thing that you do, and it kind of helps you with a specific task. A coworker or agent is someone that you can actually delegate to, and that’s slightly different. And then all of a sudden, everybody kind of becomes a manager. Even if you’re never managed anybody, even if you never managed the person.Almost inevitably you might end up actually managing an AI. Hold on,
7:27 spk_0
I’m, I’m a manager here at, at, at Yahoo Finance. So hold on, how do IHow would I manage 200 AI agents or in some cases I’m thinking back now to when I was talking to Salesforce co-founder Mark Benioff, he’s deploying agents inside of his operations. I mean, how would I manage thousands of agents and how I imagine you have to retrain a whole workforce.
7:48 spk_1
Yeah, I think it would be an interesting transformation and at the minimum, uh, um, you know, you can start uh from the tasks that you’re already doing it and then, uh, uh.Essentially just, you know, ask questions or, or, or uh or uh let’s say for example, hey, we have earning seasons please analyze uh the, the reports of the, of the companies that have uh uh um they have announced in the last week and then extract anything that has to do, for example, with uh uh you know, usage of uh of GPUs and then, uh, you know, you start basically kind of extracting that, uh, assigning it.And then think about it, the way you scale is that those agents in turn I can outsource to other agents. And so you have, you know, like the same way as your managers or managers. Like when you ask me how do I manage 200 agents.How do you manage 200 people? Well, you manage 200 people by actually maybe have 10 people that you directly manage, and then those 10 people are gonna manage 10 other people.
8:48 spk_0
How do you, how do you fire an agent though? I mean you just delete it?
8:51 spk_1
Well, I think that would be an interesting one. I’m actually one of my predictions was that in 20025 we might see the first layoffs of agents. So
9:01 spk_0
layoff of agents, but it’s actually just, just you shut them down,
9:05 spk_1
you shut them down. And, uh, and, uh, you know, or retrain them. You can actually put them back to the, to the drawing board and retrain them. What’s interesting is though, is that.At a minimum,Everyone that is never even managed the person is gonna have to develop three basic management skills, which are gonna become, I think, uh, essential for almost any job where you want to employ agents. So one is the ability.To properly describe what you want, which is not obvious. Sometimes, you know, if you do things, you kind of know how to do them. But if I’m asking you to describe in detail to someone else, you kind of have to get a very good mental model. So the ability to actually describe is, uh, is, is, is an important one. The second one is the ability to delegate.You have to be able to break up your work into chunks so that you can give to different people. So that’s another skill. You take what you do and you say, you know what? I can parallelizeze these actions and then start giving them to someone else. And the third one is the ability to supervise.You now need to understand that uh the same way as a person might not be perfect in what they do. Obviously an AI in some cases is, you know, even probably less perfect. You have hallucinations, you have process errors, and so you need to find a way for you to verify your work because also at GLA, one of the things that we really kind of cherish is the fact that we always want to have human supervision of AIs. And so those three skills are something that I think will become aUm, uh, a basic, uh, requirement in the tool set of any employee, the same way as today, I mean, you need to know how to use a computer, right? And 50 years ago, it was not a requirement. I think in a few years it’s gonna be, hey, you know what, you need to be able to manage an agent to actually be able to,
10:54 spk_0
uh, how, how, how do I hold, how would I hold like a motivational town hall with hundreds of agents? Like those days are done though.
11:00 spk_1
I think actually you know what’s interesting that you’re saying the following because what what does a motivational town hall.Gets people in a state of mind for which, you know, they tend to be more excited and more productive. But, you know, you can do that to an AI as well. The way you write the prompt, the AIs respond, uh, really, really well to the way you asked the question. And so, in, in that case, it might not be a motivational town hall, but it might be you realizing, for example,And believe it or not, it is, it is true that if you’re kind to an AI, the AI will actually give you better answers. I’ve verified that myself, and it’s true. For some unknown reason, being kind to an AI is a good way to actually get them to actually give you a good answer. Maybe because, uh, in the training set, uh, you know, uh, like, there was a bias towards, uh, in literature, etc. about, you know, a certain type ofExpression or kindness that was correlated to good material being trained upon. I don’t know. But there are, there is a psychological element. And so it would be hard, it would be actually, it, it would be wrong to think about an AI as something that is just as mechanistic as a computer. A computer program is syntax error, so just tell me the same thing the same way all the time. It’s not the same for that.
12:26 spk_0
All right. Hangwith us, uh, Marco. We’re gonna go off for a quick break. We’ll be right back on opening bid.All right. Welcome back to Opening bid. Of course, Opening Bid sponsored by our friends at Vanguard, having a fun chat here on, uh, AI and and investment banking and the banking business with Marco Argenti, Goldman Sachs, chief information officer. So let’s, before we close or or I guess close the, the loop as one would say on the investment banking perspective, would it, is this going to help?Reduce the amount of hours investment bankers and analysts spend scrutinizing a deal. I mean, you know, I have to tell you this. I mean, some folks spend 80, 90, 100 100 hours on deals a week. I mean, these are big transactions, like, how does it make their lives better, this AI?
13:14 spk_1
Listen, this is just another transition like before, uh, uh, Excel, after Excel, before you could use Google, after you could use Google or a search engine. So there is a certain pattern or Yahoo. That’s right. Absolutely, absolutely. Why did I say Google?
13:29 spk_0
It’s OK.
13:30 spk_1
Of course, you can use, uh, Yahoo. Um, and so I think that will happen and is happening, uh, uh, naturally. And, and so it’s more about, uh, what part of your job, uh.It’s like something that you really feel that you act the.Unique value. It kind of elevates your work rather than doing repetitive tasks. And so I think the bar is just gonna increase. So we have uh this uh uh uh platform called the GSAI which is uh a platform that we built, uh, uh, to give access to the latest and greatest models like Open Air, like GPT, like Gemini, like Anthropic, etc. in a way that is safe and compliant for a regulated industry, etc.And then uh the tool called GSAI Assistant, which is essentially a chat, which gives you access to all those models, but at the same time, it links to all the data or some of the data that is relevant for you that lives inside the bank. And so what we see is that people that are exposed to that, first of all.You can you can literally see this learning curve that it goes super, super fast to the point that we get people using it uh like more, like 40, 50% a month growth in terms of how much they use it.But also you see what kind of questions they’re starting to, to ask. And so that’s how, you know, you see how their work is actually in their attitude started to shift. So at the beginning.You ask uh more like basic questions. For example, you want to know an acronym or you wanna know an explanation that you wanna kind of try to explain to a client, or for example, you have a portfolio of stocks inside the basket, and the client is asking you, can you tell me some information about those stocks, and you can ask an AI, etc.And then as you go to more uh deep usage, you’re starting to see what I said before, which is, hey, why don’t you do this for me? Why don’t you retrieve a chart? Why don’t you create a presentation that will include, for example, some macroeconomic indicators, etc. etc. So the sophisticof usage becomes a little bit like the interaction with a colleague. At the beginning, you kind of exchange ideas, etc. and then at the end, it kind of becomes entrenched in the way you work. And I think seeing that evolution is really, really fascinating.
15:51 spk_0
AI is really taking hold, uh, in a lot of companies. Has, has AI taken a a someone’s job or want someone’s job inside of a Goldman Sachs yet?
16:02 spk_1
Well, I think, uh, uh.In terms of tasks,Yes, we’re seeing shifts, for example, for developers, uh, we’re seeing, uh, uh, we’re starting to see that a lot of the mechanical jobs versus uh quote unquote, like, for example, migrate from a version, an old version of Java to a new version of Java, migrate to a certain web framework to another one, write documentation, write unit test, etc. is starting to be taken by the AI.But then I wouldn’t go, like I wouldn’t go as far as saying taking an actual job is that because the, yeah, no, I, I, the, the answer is so far no because there is so much more that we have sort of a below the line, uh, you know, every year when we do our budgeting and there is so much, uh, you know, more that we wanna do because it’s a competitive environment. And so at the end of the day.What we’ve done so far is really to reinvest the extra capacity which will give us additional competitiveness and I think for the foreseeable future in a competitive market, most companies will actually decide to reinvest.
17:11 spk_0
Do youthink AI will increasingly take jobs outside of Goldman Sachs?
17:19 spk_1
I think what, uh, reflection that I, that I, that I’ve been having is, uh, you know, there has been a lot of transitions, uh, over the years, right? Before computers after computer, before mobile after mobile, before the cloud after the cloud, before outsourcing to other, uh, you know, countries, to etc. And inevitably, the pattern has been, uh, you might have some job displacement at the beginning.But then those industries tend to create a lot of new jobs. Just look at, for example, like how we have like, you know, say 12,000 developers within Goldman Sachs out of 45,000 people. If we didn’t have computers, we didn’t have, we wouldn’t have those people. And so, generally, it rebalances itself.And I think that would be the same for AI except.There is a certain velocity. So like, you know, the transition to computers, it took decades. Transition to the cloud, it took, you know, several years. The transition to outsourcing, it took several years. And so there was more time to reabsorb, reskill.This transition is happening very fast. It feels fast. It feels fast. It feels like there is month instead of, of years. And so I think, uh, you know, one of the things I’ve been reflecting is, uh, you know, can the job market shift as fast or could you have a deeper displacement, uh, you know, when, while those new skills or while those new jobs get created. And that’s something I obviously don’t know the answer, but I think it’s one of the most, uh, uh, poignant, uh, questions that we should ask ourselves. I think
18:47 spk_0
there, there’s.There’s not just one day where we wake up and AI agents have all our jobs. You think it’s just a slow, it’s a slow evolution?
18:56 spk_1
Yeah, it, it’s gonna be an evolution. I don’t think I, I don’t know how slow it’s gonna be. In fact, I think it’s happening very fast to the point that, uh, if you think about it, like 6 months ago, we almost didn’t think about uh agents. Now it’s all, it’s all the rage. And one specific moment in time was, uh, you know, saying Q1.When, uh, a lot of this uh uh uh AI model providers starting to uh introduce uh reasoning models.Like, you know, the 0103 or a deep sick or uh all the others.Which means they’re not just answering your questions. They’re actually starting to think about it, how to answer. They can do plans, so they can create uh uh uh you know, like uh processes.And that is what really created agents, because an agent at the end of the day is, you give them a question or a task, they create a plan and they know how to execute that plan. That’s reasoning in a way. So, the biggest shift has been uhModels that are capable of actually reasoning and doing tasks on your behalf. And I think if you look at the if you put those in a timeline, you see that the acceleration has been pretty, pretty remarkable to the point that today pretty much all major models have reasoning capabilities and it can be used to create agents.
20:22 spk_0
We’ll have to get your final thought on this one, and I think you’re the perfect person to answer it given of course you work at Goldman, but of course Amazon.How far are we away from the day where we have a robot, an actual robot, let’s say they like the Tesla bot.is out there greeting clients.And then taking that client to a power launch to close a deal. Uh, is this a real, is this a real thing in our time?
20:47 spk_1
I first of all, I don’t know how much clients will enjoy that experience.
20:51 spk_0
Hello sir. But, but like,I mean, I see these robots, Marco. They’re really lifelike and they’re dancing around like
20:58 spk_1
I’ve seen a lot of, uh, robots, but my prediction is that, uh, uh, uh, especially humanoid robots, uh, first they’re gonna be used, uh, in, uh, in industrial settings like, uh, you know, for example, in a factory to assemble things, to move things around.I think human interaction, uh, is something different and and so personally I don’t look forward to that day for sure but I’m gonna be greeted by a robot.
21:25 spk_0
Well, it could be like if I have like a robot companion. I mean, I think Elo makes some good points here on the robot front. Maybe it’s just my companion. Maybe the companion I send down to greet the client coming into a Goldman Sachs building.And then maybe I’m the one that just launch.
21:39 spk_1
It’s possible. I mean, we’ve seen, uh, for example, many buildings in many countries have, uh, robots that are delivering packages. That’s absolutely great because, uh, you know, they’re extremely efficient, and they don’t misplace packages and the interaction is very simple. So I think uh.Uh, uh, you know, I think I can see a bright future for robots, uh, but I’m thinking that, uh, a certain type of warmth and human interaction, especially in, uh, in, in businesses like banking or investment banking, I think are gonna be for what it’s
22:10 spk_0
worth if I ever take a company public, I want to be greeted by a human. If I’m giving the bank all this money, I want to see a human and I.to buy me lunch. I leave it there to, uh, hear about some of the things you’re working on. Marco Argenti Goldman Sachs, chief information officer. Good to see you. I appreciate it. Thank you
22:24 spk_1
so much. Thank you. All right,
22:25 spk_0
of course, that’s the latest episode of Opening bids sponsored by our friends at Vanguard. Please do hit us with all that, uh, love out there on the podcast platforms. The thumbs up on YouTube. Love your feedback. I learned a lot from it. We’ll talk to you soon.