← Redefiners
Redefiners
Banking on AI: How Goldman Sachs CIO Marco Argenti Is Rewriting the AI Playbook
00:41:34
The world has gone through several technology transformations in the past 30+ years. From the launch of the Internet to the rise of mobile, cloud computing, digital transformation, and now AI transformation. In our second live recording of Redefine...
Loading summary
Transcript107 lines
- [00:00]
A
Call them change makers, call them rule breakers. We call them Redefiners.
- [00:13]
B
Hello, everybody. Welcome to Redefiners. It's Clark Murphy with Russell Reynolds Associates. We have a special treat today. We are live at the Goldman Sachs studio in New York City. Welcome, welcome, welcome. Just a reminder to all of you, if you'd like to see any of the episodes of Leadership Lounge or redefiners, they're on YouTube. And to our audio listeners, please hit the button. Wherever you get your podcast, give us your feedback, we'd love to hear from you. Today we continue our dive into AI and particularly how leaders will lead differently in AI with an executive with one of the great financial services firms of the world who's joining us to talk about that. But equally we're going to talk about his interest in music and how AI also is involved in cancer research and other things that are really important to him. Our guest today is Goldman Sachs Chief Information Officer Marco Argenti. And Marco is a member of the management committee of Goldman Sachs, the Enterprise Risk Committee and the Client Committee. Earlier, he spent his career with Amazon Web Services, also with Nokia, but as importantly, started dreamware, his own startup, which he sold successfully. So we're very fortunate to have a breadth of experience with Marco. Marco, welcome to Redefine.
- [01:26]
A
How are you?
- [01:26]
B
I'm very well. Good to see you. Thanks for carving time out of the world. You've been traveling everywhere, I gather.
- [01:31]
A
I know a little bit.
- [01:32]
B
Well, listen a little bit. India, the west coast, everywhere, everywhere else. So. So, Marco, an Italian who loves music, who's dedicated to healthcare research, you're sitting in New York City at Goldman Sachs as cio. How'd you end up here?
- [01:47]
A
Luck.
- [01:48]
B
Luck. Luck works for everyone. But my father said the amazing thing is the people who work the hardest, they have the most luck.
- [01:54]
A
So, I don't know, sometimes luck and hard work are indistinguishable. Yeah.
- [01:59]
B
Like, to your point, how'd you end up in financial services? Goldman Sachs, New York City.
- [02:04]
A
One thing about my career is that I kind of stuck to technology for my entire life. So I actually never really changed. I changed sector where you apply technology. But I was reminding someone this morning that I've been writing code. I no longer really do production code, but I've been writing code for 50 years now. So, I mean, the half century coder club and lots of change there. But I always kind of tried to keep my roots of a technologist. I didn't try to say, you know what, I'm gonna jump in to try to become I don't know, a management consultant. My field is technology. But then obviously technology changed so many times in the journey. And you know, I think a shift in my career has been when I went from being a sort of a producer and provider of technology. Like being on sort of the seller side.
- [03:03]
B
Yeah.
- [03:03]
A
To more like a consumer of technology and being more on the buyer side with Goldman Sachs, but also from being a provider of tools to the disruptors, to actually being leading the disruption from the inside. And you know, many people are asking me, how did you go from AWS to Goldman and what drove you there? And that was really the desire of seeing the transformation from the inside. And that was kind of pre AI in a way. Yes, yes, just about pre AI. And then this whole revolution actually came. And so that was almost like a mindset shift where I wanted to see how technology can kind of become part of the strategic agenda of a company that doesn't produce technology per se.
- [03:54]
B
Financial services hugely different than the giants of telecom. You work for one of the giants of telecom in the day. What's it like in financial services? What's the same or different?
- [04:06]
A
Sometimes I joke that when I moved from AWS to, to Goldman, he felt a little bit like Ted Lasso. I thought it was the same sport just because it has the same name.
- [04:20]
B
Yeah, right.
- [04:21]
A
Your kind of technology. Okay, great. But then things can be fundamentally different, although there are also a lot of things that stay the same. And so I think it was an interesting transition because it was more about me having to adapt to a new world, but also at the same time actually having this, you know, like having this task of driving cultural transformation so that a bank IT department will look more like a tech companies, you know, engineering department. And I think that is really what was very, very exciting for me. You know, differences are profound. First of all, it's a regulated industry, is one of the most regulated industry. The purpose of the company is not to sell technology in a way, but it's to sell financial services. And then the question is, how are you being an enabler without being the object of the business, of the subject of the business itself? And so the relationship, navigating the relationship between the technical team and the non technical team and being for me in a company for the first time where most people are non technical. It was a big transition, although the technology team is pretty big. I mean, we have well over 12,000 engineers at Goldman Sachs.
- [05:45]
B
12,000 engineers?
- [05:46]
A
Yeah. Actually that's a little bit of a conservative number, but let's say 12,000 out of 45,000 ish employees. And so kind of one out of three is an engineer. One mechanism that really I tried to focus, that I kind of borrowed a little bit from Amazon, is the working backwards mechanism where you're asking engineers to really put themselves in the shoes of the customer or the client and really ask yourself client centric questions such as who's your customer? What is really the problem opportunity that you're trying to solve? And how do you know that there is a problem opportunity? How do you measure the return on investment, how the experience looks like before you write the first line of code? And I think another way to put that engineers tend to be naturally attracted towards the how. Engineers love the how. Sometimes the how is, oh, I found this new tool. It's amazing, I love it. I just want to try to find where I can use that. And so the classic sort of a natural mentality of engineers is how, what, why? And I think what, you know, working backwards means is that you do why, what, how?
- [07:04]
B
Right, right, right, right.
- [07:06]
A
And so that I think was also something that was one of the big directives of, of, you know, like a cultural transformation to write. Because definitely you can be technical or non technical, but you definitely meet in the conversation at the why point. Because the why is where the client is and what the business opportunity is.
- [07:26]
B
When you came to Goldman Sachs, did the reversing, the how, what, why, it started when you arrived or you think it was already here. How much cultural work did you have to do with the engineers to get it moving faster and more successfully At Goldman?
- [07:43]
A
I think the why is really in the ethos of Goldman, because the client centricity is so strong in here. So it wasn't too much of an overall cultural shift because engineers were exposed to that. But I think what we really worked on was kind of to take it at scale, to have mechanisms that are consistent across the organization, to have people to actually write down those specs in a way that helps other people understand and have conversations around that. It's more like how do you turn a bunch of sort of, you know, like great individual efforts into a well run factory at scale? Yeah, because that's also one of the things that I learned at Amazon was that designing for scale makes a difference in really the way you do things and having consistent processes across having pretty much the same way of executing and doing things. If you move from one team to another so that it's easy for you, you find a familiar environment. And then one of the very first things that I did that wasn't there for engineers was to put down some tenets of what it means to actually be an engineer at gom. Right. And what are the things that really matter, like leading with data, build with purpose, which is like the, why promote inclusivity? But having these 10 things that people can look at and any employee can look at and say, you know what, if I kind of apply those tenets in a discussion, there is an alignment of underlying beliefs that will actually help us go somewhere. And I think that was one of the other tools that I really, you know, that wasn't really there. But then I think the. It was not a very hard sell because of the sort of inherent client centricity of the firm.
- [09:47]
B
Got it. So I'd love to start broadly and go to the specific. From conversations you and I have had over the last couple of years. The kind of rage and energy of AI in the last year is not something new to you, obviously. Where do you see the opportunity and power of AI, of AI in the world today? And we'll come to financial services in a minute.
- [10:10]
A
It's almost where the Internet meets books meets tutoring is that intersection, you know, because on the Internet you generally have more of a lightweight type of information that is very consumable and very fast.
- [10:26]
B
Yep.
- [10:27]
A
If you want to go deep, you have to go and actually read the book. And if you want that, if that book, for example, is too difficult for you, you probably need to get a mentor or a tutor of some sort and it's going to help you like walk that ladder that takes you to the understanding of what's in the book. All those three things, A, the accessibility of the information, B, the depth of the information, and C, the adaptability of the information to your own knowledge, are what the AI excels at. Okay.
- [11:01]
B
Yep.
- [11:02]
A
And so it is, I think already, even with all the limitations that were very clear, the most powerful learning tool that there is for individuals, for corporations, for students, for anybody. So that to me is one of the profound long term change factors. Okay. Where almost everybody, like someone else said, could have individualized tutoring, which is kind of the gold standard of education. Okay. Knowledge has been a domain that we've been kind of exploring. And the very first AIs were mostly knowledge centric AI. You ask a question, you ask a follow up, and you keep asking a follow up. And this thing takes you down the rabbit hole better than anybody else. And then you can go in and out and so forth and it adapts to your level of knowledge. Great. In the Last year, I would say the landscape has shifted quite powerfully because two things have happened to AIs that weren't there before. One is memory. Now, the AIs, they kind of adapt with you as you start using them. Before, it was like 50 first dates, right? Every time you talk to an AI, it's like, hi, Clark, never seen you before. And by the way, the way it was done was actually kind of smart because obviously the AI doesn't really change its weights. It's not that you train it continuously. It will be too expensive for you or for each of us to have our own model that has different weights. But then every model provider understood that by adding a sort of a scratch pad on the side, memory scratchpad, the AI itself could annotate things back to the Adam Sander metaphor that he was putting all these post IT notes without changing the brain of the person. The AI has its post it notes that now they can read and they come into the prompt before you ask the first question. Okay, so that is, that was a major breakthrough. And you can tell, I mean, that somehow they tend to adapt to your personality the way you want to hear things and so forth. So I think that is, that is, that is there has been a major breakthrough. And the other one which happened, you know, let's say late 2024 and then very powerfully in the first quarter of 2025, is going from AIs that can answer questions to AIs that can actually plan and reason. And now if you look at any of the AIs at GPT or Claude, or you name it, or Gemini, when you ask a question, the very first thing that you see is it takes some time to reason. And then if you look at the reasoning, which you can generally see, you see the chain of thought. And it will ask itself. Okay, Clark has asked me to understand, I don't know how color derivatives work. Let's first define what means by color derivatives, et cetera, et cetera. And then let's goes and out and checks sources of information, decides which one, you know, if there are enough sources of information that agree on a certain. So there is a whole plan. And then he summarizes its own. Now I'm ready to go forward to, to, to go forward. And that is a breakthrough. That's a different way. That's almost like a different AI. And then if you want to extract that to kind of finish, you know, the question is, you know, we were starting to use the word agent almost to a fault that we're using it so much. But what really means is like you extend the concept of reasoning into the concept of okay, now I can actually take action, I can do things. It's not about answering things. It's actually I can go and potentially, you know, I could book a flight for you or if you want, if you ask me, okay, I want to do that. Is it okay? And then it goes and does it. Or I could do and create an entry on your calendar or I can send a message to someone to say that you're late. And then in the, in the corporate world they can do a lot of things. There are kind of, you know, from acting for example on a code base and actually create code and creating a pull request or many other things, summarizing documents, creating a summary and then sending this summary to a newsletter or you name it. And now you have AI that does things on your behalf with a certain degree of autonomy. And of course we're very careful, especially like back to, you know, what's different in a regulated industry, etc. Is that we put very strict boundaries around what it can do.
- [16:13]
B
Right.
- [16:14]
A
And always having a human in the loop. But still you can almost like, you know, consider it almost like a coworker.
- [16:21]
B
So this comes I think to the one of the core questions about leadership and you've spoken pretty eloquently in other places about agentic AI agents. How do you think perhaps in a non regulated industry might be easier to talk about? How are leaders going to have to learn to lead? The opportunity of leading in this concept of AI agents or digital twin or something else. What does that mean to leadership, do you think?
- [16:48]
A
The first thing that I want to observe is that what's going to do to people that are today not required to be leaders. So what does it do to non leaders? And then we get to leaders.
- [17:01]
B
Yep.
- [17:02]
A
I think when you're starting to have agentic co workers that you can potentially like hire yourself up to a certain number, but mostly limited in cost, then I think everyone, even people that today have never managed people are going to have to develop at least three critical leadership skills. The first one is the ability to clearly explain what you want done. The second one is the ability to delegate and not necessarily to one person, but maybe to multiple person or multiple co workers. Which means, for example, how you break it down into units of work that can be executed in parallel and what are critical paths and so on and so forth. And the third one is you need to have the ability to supervise. You cannot blindly trust, especially AI agents and so you need to be able to put together mechanisms for which you can audit, you can verify, you can have other agents who check on other and so work of other agents and so forth. And so I think that will become a new requirement almost for everybody.
- [18:25]
B
Interesting. Yep.
- [18:26]
A
The same way as, you know, like, knowing how to use a computer became a requirement pretty much for every job. Fundamental, basic leadership skills that are going to become the purview of people that today are not worried about managing people, but they're mostly worried about doing work.
- [18:45]
B
Okay. Yep.
- [18:46]
A
So that's point number one to leaders that are already managing people and other leaders. I think this raises a lot of questions, which is, first of all, there is a skills question. So you need to kind of, as a leader, enable your organization to go through this journey. The second one is how do you actually enable or create or exercise that muscle group inside your organization? Because it will require training, it will require a new curriculum almost from, you know, when you onboard a new employee to how you retain employees and so on and so forth. How do you do performance reviews? I mean, it's, it's, it's in the world. Yeah, it's everything. Also, how does your pyramid change what we call pyramid? For example, the mix between junior and senior people. Is juniorization the right way? Or are you thinking that maybe because people need AIs need supervision, are AIs and you, you, you read that quite a bit right now there is a big debate of, you know, maybe the people that are mostly affected by the AI are actually maybe the junior people because they're the, you know, those tasks are probably the first ones that the AI is going to be able to replace. I personally am quite optimistic of, you know, like, because I've seen examples of a lot of, you know, people that are kind of AI natives that are doing extremely well in the world of AI. And so, but this will be another question that any leader needs to ask themselves, which is how do I actually change the mix between seniority and general population? And how many tasks can I give? Because one of the things, remember, the third prerequisite is the ability to supervise. The AI is most likely going to be your most junior employee. But then what level of seniority do you need in order to supervise an AI? That kind of changes the game a little bit. Another interesting thing which I don't see talked about too much is how does it change the span of control? What if you have AIs that are going to help you do the job as a manager? So you could have potentially like, you know, a lot of the supervision could be done by AI. A lot of the approvals of staff, of vacations or whatever can be done by AI. The AI could help you with performance reviews, the AI could help you for understanding what are potential red flags and so forth. And maybe then the new span of control could be higher. And then the last thing that I'm going to say is as a leader, you kind of need to see where the puck is going regardless of the industry that you're in.
- [21:21]
B
I agree.
- [21:23]
A
And I think this kind of changes the balance a little bit between how much time a leader spends on the today versus how much time needs to spend on the tomorrow. And I've seen also in my little world that I've been spending more and more time and telling my leaders to spend more and more time trying to think of how the future is going to look like and what are the areas that we take for granted today that might actually be disrupted by also and primarily on the business side, not just on the people side. We'll be right back with Marco Argenti.
- [21:58]
B
But first we'll hear from Jennifer Flock.
- [22:00]
A
A leadership advisor in our Paris office.
- [22:03]
B
Jennifer will explore what it takes to.
- [22:04]
A
Build transformation capabilities that last here's the.
- [22:08]
C
Problem many CEOs face today. Their teams are exhausted by transformation. Just as they finish one major change initiative, another urgent shift demands attention. They are burnt out. Yet the pace of change continues to accelerate. The traditional approach of treating transformation as discrete projects with clear beginnings and ends simply isn't working anymore. So how do effective leaders solve this problem? They stop thinking about transformation as something they do and start building it into who they are as an organization. In our latest article, we identify five specific leadership behaviors that enable perpetual transformation. We also reveal how the best CEOs build always on capabilities that move organizations from adapting to change to creating it. To learn more about building transformation into your organization's DNA, you can find the full article via our show notes or by visiting russellreynolds.com and now back to.
- [23:15]
A
Our conversation with Marco.
- [23:19]
B
This concept of focusing on the tomorrow. Yeah, how do you you're at the edge versus most of our listeners. You are at the edge of learning and living about the impact for tomorrow. But how do you stay ahead? How do you think about where the puck's going to be?
- [23:37]
A
For me, my trick is simply that I talk to a lot of people and I talk to a lot of people that I respect and actually find the time to put in my own calendar sitting down with people that I consider Leaders.
- [23:54]
B
I agree.
- [23:55]
A
And I just talk to them.
- [23:56]
B
It's the best. Smart people or smart people.
- [23:58]
A
And now I have to say that maybe up to. I don't know, that could be clients, it could be vendors, it could be CEOs. And I spend almost like 15 to 20% of my time actually doing that. That being over indexing on that. Also traveling. You go to a different country, you go to a different reality. You go to Japan or China or Europe or the Middle east, and then you see maybe new things that you haven't thought about. So kind of getting out of your bubble, I think, is very, very important. You know, obviously in America, we have leadership in AI, but it is a global phenomenon. Yes, it is definitely global phenomenon.
- [24:31]
B
Yeah. Somebody. Somebody. Actually, Jim Coulter said, we're all focused on the super bowl, and it's amazing, but the World Cup's five times bigger than the Super Bowl.
- [24:40]
A
That's great.
- [24:40]
B
I like that.
- [24:41]
A
Absolutely.
- [24:41]
B
I like that. So here's. I have a question of balance. One of the issues is being in an institution with 45,000 people, that you're trying to create change and a pace of change. So adoption and culture is one of the big challenges. At the same time, there are massive expectations of you and your group to create faster change. So on the one hand, you're like, come with me. Hurry up. Let's go. On the other hand, everyone's saying, how fast will you get us there? How do you balance that?
- [25:15]
A
You know, there was one point in my career where someone told me to read a book called the Goal, which is a famous book. It was actually Jeff Bezos was giving it as a recommended reading. Okay. And, you know, if there is one thing to take away from that book is the fact that if you want to understand how to, you know, make something faster, you need to, you know, be very relentless in always understanding what your bottlenecks are. And what I personally like to do is to go, first of all, the. Working backwards actually helps a lot, because when you have people actually write down and have an rude FAQ where they ask themselves, why can't I do it faster? And then they list blockers, you find out that 80% of the time, there are blockers that are more like by convention rather than by necessity. Think about this by convention, not by necessity, just because we do things in a certain way. And then there are certain people that cannot change those rules. And other they can. And so maybe those rules are something that you can reinterpret or you can just have a conversation between people that Generally don't talk to each other. So generally that's a good thing to try to go deeper into what are the bottlenecks and try to have conversations around those bottlenecks. And the other one is I'm a big believer in standardization. Standardization and automation is what really gives you speed. One of the learnings, and this very much in general, is the fact that when given a business requirement, the first instinct is to go and build a custom solution, right?
- [27:00]
B
Yes, always, always.
- [27:02]
A
And so then you end up with, okay, fine. And in order to build that 10% that is different, I'm actually taking 90% of the time because you need to go through approvals and take risk reviews and all that. And it's the same where you actually want to build something. First question you need to ask yourself is, can I actually do it with existing tools rather than with having to build a new solution? Now, this is a non AI statement, but you know what? It actually becomes so much more relevant in the age of AI. And why is that? Because computers tend to be very inflexible, right? Because they're programmed in a deterministic way, right? You have a bunch of instructions and those instructions are generally quite rigid. And when they do one thing, they do one thing. The reasoning is flexible. The underlier, like the tools that you have at your disposal, like for example, knowing who you are having access to corporate directory, having access to information, etc. Is something that can be common to many, many, many use cases. And then those use cases can actually be solved by something that almost has not only reasoning capabilities, but there's certain common sense, which is the ability to see patterns. Okay? And so I think that's really what we need to be prepared of, which is that we're going to have a new approach to the world of software and to the world of applications that is more about how do you ask the question what kind of task you can actually describe and delegate and what tools you give to your disposal to your AI. And then at that point, a lot of the bespoke applications are never going to be written. And so back to your original question. I'm finally closing a circle. Imagine if a business use case that is similar enough to something that the AI has seen before and the business person can just do it with the AI without even having to potentially write or ask to write a single line of software. That would be the ultimate goal.
- [29:14]
B
So you have been at the forefront, really probably haven't thought about it this way. In the telecom world, when Nokia was where it was to be aws, certainly financial services. Goldman has typically always been where it's going to be, where it is. Has there been a redefining moment in your career for you about who you were going to be and what caused that to happen?
- [29:44]
A
So when I joined aws, the first thing that I was given, believe it or not that there were still pagers around. Probably the last pagers I ever used. I was given a pager.
- [29:54]
B
Really?
- [29:55]
A
Yeah. I was literally as part of the.
- [29:57]
B
You cutting edge guy.
- [29:58]
A
You, can you believe it? A pager that would ring me with a battery physical that was substituted by an app maybe a year later. But the pager at that time was still the most reliable way to be alerted. Why? Well, because first of all, nobody had fixed lines anymore, so you can call you at home if you were sleeping. And then also back then, no cellular phone had the ability to kind of receive a message which will bypass your silence or whatever and just ring in the middle of the night. Now you can do it. And even back then, a year later, finally we got rid of the pages. So I was at the tail end of that. But what was the significance of that was that no matter how senior you are, if something happens, you get paged in the middle of the night, you wake up and you own it. You're on a support call with your engineers, you're going to communicate to clients, you're going to have to basically kind of manage that availability. And it's you. And that's the ownership principle that leaders are owners, they don't point fingers. There's not a collaborative thing is a collaborative things, but the ownership is yours. So you can collaborate all you want. At the end, the bucket stops with you. When push comes to shove, you have to own the design, you have to own the support, you have to own the maintenance and all that, because all those things, they actually drive your design in the first place. And all of a sudden the developer that doesn't want to be woken up in the middle of the night is going to write software very differently, very differently. And then at scale they will make night and day. And so that's what I've been trying to do afterwards, which is always hold leaders accountable. Pushing ownership as possibly the most important leadership principle for a leader, which is that this buck stops with you. And you can delegate work, but you cannot delegate accountability.
- [32:04]
B
That's a great line. Can't delegate accountability. So switching gears a little bit, you're very passionate from a personal standpoint about cancer research and having lost particularly one of your Fellow musicians, how does the book, the knowledge, the tutoring, the extraction, the action, how does that work about accelerating cancer research and outcomes?
- [32:31]
A
So I was in the board of the Pancreatic Cancer Action Network for a while, and then I joined the board of a research hospital called Fred Hatch Cancer center in Seattle. It was actually three years ago where AI was so new that there was a lot of talk about, oh, my God, AI is dangerous. It can be the end of the world is going to revolution is going to turn against humans. There's going a lot of doomsday type of scenarios. Remember the super intelligence paperclip machines and all that. But so in that climate of a little bit of, you know, fear of AI, I was more like, okay, if AI is that powerful, then the flip side of that coin is we should leverage that power to really solve the biggest problems in the world. And if I list, there's many problems, but if I list something that touches every single family in the world, with very few exceptions, is cancer. Okay? And therefore, I think we have the moral imperative to do something about it. Now that we've been given or we have created what's possibly the most powerful technological tool that we've ever seen.
- [33:38]
B
Right.
- [33:39]
A
If you ask all the CEOs that you talk to, inevitably they will say that we have not seen a revolution of this magnitude in technology ever. Even bigger than the Internet. Some really not only hope, but kind of conviction that AI could do something major for cancer and for cancer research emerged. In me, there is a press release where you see all the companies that are involved in this effort, which we call the Cancer AI Alliance.
- [34:10]
B
Someone should look it up. It's. It's some of the greatest companies in the world.
- [34:13]
A
Yeah, yeah. And so you have some of the best technology companies in the world and some of the best research hospitals in the world. Okay. And you can find more information on cancer alliance AI.
- [34:25]
B
There you go. Canceraliance AI.
- [34:28]
A
Curing cancer is a difficult even concept to explain because, as you know, there is about 1200 known type of cancers. So it would be like saying curing diseases. It's just too broad. But also there is a lot of specificity there. And also finding new molecules and new drugs, although very desirable, the drug pipeline is long. It takes years, decades sometimes.
- [34:54]
B
But you're combining the greatest cancer research centers with some of the greatest chip manufacturers, software providers, and technology companies to accelerate.
- [35:08]
A
And that's the thing. We're trying to inject speed in the system, first of all, like a 10x speed. That will help. So today if you have cancer, depending on where you are, who you are, which system, 30 different outcomes, you're very different. Yeah, yeah, yeah. And there is an area under that curve, whether you are like Clark Murphy and you are in New York, whether you are someone in rural, some other country or whatever, there's a huge gap in that area under that curve. There is a lot of years that you can extend quality life. So one of the goals that we've established for ourselves is to give back a million years of quality life in 10 years.
- [35:54]
B
Give back a million years. Talk about giving back. My goodness.
- [35:58]
A
So that is the bold, ambitious goal. It's gonna be tough, but it's not impossible.
- [36:04]
B
But again, for our listeners, CancerAlliance AI, this is the greatest research centers there are, with some of the greatest technology companies, doctors and. And people of technology. What a great thing. So, last question. Music. I don't know. You travel all over the world nonstop, which I see firsthand. You are very focused on also cancer and this cancer alliance. But you love music, and you find time for music. In fact, you have, I believe, your own recording studio in Queens, where I heard the Lumineers play live not too long ago.
- [36:40]
A
It's in Greenpoint. Greenpoint, yeah, in Greenpoint.
- [36:43]
B
So your studio's in Greenpoint. Tell us about your love for music. You have your. You're in a band still?
- [36:48]
A
Yeah, yeah, yeah.
- [36:49]
B
Okay.
- [36:49]
A
Occasionally. But, yeah. So it's part of the same initiative of cancer in the sense that, you know, like, we used to have a garage band with a bunch of tech, you know, executives in the basement trying to be cool in a way. And one of them, which was the singer and that was a CEO of. Of an AI company back then, six years ago, one day he had some severe back pain. He thought he strained or a muscle or something. And then the next morning, I got a call, and they found stage four pancreatic cancer. And he died literally eight weeks late.
- [37:27]
B
Oh, my gosh.
- [37:28]
A
And so we decided to say, okay, you know what? In his memory. But also, music is one of the biggest triggers of emotions. And what if we channel that emotion into a mission? And so we started to go out and play. We started to raise money. We started also to partner with other bands to kind of, you know, like, bands come and play, and then people donate money to the mission. And so that started with pancreatic cancer Action Network evolved into Kaya. And that's why we got people like the Lumineer. So we got people like, you know, Katherine Russell. We got people like, you know, we have many really great artists that came our way. And you know, I mean, just with that alone, we raised, you know, close to $3 million. I mean, and, and you know, and, and, and everything counts. And so, you know, if I go on stage, it's not because I want to be famous, it's just, you know, I go on stage because I know that people actually react to that. And then when they are in that state, you know, they're also reacting to the message that we're sending to them. So that's kind of part of the same passion.
- [38:28]
B
It's a successful message and, and, and you get enjoyment out of it.
- [38:31]
A
Yeah.
- [38:31]
B
In progress. So we end every podcast with some rapid fire questions. So whatever comes to your brain right away, instantly, which is a fast moving brain. So you sit on the board at Carnegie hall. You love music of all types. If you could have dinner with any musical artist, alive or dead, who would it be?
- [38:47]
A
Leonard Bernstein.
- [38:49]
B
Oh, there you go. There you go. What's the best piece of advice you've ever received?
- [38:55]
A
Someone quoted Mark Twain, they say that, you know, I had a lot of trouble in my life, most of which never happened.
- [39:04]
B
If you could wake up tomorrow and gain one skill or capability, what would it be?
- [39:10]
A
Speaking English properly.
- [39:13]
B
People love the accent. You're not getting that one. It's not a chance. So you were a guest professor way back when, early in your career. If you could teach one subject now, what would you teach?
- [39:27]
A
How to reverse the how, what why? So the working backwards for engineers, how to attack a problem and how to look at software from the customer's viewpoint. There is a lot of unofficial literature around that and it is actually one of the first things that we want that we teach to people that come out of universities, which is a. You need to think slightly different about software. It's not just about data structures, etc. It's about having a clear mental model, is about asking the right questions. And so how to build software in the new world of AI and in the new world of, you know, like where the why becomes the most important thing, would be a super fascinating subject to teach.
- [40:06]
B
There you go. Well, I think we'll end where we started with what you talked about is people are typically how, what, why you and the client centricity of Goldman Sachs and AWS starts with the why, not the how and goes backward. I think this concept from the power of AI is the Internet and the book with the tutor, except the power is the extraction to then action.
- [40:33]
A
That's right.
- [40:33]
B
And what does that mean? And how do we take advantage of that action? But leaders have to be owners of working with these tools and maximizing these tools.
- [40:43]
A
That's right.
- [40:44]
B
And giving those who aren't leaders tools. Potentially, though, they have to learn to supervise and giving other leaders probably much vaster span of control if they can manage the AI tools twins what it might be. The upside is enormous, but the verification and the authenticity is critical to making it that powerful. At the same time, we should all have the mission around the cancer research and that we think about what tools will do for us and productivity and profitability. We also have to think about what tools do for mankind. And that's a lot of what you've thought about. So thanks for being who you are and joining us today in Redefiners.
- [41:28]
A
Thank you so much.
- [41:29]
B
Take care.
- [41:29]
A
Thanks for having me.
- [41:30]
B
Ciao.
- [41:30]
A
Thank you. Ciao.