Joe Fuller: How are investment firms rethinking workforce strategy as automation, evolving regulatory requirements, and technological disruption reshape the business? Demand is rising for expertise in data science, ESG, and digital product development—while many operational roles are shrinking—according to industry research and data from the U.S. Bureau of Labor Statistics. In response, firms are reexamining how they attract, develop, and organize talent to align with shifting investor expectations and a rapidly changing competitive landscape.

Welcome to the Managing the Future of Work podcast from Harvard Business School. I’m your host, Harvard Business School professor and non-resident senior fellow at the American Enterprise Institute, Joe Fuller. My guest today is Nitin Tandon, chief information officer of investing management giant Vanguard. We’ll talk about how the firm is marshaling its tech workforce to support its core business and its forays into new markets. We’ll look at how artificial intelligence is changing the skills equation for everything from Vanguard’s digital infrastructure to how it advises clients. We’ll consider the evolving role of the CIO from service provider to business partner. And we’ll discuss the changing geography of the firm’s talent strategy. Well, Nitin, welcome to the Managing the Future of Work podcast.

Nitin Tandon: Thank you, Joe. Pleasure to be here.

Fuller: Nitin, you’re the CIO of investing giant Vanguard. Tell us a little bit about your career progression and how you find yourself running such a big digital organization.

Tandon: Sure, Joe. I joined Vanguard in 2019, about six years ago, as chief technology officer, and I got into my current job as chief information officer in 2021. Prior to Vanguard, I worked with Deloitte Consulting for 17 years. I was a partner in financial services technology, and I led the cloud practice in financial services before I joined Vanguard. In fact, Vanguard was a client of mine for about three years before I ended up joining Vanguard in 2019. And in my experience in working with Vanguard in those three years, I found the unique ownership structure as well as the mission-driven culture pretty unique across all of the financial services industry—not to mention that they had embarked on a cloud transformation, which was very attractive to me at the time. Before Deloitte, I was with Citigroup in Southeast Asia for about four years.

Fuller: Well, it’s always very flattering for a consultant to get that that offer to join a big, advanced client and move from the world of giving advice to the world of actually doing and managing. 2021—that’s, of course, in the aftermath of the worst of Covid. That whole period was a period of dynamism and migration to the cloud for lots of big companies and also, in the run-up to Covid, big digitalization efforts. Tell us a little bit about the evolution of Vanguard’s digitalization and cloud migration strategy. And how did they play out, not just technologically, but through the types of skills you have to have in your organization and the training and development you have to do for your colleagues who are not technologists?

Tandon: Yeah, absolutely, Joe. Our cloud transformation journey wasn’t quite driven by the pandemic. It was driven by coming to the end of our last phase of technology strategy in 2019 and then figuring out what’s next for us. Our web experience, our mobile experiences, had decayed and were due for a refresh. Our enterprise technology strategy had three distinct objects. We wanted to increase the agility with which we responded to our business and our clients’ needs. We wanted to use data insights and AI/ML [machine learning] to offer better insights to our clients and advisers. And then, of course, no tech transformation is complete without a focus on talent. On talent, our big focus was working in new ways. There’s no point working in new technologies with new processes if you’re still working in old ways. We had cross-functional product teams with business technology, CX [customer experience], engineering all working together toward client-specific goals and objectives. Think of an onboarding journey, or think of a client platform. So Vanguard said we want to be 80 percent-plus on the public cloud by 2024. We ended last year at 82 percent. And you might ask, like, what that led to, in terms of benefits. We launched a brand new website and mobile app in 2022, and our CSAT [customer satisfaction] scores are the highest ever, with a 70 percent reduction in major incidents, post-transformation. And then, if you look at time to market, what used to take us months and quarters now takes us days and weeks in terms of our responsiveness to our clients’ needs. So our developers today deploy code every day to production, and we’re able to move at a much faster pace when it comes to the ability to respond to our clients’ needs.

Fuller: Nitin, was the introduction of those multifunctional teams and the move to the cloud, did that put a big burden on the organization to not just learn—accept the need to work differently, but to learn how to do it? And how do you manage that transformation?

Tandon: It was definitely a big transformation both from a technology perspective as well as change-management perspective. So there was a top-down positioning from leadership which says, “This was important to us.” We put a set, we put a date, we developed a strategy, and we helped people understand why. And once we did that, we’ve seen the benefits. But to your other point on training and upskilling, Joe, yes, we had to train thousands of people on new ways of development in the public cloud. And we weren’t just lifting and shifting workloads. We were refactoring applications to be cloud native. So we had a three-pronged strategy, Joe. We went and hired some cloud architects who had done this before—because we hadn’t—just as evangelists, internally, who could help our teams understand how do you refactor applications on the cloud. We also leveraged our sourcing partners, who helped provide capacity at scale that was required to modernize all the applications. But then, most importantly, we went through an extensive retraining program that had a few different elements. The first was, we built role-based learning. So whether you’re a product owner or if you’re a cloud developer, different roles had different learning curricula. We incentivized certifications. So we went with AWS as our cloud platform of choice, and we incentivized people to get AWS certified. We introduced a badging system, which celebrated people who attained a certain level of mastery over the technologies and concepts. We also had a number of different learning events interspersed throughout the year. We wanted to reinforce that learning culture, Joe. So one of the things we did was, we said, “We’re going to consolidate this into a three-day learning conference, and we’ll call it ‘Unlimited.’ And we’ll invite everybody across the company to immerse in learning on those three days.” And that has been a huge success. You know that that sends a message to our crew—we call our employees “crew”—that, especially in this day and age where the half-life of skills is rapidly becoming shorter, you have to continuously learn and relearn your skills. So in those three days we choose a theme, and we really immerse in learning not just with technologists, but also with our business community. And then finally, celebrating success was important. So once we started to move some applications successfully, we helped showcase those wins to people to understand so that others could draw inspiration as well as see how to move their applications to the cloud.

Fuller: It certainly seems that that the ability to manage these types of technological shifts through the lens of making the transformative experience that’s not just cutting over from one platform to another seems to correlate with success. Certainly in our research here at the Managing the Future of Work project, we’re now seeing that AI experiments in large companies that have significantly more associated investment in transformation skills and transformation systems are doing much better than those that are not, even though it seems like the technical merits of the projects going in on day one are equivalent. So there’s something about managing the associated human processes that really seems to be integral to unlocking the value quickly.

Tandon: I couldn’t agree more. If I look at all my transformation experience in consulting, as well as at Vanguard, it’s a cliché probably to say this, but yeah, technology is the easiest part, you know, of any transformation. It’s people and changing ways of working that is the toughest part of transformations. A few things are important, in my experience, Joe. First is helping people understand why. People want to understand the impact of their work. They want to see the value in their work. So linking the transformation back to their purpose, we found, is immensely important. Second is creating an environment where they can collaborate and innovate. People also like solving challenging problems. But they can’t do that in a siloed organization, so the ability to work in cross-functional teams that are then empowered to innovate is a big unlock. Of course, you have to train and provide coaching. But purpose, collaboration, and culture, to me, are big drivers of a successful transformation.

Fuller: Were there any particular challenges you ran across consistently or “watch outs” that you’ve now cultivated as a sixth sense to keep in the back of your head as these new technologies emerge and as the pace of transformation picks up?

Tandon: If I look at our cloud migration progress from 2016–2020 and then 2020–2024, right, one of the big differences first, of course, there was some learning in the early days from 2016–2020. But I think setting top-down direction, clear direction, and helping people understand why you’re embarking on a transformation journey, to me, is perhaps one of the most important things that I focus on now. Second, especially in IT, you want people to understand this is not just an IT project. It’s a business initiative or a business transformation, because more and more technology is providing differentiating business capabilities. And you can say a cloud transformation is an IT transformation and AI transformation is an IT transformation, but in reality, it never really is. You’re moving to the cloud because you want variable cost, you want faster time to market, you want more business agility. So I’ve found, both in our current AI transformation, as well as our cloud transformation, working hand in hand with the business and helping, whether it’s line-of-business leader or business stakeholders, understanding this is an enterprise transformation, not an IT transformation, is also pretty key.

Fuller: How does that apply to the way you and Vanguard are approaching generative AI? We’re talking in late April 2025. Every quarterly board call for every public company, there are questions from analysts about generative AI. I can’t pick up a newspaper without reading some opinion column or news column about generative AI. Certainly at our school, many of my colleagues are working hard to understand the management challenges of using generative AI in the disciplines that they study. What are you up to? Where is it being applied? And what are your early lessons learned?

Tandon: Yeah, AI is perhaps one of the most disruptive technologies since the Internet, and it has significant implications in our industry, Joe. Not only can it drive higher levels of productivity, but also transform all the ways in which our clients interact with us. At Vanguard, we’ve been experimenting safely with AI for the last 18 months or so. We started off with productivity use cases. So we focused on developer productivity, marketing content creation, as well as contact center assist. And we’ve seen some interesting results. Like, we’ve seen 10–15 percent productivity gains across those experiments. And that positioning was deliberate. We said in the first 12–18 months, we want to learn the technology. We want to see what its promise can be. We want to see how we can use it safely. And then we will place bigger bets. We are now expanding it to other areas as well—helping using AI and client experience. So think about all the ways in which our clients interact with us. You know, web experiences will be far more intuitive, far more interactive, using generative AI. In fact, I often say this internally: Would a website of tomorrow look the same as a website of today? Do we need structured navigations of websites of today tomorrow? Or can you just have a simple text box which says, “Joe, how can I help you today?” And Joe says, “Oh, I’m curious what my balances are, or I’m looking to retire at a certain age.” So just like you would have a conversation with an adviser, you know, web experiences can become far more intuitive and interactive. Second, if you think about phones, you know we can, of course, contain a lot more calls by providing first-touch resolution using AI, because AI can understand intent and issues better and help resolve or help resolve those issues faster. And where it can’t, it can help us redirect those calls better to our crew, who can then help the clients resolve problems for highly, highly complex issues. And then mobile as well. I mean, imagine being able to just have a conversation, hands free, in any language that you want about your financials. So Gen AI has immense applications when it comes to client-facing technology. And that’s one area we’re investing in. Another area is enabling our crew better. So think of financial advisers who spend today up to 50 percent of their time just preparing for client conversations or administrative work related to client conversations. Using Gen AI, we can offer them better data to have those client conversations. We can also do in-time coaching while someone is on a call and provide them with, let’s say, the latest economic models or perspectives from our chief economist. And then also help with things like call transcription and summary and action-items apps. So significant potential in helping our crew, client-facing crew, help serve our clients better. And then, finally, we’re also investing, Joe, in, you know, safely scaling this technology. Because, for all its promise, there are things like hallucinations, you know, bias explainability, especially in the financial services industry. So right now, we have a human in the loop when it comes to our use cases. But before we are able to exclude the human from the loop, we want to be sure that we would be able to meet regulatory and legal requirements. But not only that, we feel comfortable with the results of AI that we would embed then into our client-facing systems.

Fuller: How is this manifesting itself inside the CIO organization and, specifically, in terms of software engineering productivity? We’re hearing reports from a lot of companies of very, very significant improvements in the productivity of their senior software engineers. And we’re also seeing a, we don’t think, coincidental reduction in the number of job postings for entry-level software engineers. Does that track with your experience?

Tandon: We deployed GitHub Copilot last summer, and I’d say we we’ve seen about an overall 10–15 percent productivity improvement, but across the life cycle of the software development life cycle. But if you look specifically at coding time, about a 27 percent improvement in coding time. So I’d say, yes, we are beginning to see the productivity improvements from technologies like GitHub Copilot. We are expanding it to focus more on the end-to-end software development life cycle. So I think, as we scale, that we’ll probably see even better results. I’ve heard people talk about 40, 50 percent productivity improvements. We’re not there yet. Can we get there? Potentially. But look, I think, just when you talk about the impact on IT, doing software development faster is, of course, one aspect of it. But it impacts a lot of other areas as well. Like, we are looking at a full software development stack. So we have a cloud-native development stack, but now you need to think about how do how does agentic AI and LLM orchestration fit into that stack? So we’re working on improving that software development stack to be more AI-native now. We have to look at data strategy and data governance, ethics, and risk management, and, with the introduction of AI, also the entire data stack. How do you get better AI-ready data and then how do you govern and manage that data given the implications? That’s a big focus area for us.

Fuller: Is it changing your day-to-day job materially?

Tandon: Yeah. So I think, if I look at two aspects of my job, one aspect of my job is helping our business provide different shared capabilities to our clients using technology. So I think CIOs have moved from service providers to more business partners, I would say, a long time ago. At Vanguard, I’m part of our leadership team working with our CEO. And I work pretty closely with our business leaders on how to use technology. I think AI is going to require a big cultural and transformational shift across the company. And we as a leadership team understand that. The second part of my job is providing the systems that enable our business processes, and for the reasons I just outlined, your entire software development stack is changing. How you manage data is going to change. Talent, the people you hire, what skills are going to be required in the in the future—I think significant implications across both the demand and supply side of a CIO’s job.

Fuller: Let’s double-click on the talent question. The No. 1 open job position in terms of absolute numbers in the United States for jobs that include the word “engineer” in the job description is for “prompt engineers” right now. Where have you sourced talent for Vanguard historically? And is the need to bolster AI skills causing you to look at other geographies? Where’s the talent, now and in the future, going to come from?

Tandon: Sure. So we have we have presence in U.S., Australia, London, Canada—business operations and business offices in those locations and then also in India. But in India, it’s through our sourcing partners, and it’s mainly IT. We recently, last year, announced a new office in Hyderabad, India, because exactly partly to address what you were talking about, Joe, which is, if you look at the talent demand and supply over the course of the next decade, India is the only market that offers you a net surplus. We want to hire our own talent and develop our own talent in that market because we think it’s going to be critical. Also, in 2020, this is before the Gen AI boom, we announced a partnership with the University of Toronto to open an AI center of excellence in Toronto, and that’s something we’re going to be looking at scaling going forward. So, to answer your question, yes, we do have specific locations that are focused on specific skills. Like, in Dallas, we built out adviser technology; in Toronto, AI technology. India is going to be a global capability center. So AI is going to be one aspect. Mobile development is going to be another skill and capability we’re going to look to develop there.

Fuller: And how do you find coordinating global operations? Are you are you giving each geography a Toronto-like specific responsibility? Or are you having teams where the sun never sets on them and there are lots of handoffs there? There are a lot of different philosophies about how to manage a dispersed technology group.

Tandon: Yeah. So, we have, we have distributed delivery, even if you look at our various locations in the U.S., itself. So we have Malvern, Charlotte, Dallas, just as three examples. And we have teams who are spread across those, which leads to distributed delivery. When it comes to global operating models, if you look at our investments and if you could look at technology operations, those two critical functions, we have a follow-the-sun model across the globe. And if you look at just software development, that we’ll have distributed teams, you know in India, in U.K., in Australia, in Canada. We like for our sites to have an identity, but if I take an example of Dallas, we started off as an advice hub. Or Charlotte, which is another site that started off as a personal investor or retail hub. But now you’ve got various capabilities, including information security, personal adviser services, FAS [financial adviser services]. So it may skew towards, you know one capability, but just the availability of talent or the diverse availability of talent in the area has allowed us to use it for more than the intended or the initial purpose.

Fuller: You have an office called the “Chief Data and Analytics Office.” What does it do, and how does it relate to your organization?

Tandon: I mentioned one of our pillars in the technology strategy was insights. And four years ago, Joe, we were spending a lot of money on data and analytics, but we didn’t think we were getting the value we liked out of that investment. And that was when data and analytics was distributed across each division across the company. So what we did is consolidated data analytics functions in teams that were sitting across the company into a CDAO or chief data analytics organization. We hired a CDAO, Ryan Swann, and we incubated that within IT—the idea being we offer our crew a better career path, we prevent duplication of capabilities. We put more structure in driving value out of data. And the results we’ve seen have been pretty impressive. That organization has been tracking, in partnership with the finance business, the value that our efforts have generated. And over the last three years, they’ve generated $450 million in value through the use of data analytics and AI/ML.

Fuller: Does its ambit also run into analyzing the internal operations of Vanguard, such as what kind of skills do we have, and what kind of skills do we need? Or is it pretty much focused on process improvement and customer success?

Tandon: As it relates to data analytics skills, yes, like they have helped us, you know, improve our rigor and discipline on data engineering, data analysis. You know, machine learning, engineering, data scientists, et cetera. But if you’re talking more broadly skills analysis across the company, that is more a function of our HR team. What a CDAO will focus on is enabling things like hyper-personalization for clients. So let’s say you’re going on a Vanguard website. One of the areas we like to differentiate ourselves, Joe, is just through our mission. We are very client focused, and the sole purpose of Vanguard’s existence is to maximize individuals in investment returns. So while we want our client experience to be just as intuitive as anybody else’s, we are constantly focused on, “How can we make Joe’s returns better?” So if you have money sitting in cash, which you haven’t invested, we’d want to remind you. If there’s an opportunity when markets are down to do tax-loss harvesting, we want to nudge you. So you know the data analytics organization works for those application teams or the business teams to really identify opportunities like that which can drive better investment outcomes for clients.

Fuller: Well, I certainly don’t want anything to stand in the way of maximizing returns for Joe. As we conclude here, how do you think your function—and particularly your talent base—is going to evolve? If you go back to, imagine when you’re working at Deloitte in 2015, it would have been hard to imagine the type of change that’s occurred between 2015–2025. As you look forward, what do you what can we expect? What do you anticipate are going to be challenges and opportunities between now and 2035?

Tandon: It’s stating the obvious, but AI is going to be a big influencer in any future business strategy or technology strategy, needless to say. And I expect everybody by 2035 to be far more digitally savvy and far more AI proficient. But I think a couple more skills or focus areas are going to be important as well. One of them, which we are focusing on right now, is digital product management. So as companies work in more new ways and get further along their cloud transformation journeys, digital product managers are defining what to build. They are the ones who are defining the experience, aligning teams to new innovations. So that is a skill—product management and digital product management is a skill we at Vanguard are trying to get much better at. But I see it as a key focus going forward. And the second is, to our conversation on change and transformation management, I think the degree of change you will see driven by AI is going to lead to a lot more focus on change and transformation management. I see a hybrid world in which humans and machines coexist. I see there being, you know, lesser layers between data and decision makers. I see, for Vanguard, I see us having a more global workforce as we look to scale our India office as well as drive the Vanguard effect more internationally. Those are some other things I would predict for 2035.

Fuller: Well, you make a lot of change sound like something that is well under control and something that that will lead to a lot of benefit. Nitin Tandon, CIO of financial services giant Vanguard, thank you so much for joining us.

Tandon: My pleasure. Thank you for having me, Joe.

Fuller: We hope you enjoy the Managing the Future of Work podcast. If you haven’t already, please subscribe and rate the show wherever you get your podcasts. You can find out more about the Managing the Future of Work Project at our website hbs.edu/managingthefutureofwork. While you’re there, sign up for our newsletter.