Building the AI-Powered Workforce of the Future with Ganesh Ramakrishnan | CAIO Connect Podcast

If at all AI gives you a much bigger canvas to dream big and once again a businesses especially as I see many many leaders struggling with what AI can mean to them because they get very confusing signals from the market about AI and it's the CIO's role to translate AI into business impact which is relevant to their business. So one has to be very careful. This is a time when there is a great ferment and turmoil and in the next 5 to 10 years you're going to see many many large story names fall and bite the dust and new ones coming and finding their place. So if you think about candidates today, they come to our website, they apply for jobs. But what we're trying to do is instead of just taking in an application and telling them we'll get back to you, we will have a virtual AI agent which is a coach, a candidate coach or a talent agent, a talent agent which is working on behalf of the candidates. And believe me, in every single industry, there are gamechanging stuff. We are only limited by our imagination and our courage to step forward. We are always restricted or restrained because of the fear of failure. Now AI represents such distinct difference in technology evolution and development that unless you understand it yourself. So I always say the leader has to devote a lot of time to understanding the mechanics.

Thank you so much Sanjay. It's a pleasure to be on your show.

Yeah, that's a great question. So I think as you can see, I have a very technical background as an engineer but I rapidly forked out and did an MBA from the Indian Institute of Management in Bangalore which again is a very prestigious organization very similar to the IIT. After that I went into technology management which is a little different from core technology or detailed coding and such like. What I immediately got was very rich experience in banking. I joined an American bank, a Wall Street bank called Citibank, which people in the US would know very well, and I spent the next nearly 25 years in Citi not in just one role but in multiple roles spanning technology, operations, transformation, large program management and also setting up new businesses in Eastern Europe, Northern Africa and in fact in Russia too. So it was a wide variety of experience which shaped me and gave me a sense of how operations and technology can enable businesses. And so I got to really understand the strategic significance of these areas in building a business which normally you think are hygiene factors and things which everybody needs to do but if done well and if done with the right talent and strategically it can be a very important competitive lever. So that's the important thing I understood. I'll give you one or two pivotal moments since you asked. One was that when we were setting up our infrastructure in Eastern Europe, we were setting up new businesses. The cookie cutter approach that we employed, which was we had all the components which we were able to put together to build a financial institution from scratch and launch a full-fledged business within one year, it was nothing short of amazing. And the reason we could do it is because we had all the components and we had all the talent and we could bring it in and build up a full institution. And that really taught me if you have all the underlying components and it's well defined, your technology stack is well defined and your business model which accompanies that is well defined, you can really put together something very quickly and run it very professionally. So that was a very important moment for me and that taught me the significance of these underlying disciplines.

Yeah, great question. I think there are two key levers here. One is a recognition that it is not just a hygiene factor or table stakes but it's a very critical component of competitive advantage and of enabling business. So unlike in the past, it becomes an extremely critical component for you to power your business and that is a realization that CEOs have had and the boards have had and that is why the CIO position has been elevated and often now reports into the CEO and is part of the management team. The best way for the CIO themselves to think of their role is enabling and developing business capabilities. What I mean by business capability is a mix of people, process and technology and there is no one better than the CIO to bring it all together. It needs a mix of business partnership, process partnership and technology partnership to put together very tightly and have a business capability. And these kind of mix of business capabilities is what defines a company or a firm. So if you're able to put this really well together, you can get massive competitive advantage. And most companies today don't realize this well enough. In the leadership team they think of their business distinctly differently from the process and technology and that's a problem. So if you bring this together and as I said the CIO is in a very unique position to have that higher order thinking and to demonstrate that with their leadership they can bring these disparate disciplines together and create a very powerful fusion. If you do this then you're naturally elevated to the C-suite because that's where this kind of thinking belongs and you're able to shape something very distinctive. In the age of AI when the power of technology is now multiplied many fold, the kind of business capabilities you can think of is distinctly different. It's not only incremental. Normally what happens you make incremental improvements, you do something a little faster, a little better, but today you can do it completely differently. You can redefine the way you do it and often times you can even redefine the kind of business you are in. If you have a broad enough definition, you can think of managing your overall mission in very distinctively different ways because AI allows you to do that. So if at all AI gives you a much bigger canvas to dream big and once again a businesses especially as I see many many leaders struggling with what AI can mean to them because they get very confusing signals from the market about AI and it's the CIO's role to translate AI into business impact which is relevant to their business. There's no point in talking about LLMs and so on if you can't translate it into what it can do for you in business capabilities. So that's what I would say defines the role of the CIO.

Absolutely. The role of the CIO and the reason why the CIO should be in the C-suite is because they play a transformation role and they're able to think about the business strategically over a number of years and not about getting technology working. So that's the hygiene. They need to have technology working. They need to make the changes. They need to add on to new features and so on. That doesn't change. But the fact that you're able to redefine, reshape the value chain and think of it distinctly differently over time is really the role of the CIO and they have to bring together not just the technology expertise even the AI expertise but also bring together other strands of the business to find out how AI and technology can impact them. For example, every one of our businesses has an engagement with the customer or the client. The way we engage with clients can be fundamentally redefined through AI. The way it can be done needs to be exemplified by the CIO. They are the best positioned to be able to say this is the art of the possible. I always use the word art of the possible. These are all possible because AI gives you the power and we might make a few mistakes but that should not prevent us from dreaming big and launching very courageously and boldly. If the CIO does not show that courage and that ambition even aggression then the rest of the C-suite will just go to the lowest common denominator. They're not going to advance and progress and then you're left to competitors who could do the same thing. Even worse, there are many startups who are redefining many many businesses, especially knowledge working businesses are under attack today. The fact that you're very large and entrenched does not prevent you from being upended by much smaller and AI native startups. So one has to be very careful. This is a time when there is a great ferment and turmoil and in the next 5 to 10 years you're going to see many many large story names fall and bite the dust and new ones coming and finding their place with very distinctively different business models but kind of fulfilling the same mission. So unless you're paranoid and the CIO has a very important responsibility in this, you will not survive and that's why it's no longer a nice thing to do. It's a must.

So it's a plus plus. I think it's very useful for them to have a good understanding of the potential that all of these new technologies offer. So when you think about cloud as an example, cloud is a highly technical area and you can go very deep technically, but it's much more important for the CIO to understand strategically what does a cloud offer businesses. The way they need to think about the cloud is it allows them to build the organization on a pay-per-use model. Unlike in the past where you had to make significant large investments and build the organization, here you could start small and you could expand as demand increases. That is fundamentally what cloud allows you to do. The second thing it allows you to do is to build the entire business logic in a very different way from the earlier paradigm where you had to buy or build monolithic software, bring it in. You could actually take smaller pieces and build your entire business value chain. So this is what cloud allows you to have. The reason I say this is that the CIO needs to know the true significance of what these new technologies can do to their organization. So it's a strategic bent of mind on what technology means to their business model. That is what they need to realize really well. They need to have a happy blend of some technical knowledge, the potential to ask the right questions to their teams and build it up, but have a very keen understanding of what these means strategically and have a mind map to say that these technologies are going to power my businesses in a distinctive way. You have a choice of multiple technologies and you need to be very clear about what is the path for your business and that is the decision that you take in the C-suite with the rest of the leaders and then you pick and choose your model and build up your technology or digital core so that on that sound digital core you could then build up the layers which really are competitively differentiating. So we know data is important, analytics is important and unless you have a core digital core which builds out the right kind of data, clean data on top of which you can have analytics, on top of which you can have machine learning and then on top of which you can have AI, you need to have the realization that you need to have a strong foundation and based on that foundation you can build it up. All of these are fairly technical discussions which some of the C-suite teams may not understand or appreciate. It's very important for the CIO to translate this into business terms and to make a compelling argument why certain investments are essential, especially what I call foundational investments which are very difficult to justify. People can understand if you're enabling a business capability which is out in the market but how do you justify cloud investments? They are infrastructural in nature. The ability to translate that into business speak, into business advantage, into competitive advantage is a very important skill for the CIO. It's not enough if they just know it. They need to be able to bridge that link or make that link between technology and business strategy.

So we have positioned really well because we do a lot of knowledge work. So we use a lot of information about talent. We also have information about the opportunities and what clients want. Our business is to match the right talent to the right opportunity and we need to do it at scale and we need to do it at the lowest cost. So this is a core business model. If you think about AI, AI is able to do cognitive work unlike in the past technologies which can't do cognitive work. AI is able to do cognitive work and it is starting off with lower cognitive work and then it's going to keep on going up the ladder. So when we think about understanding the profile of millions and millions of résumés which represent talent and look at thousands of opportunities which clients entrust to us, the ability to understand these two and then find out the right match is something that AI can help significantly. Let me give you some practical examples. What does a recruiter do? They do many tasks in trying to take a requirement from a client and put it out on different channels and allow talent to look at it. So they rewrite the job description. They kind of seek out different candidates. So we have AI tools which are able to understand at a very high level what a client wants and translate that into very specific language in multiple languages not just in one language and then throw it out into the web. So this ability to very quickly translate client needs into digital channels is what AI is enabling us to do automatically. The second which is very ambitious is the ability to take a resume which is written in multiple different formats. Every candidate writes their own resume in different formats and understand really what are their core skills and break it down into a standard format. In the way that it's broken down into a format, it allows machine learning algorithms to act on it. If you think about it, we have a history. We have a 75 year old history of matching talent opportunities. So we have very good history and we know what a successful match looks like. So if you think about machine learning algorithms based on this history, they're able to learn really well and once they've learned it for certain domains they're able to apply it for new skills. So if you think about software development as an example, we are one of the largest providers of IT talent in the world. If you think about the past 20 years and we look at our best recruiters who fill opportunities the best way, we take a look at their history, feed it into a machine learning algorithm and then it is able to suggest on its own the right kind of candidates for every new IT role and it helps our recruiters get rid of the grunt work of going through hundreds and hundreds of résumés and searching. So we have now deployed machine learning algorithms which only get better with time. The reason we are able to do it is because we have a lot of data. Data is extremely important. It's not something a startup can do even if they have machine learning engineers and smart people. They need a lot of data and so we are well positioned as a large organization to leverage that. So this is just another example and there is a third relating to pricing. If you think about the labor market, the labor market is all about supply and demand and it changes from one region to another even within the same country. To be able to understand those exact price points based on the dynamic nature of demand and supply and then price our talent appropriately is a third one and that also uses machine learning models. So I'm giving you a couple of examples but I'll give you a fourth example which is intimacy with candidates. So if you think about candidates today, they come to our website, they apply for jobs. But what we're trying to do is instead of just taking in an application and telling them we'll get back to you, we will have a virtual AI agent which is a coach, a candidate coach or a talent agent. A talent agent which is working on behalf of the candidates. How nice it would be for a candidate to say I got somebody working for me for my own interests and trying to find the right job for me and giving me advice about what skills I need to do and even counseling me if I'm on a low and helping me get the next role. So we're trying to do that and the large language models of today are an excellent fit for this kind of role. So some examples on how we can revolutionize this area through AI.

Exactly. So what AI is able to do is personalize the engagement and interaction. So it is an agent which is dedicated to you based on your personality, based on your skill sets and based on your goals and it provides you fairly personalized advice. So as an example, you are a warehouse supervisor. But did you know that because of the skills that you have in terms of talking to people, you could get into a customer service agent or you could get into some adjacent role that you were never thinking of but with a little bit of training you could get into that. So this is what the personalized one-to-one agent can advise you on your career path and based on your skill sets, your likes and dislikes, it can do it and then serve up the jobs that work for you. So this is what we're thinking of and in my view it's truly exciting because today if you look at the workforce of today, they really are craving for something authentic, something personal and searching for a job is really exhausting, can be demoralizing because you get a door shut on you every day. So having somebody who has your back is something which is morally uplifting and it kind of works for the candidate practically as well. So this is an idea we are thinking about now. You haven't rolled that out yet, it's something you're working on right?

Exactly. And our view Sanjay clearly is that exactly as you said AI augments human beings. So you have the human at the end of it, they provide even more personalized and expert advice but when they are not available their AI personas are available. So they kind of bridge the gap in terms of timing, in terms of focus and so on and they serve up a lot of stuff for actual human beings. So when it comes to the final job, they'll be talking to a human being. They'll be talking to a recruiter. So recruiters cannot be available all through the day at every point of time for all the candidates. So we think that if played really well, they can be powerfully augmenting humans and we believe that a human in the loop is very critical. But the rest of the loop is created with AI based enablers and technologies. So this happy coexistence between the two I think can be a winner. And I would invite all listeners, my colleagues and other people in the industry, to really think about gamechanging areas in their own industries. And believe me in every single industry there are gamechanging stuff. We are only limited by our imagination and our courage to step forward. We are always restricted or restrained because of the fear of failure because when you do something dramatic with a new platform or technology like AI, they're bound to be challenges. They're bound to be failures. It's never going to be smooth. But if you don't start early and this is a very important message that I have and this is just my practical experience. Your early steps are never going to be easy. It's not going to be uniform. You will have more failures than successes. But through that you learn personally and organizationally. You build muscles. AI technology keeps changing fast. What was not possible today in six months is possible. But you're thinking about it. You've tried many ways. So you're on top of it and that's what allows you to progress as rapidly as a technology. And when there are disruptors who come after you, you're well positioned because you built those muscles. So that's why I say there is no better time to start than now. Even if technology keeps changing, doesn't matter. Start now. Work with it. Don't bother about failures. Do your POCs, do small experiments, but dream big. Have a very clear focus on what you want to do and don't dream small, don't think small is what I would say.

So let me confess that the general experience of large companies is that there was tremendous excitement when the LLMs came and everybody started dabbling with it. There were hundreds of experiments all around the organization but that did not have a material impact on the bottom line. Everybody was doing something. People were doing powerpoints and transcription of calls and writing essays and writing emails and so on. That's all nice. It's good to be a little better than before. But they didn't have material impact. So what I would say is that focus on four or five big themes. Find out those four or five big themes that you can hunker down, double down on. These would be thematic. As an example, in our industry, that candidate journey is a big theme. Revolutionizing the candidate journey is a big deal. And you won't be able to solve all of it on day one. You have a dream, you take it step by step, but you know that over a couple of years, you're well set to solve it. In every one of the industries, I'm sure C-suite people and CIOs are able to identify those four, five, six areas they want to really double down. So identify, ensure that there's good business alignment and agreement and the other business, sales, marketing and the functions and the brands, they agree that these are the ones to think about. Definitely it should be led from the CEO. If you agree on that and the organization comes behind it then you can create something truly groundbreaking. If you don't do it because people are skeptical or the CEO says it's not a big deal then they'll find to their discomfort that years down the line they will get taken over or superseded by competition. So this is the time to do it and there's no choice at all.

So I think it's been oft repeated that data is everything and I'm sure all CIOs know that. I would expect even before the AI revolution we have had the data revolution so this is not new at all. So they should be well on their journey in ensuring that they have reasonably good data. Now in the old days the only way to get the data was to bring it all together in a central place, construct a data warehouse, clean it, have very structured data and then utilize it. Thankfully today that's no longer the case. You could have distributed data. You could have data which is unstructured. You could have data which is not very clean. But now we have technologies to build that into a data fabric. We have technologies which can go and query and interrogate the data and understand it from where it is without going into lengthy cleanups. So all those are there today. So we are actually very fortunate. So we should not waste time. At the very least CIOs should know where their data is situated and what the data means. That's extremely important. They should have an inventory of the key data they have. And then there are technologies which can build the fabric and build on top of it. So not a moment to waste because data is the key and your data is your competitive advantage. Especially if you're a large organization which has scale which has been around for a long time. The one thing that you have which startups don't have is data. So better hang on to it. I would say be very careful about sharing the data with outside. There are lots of people who will come knocking on your doors wanting to provide you technology but sometimes when they provide technology they take out the data and utilize it. It doesn't mean that they take the data and publish it elsewhere but they take the intelligence behind the data, they build models on that and the models are then utilized with other customers, other places, even competitors. So it's very crucial that you just don't just own the data but also the intelligence that comes out of the data.

Great question. So the funny thing is that with AI coming in, I think what is becoming more and more important are critical thinking skills rather than deep domain, deep knowledge of techniques, of languages, of libraries, of deep AI algorithms because AI is able to substitute for that but what you need is critical thinking skills and a good appreciation of what is a possibility of what AI has to offer. I'll give you a good example. We have heard about agentic AI and agentic AI is a way which can take processes and kind of automate them. Even six months before you needed a lot of skill sets to write an agent, to develop an agent. That's no longer true. You need to know the understanding of the process. Agents can be now built by AI itself. Any of the current large language models are able to build fairly good agents and you need just higher order skills to be able to test the agent, ensure that it's ethical and then deploy them. But you need to have the appreciation of what agents can do. So what I would say is that the culture that we need in our teams and let me tell you that you would think that technology teams are going to be the best and jump into AI headlong. That's not true. They are also very traditional. They are also orthodox. If somebody has been in a particular way managing a particular application for 20 years, they would tend to do it the same way even if much better techniques were there. So the ability to rethink how technology is developed, maintained, deployed is a cultural change that needs to be done and the cultural change has to be led by the people in the front. The leaders need to do it. I spend a lot of time going back to college days where I learn about all of these and I learn to apply on how the software development life cycle can be animated by AI and so on so that I can demonstrate to the teams and then they take heart and they take courage to flip and change. So critical thinking skills are critical. The other part is seeding AI related skill sets, higher order skill sets within the team is important. AI skill sets are not easy to find. They're very expensive and difficult to retain. So what is important is that to be able to attract that you need to be able to demonstrate what you can achieve with this. The amount of data you have and the ambition that you have is what brings people to you. If you're able to demonstrate it, we think in Manpower Group, we have both. We have massive amounts of data. We put between two and three million people to work every year. We have more than 15 million résumés in our database and we have the ability to understand the performance of the people we put to work. So we have tremendous amount of data and we have the ambition to reshape the industry. That's what brings talent because they look at this and say this is what I would not get in a startup because I won't get the data. So I think that's very key to set your stall appropriately.

Another very great question. If you think about it like you said, we are at the center of this because we are one of the world's largest private sector employers. Like I said, maybe 2.8 to 3 million people we hire every year. We provide employment and doing it in a fair, ethical, explainable manner is really really important. No matter what AI systems we make, we should be able to demonstrate not just tell it ourselves, be able to demonstrate that those decisions that even these models make are ethical, they're explainable and there's no bias. So it's very important that as we build the models, we keep this by design. So you don't just build it and then test it for bias later. You ensure that as part of your development process, as part of building it, this is built by default, built by design. At every point when you're developing models, when you're developing large language models to analyze, you always keep this at the forefront of your mind. Then we have a set of ethical rules which govern everything that we do. One part is our kind of ethical standards. But the other part is also the technical standards. So we look at what are the technical protocols, the red teaming, the compliance related items that we have. So we have put together what we call a risk framework for AI development. No matter we build or buy, everything has to go through the same risk framework. So in a way we have centralized our standards and governance but we have decentralized the development. So people can develop AI applications across the organization but every one of them has to go through the same governance standards and principles. So this way we are able to manage this in a very intuitive manner.

Yeah, I can understand the difficulty because this is a very different way of doing things but I would suggest to all of them that the current set of business metrics, SLAs, KPIs and financial metrics, whether it's AI or non-AI technology, the way of measuring it is not very different either. It has to do the job at a lower cost or it has to contribute to topline growth or it has to ensure that the experience is much better or it has to manage controls and manage risk much better. So these are very critical components for any business and to the extent you're able to impact one or more of them positively you have the ability to tell a story. The key thing is that you need to be able to have very clear metrics. So when you start out an AI program or AI project, success is not defined by putting in the AI technology and making it work. Success is defined by achieving a business outcome. So the project should be not AI. It should be what is a business outcome and incidentally enabled by AI. So it's no different from a non-AI program. It's the same. So focus on outcomes. Don't focus on the underlying AI at this point of time would be my advice.

First is women have been traditionally underrepresented in technology. I think it's unconscious bias unintentional. Women also self-select themselves sometimes out of STEM related disciplines and so on. But even more so than before Sanjay, AI systems are being made in the image of the people who develop it. That's very important to know. They are highly flexible in the way they are designed and that can lead to a very different consequence. Having diversity of thinking is very critical. So having a broad set of thinking beyond very narrow technical thinking is very important when you're developing AI. Because of the power of the platform, this is going to reflect on human decisions. For the first time we are encoding a lot of human decisions into AI and the way people think is what is being reflected in AI platforms. So that's why you need diversity of thought and you don't have a monocultural thinking which just puts you in a particular direction without you even intending it. So even more so than before we need to have diversity.

So if you think about 10, 15 years ago, it appeared that once you reached a certain level of seniority you were leading teams, you didn't need to be in touch with technical advances constantly, updating yourself because your teams would do it and you had to only think about strategy, C-suite dynamics and so on. That kind of thinking has kind of met its end because of AI. Now AI represents such distinct difference in technology evolution and development that unless you understand it yourself. So I always say the leader has to devote a lot of time to understanding the mechanics and it's not enough to have just a peripheral view, a topline understanding which a CEO can have. CEOs need not be technical. The CIO has a duty to actually get into the details to understand how this entire technology is built up and that'll also allow you to know what are the limitations because we shouldn't confuse AI into thinking it's like some superpower. Why do AI platforms hallucinate? Why would they go wrong? Why are they not good at numbers? These are things that you would instinctively know if you understood how transformer architecture is, how LLMs are and how each evolution of LLMs try to solve some of the problems but they don't solve it fully and real reasoning and thinking still eludes even the best LLMs today. Why is that? So if you understand these limitations from ground up, you'll be very thoughtful in designing it. So when vendors come knocking on your door and claiming all kinds of things, you're able to challenge them appropriately. You're able to in the short term think of the right way to negotiate the path while having very ambitious long-term goals. So I think that power liberates you, sets you free and is able to kind of guide the organization's path in a very pragmatic manner.

Yeah, I mean we all know that AI is going to fundamentally change the nature of human work and human labor. We know that certain industries are going to be impacted a lot more than others. But we believe and I also personally believe like every other fundamental technology advancement in the past, AI has often been compared to electricity because it is a multifunctional technology. It's not just for one area. It impacts every area like electricity does. Even when that happened, we found new roles, new jobs coming up, old ones dying and new ones coming up. So I would suggest that as some areas and some industries get impacted, the people involved there learn new skills. They get empowered by AI, they get augmented by AI and it's much easier to learn a new skill when you're augmented by AI than before and face the future with confidence and not fear. So as things go and we get a radical restructuring of the labor market, you'll have as many new roles as the old ones that get destroyed. So it kind of has a way of settling down and these new roles are available for knowledge workers. But underlying all of this I would think is one: develop your critical thinking skills and develop learning as a lifelong activity. Every single day one has to learn. I really mean it. You cannot just like you go to the gym every day and it's like you need to exercise every day or at least three or four times a week. You need to learn every day or multiple times a week. So if you keep up to it at every level, whether it's a junior person, senior person, doesn't matter, they can all learn, everybody can learn. I think you'll be very well suited for the future. You can face the future with confidence.

Yeah, I think agentic AI which is just at its infancy now is going to reshape fundamentally how multiple industries work. Because agents will have the ability to do back office activities, transactional activities, but also high value added activities. They can also interact with customers, people who consume technology. So it has the ability to spread, it has the ability to interact with our physical world through robots as well. So I think agentic AI is at its infancy now and it'll see a rapid development over the next years. So watch the space.

Sure, go for it.

Cloud.

Personally Python but I'm sure many people would have their preferences.

Build. And I say that because of what I said earlier about don't lose your data, keep it with you.

Actually each one has its pluses. I mean there's no need to choose one over the other because each one has its strengths based on the context. So I would just say pick one based on context.

For the most part open source because I fear the proprietary solutions are going to have very short shelf life.

So my preference would be MCP because I think it gives you the ability to interact with the world at large. But I can understand that agent to agent communication is also going to happen. So it's not that one is going to displace the other. There's a place for both but between the two, MCP.

Thank you so much for having me on your show. Thank you.