Good afternoon, everyone. I'm Anita Mann, chief strategy and corporate development officer at EXL.Thank you all for being here, and thank you to the conference organizers, and thank you to my colleagues who helped me prepare. So my work sits at the intersection of strategy, investments, and execution focused on helping our clients to move from AI ambition to measurable business outcomes. And over the course of my career, I have worked across telecom software, health care data and analytics, and now AI transformation. Throughout, you know, I've seen that technology creates the most impact when it is combined with human judgment, curiosity, and truly inclusive leadership. So I find it, very gratifying to be part of this conference because AI is changing not just how economies and businesses operate, but also how careers are built.

And for women in technology, this is a defining moment to lead the change because AI is never is no longer, at some distance on the horizon. It's already reshaping work and business. So let me, get into it now to share with you a little bit of what I am seeing. So first, we'll, commiserate a little bit about the overload, the cognitive overload that AI is creating, on all of us. And, we'll talk about, like, why it feels so challenging, but then try to finish with some hope, for the future. Alright. So AI overload. On top of everything else that everyone is dealing with in today's environment, we also have the challenge of true cognitive overload that has kind of become de rigueur in the AI era. And let's talk about, like, why is that? In in what we're dealing with, what are seemingly simple questions, are really not. They're very complex.

So you can, think about is there a bubble going on? Is AI actually ready for the kinds of things that I need to do? I don't know who I can listen to. And, you know, why does it feel so difficult to get answers? Well, I think that AI, as we're all coming to realize, it it's quite a bit more than a technology. We're not talking about a technology shift. Some people compare it to the Internet. I really don't think that's the best comparison because there's not really a networked value coming from AI right now. It's even more fundamental in my mind. It's as fundamental as electricity being widespreadly you know, being, available on a widespread basis or, maybe more like the shift when desktop computing became available at an affordable cost.

So it is a real paradigm shift affecting all aspects of business and commerce. So you would be forgiven, though, from being confused given, the headlines that we're all bombarded with, every day. You know, one thing people are still asking, is there a bubble? There are some characteristics of, a bubble being exhibited in the AI market, maybe among some speculative investments or early stage startups. But, you know, most AI players, the larger players, truly are profitable. Many investments are going into measurable tangible infrastructure. As, you've probably heard others say by now, there are no dark data centers. Everything getting built out is being used.

And I think that makes this very different than what was experienced in the .com era, which I did experience in my career. So I do remember back to that. But it is also true that this technology is moving at breakneck speed. It has truly been a whirlwind. So clock code showed us that AI can write, self test, and fix software. And then OpenCLAW came out, and you saw that AI can operate your computer for you. And that became a little bit appealing. Can the can the AI do some of the the work I don't like to spend my time on every day? But you have to be careful because you don't wanna become one of those headlines. And but new things are coming out all the time. Last month, Anthropic dropped a study saying that AI, its AI, is acting emotionally. No. Really.

Claude, stores emotional states, and it responds differently based on the emotions that it detects in the prompting. So if you push it to the point of desperation, you may be surprised with what it will do. But even more exciting from my perspective, last week, OpenAI really upped the game with a new voice model that is actually speech to speech. So they're not taking speech, translating it to text, operating on the text, and translating the response back to speech, But the model is operating on speech. So this gives us lower latency, GPT five, class reasoning, and it will be realistic and reliable for creating apps that can really talk to you. So you can see, you know, even beyond customer service type applications in enterprises like the clients that we serve, but also could be game changing in education, media, events, and and other ways.

So, also, who do you listen to? There are this map is only from 2023, and there were already nearly, 10,000 startups out there. The funding has just continued to flow into any AI startup. And on top of that, all, proven providers like EXL and all of our peers have a distinct point of view and capability set on AI. And undoubtedly, all the senior technologists in your organizations have points of view. So it it is becoming more and more difficult to try to sort through the noise and find the signal. So how does this, turn into an experience for you working in a large enterprise?

I'll just say when it comes to what you're experiencing, I'm sure there are many others, going through the same. So think about where you sit today, in your organization. If you are in a group that's been around for just a few years and has maybe a thousand or more employees, chances are good that you're facing something like this. So the strategy may not be clear and may be working against itself in in some ways, undoubtedly of legacy platforms that are controlling a lot of the business logic today and need to be harnessed and marshaled. We'll talk a bit more about data in a minute. The pace of change is not easy to keep up with, certainly not easy for a distributed leadership team to manage. And it can feel like it's, overwhelming at times, but it is not as we just talked about. It's not going away. It's getting stronger and more and more helpful.

So the only way to get through something like this is to go through it, and we're we're not going to go back to the past. One thing I do know about technology and business trends is they don't go backwards. And then I'll just share with you a a lovely quote I heard. AI is compressing labor faster than organizations can compress bureaucracy. So I thought that was a good one. Alright. So what are these challenges? The ones I'd like to talk about today, and I'm sure you may have more, and I'd love to hear more about what you find is getting in the way of scaled AI adoption in your organization. So please do feel free to share your views in the chat as to whether these are the things you're facing or there's something else that's giving you trouble. So first, misaligned strategies is, what we'll talk about. I I mentioned this on the previous slide. But also technology gaps.

As advanced as AI is, it does not come ready to plug right in to your, enterprise architecture. There's, quite a bit that it takes to get value from it. And the talent and the organizational hurdles are really they're almost two different things. One, certainly, there needs to be, a big upscaling of the talent, but then how you can help people to work together to find the value in AI is a whole other story. And then the cost, can be high, and the costs of mistakes or misguided projects are extremely high. So we'll talk about that some more as well. Okay. So starting with strategy. I have, great deal of background in strategy for different types of organizations, organizations that are medium size and large, also as as a part of some very large corporations.

So but I will say, in general, you know, over the past twenty, thirty years, strategy has been, you know, something of a centralized and then cascading, kind of a function. So what we what we saw in the past was there would be some corporate strategy that then gets communicated and cascaded through the operational teams who do their own team planning and then the finance teams that do their own planning. And then at each step along the way, there's a somewhat, local localization of the strategy that happens. And when it you're in the AI world, this is, not as effective because technology won't know what the business can take advantage of, and the business wants to plan things not knowing whether the cost of implementing them is actually going to be feasible. So while we're, you know, going through this phase, this kind of a strategy process doesn't really work in my opinion. I really think we're going to need to get on to a more coordinated approach where multidisciplinary leadership teams can, appreciate the value of working together, sharing and deepening their insights, figuring out how to reimagine something in the business, test it out, assess it, see if you're delighting your clients as you thought you were, you were going to, and then consistently assess progress and innovate some more and and try the next thing.

And the thing is none of these can kinda sit behind a a six month or a full year strategy cycle. These things and luckily with the AI technology, you can. You can actually do these in four week or three month cycles and and get things into the market very quickly so you can learn and, share and do it again. So that's something I wanna talk a little bit more about around how your business strategy is going to be impacted and and your strategy process. And one of the reasons I think this is important is as you look at who's driving the adoption of AI in your enterprise, you know, in the beginning when the power was just becoming clear, operations leaders were very engaged and wanting to bring AI solutions into their operations.

But very quickly, senior leadership also recognized, you know, not only must we protect data, but they when it comes to training a model and embedding institutional knowledge in a language model, there's a proprietary protection aspect of of that. And you don't want to, you know, just let the whole company start sharing data with vendors and having no control or or rhyme nor reason as to what your data protection strategy is. So governance and compliance kinda stepped in and said we've got a whole host of risks we're managing, least of all protecting our, institutional knowledge and property. But then, of course, AI continued to progress and people said, alright, to their CIOs and their CTOs, you have to do something. And a lot of them, you know, adopted some kind of, copilot or teaming software. And the business leaders got frustrated and said, this isn't enough. I wanna do more. I wanna do something much more transformative.

And what we have seen in the last six six to twelve months, the AI transformation for most large enterprises is on the board agenda, which puts it on the CEO's goal forms. So when we're talking to clients now, it's often the CEO that we're talking to first, And they are taking on accountability for not only the planning of AI, but the overall impact on strategy and the longer term vision of the company. So I really think at this point, AI strategy has reached the level of becoming the business strategy, and it needs to be transformational. Okay. The next, set of challenges that I'll I'll talk about is around the technology itself. So if you look at what the power of these models is, they require much more dynamic data inputs. They're working on pattern recognition and very large actually, the largest frontier models now are up to a trillion parameters, very large models, continuously learning and adapting and able to predict what might happen next.

So this requires a whole different technology infrastructure behind it to be able to scale it scale it up. And one of the challenges in this is, as I was saying, any mature organization has a large, legacy infrastructure that they're managing that could be 30 to 40 years old in some cases. Sorry. So many are, still relying on mainframes using COBOL and spending a huge amount of their tech budgets on, legacy tech. On top of that, the governance approach that has been more centralized in command and control, is important. You know, we have to maintain and even, you know, increase the compliance function around maintaining compliance with regulatory. But the traditional approach to it, is viewed as being too slow and in some cases lacking for the AI era. And then third, the data itself.

So I'm sure everyone has heard, you know, the AI is only as good as the data that is going into it. Well, your data may be fragmented across, a thousand or more systems within your enterprise, and you'd be managing, huge volumes of unstructured data that you do want to tap into so you can begin to, categorize and graph the tacit knowledge in your organization. Five petabytes is like equivalent to 10,000,000,000,000 books. It's a huge amount of information. And then this is, from a survey across, organizations with at least 5,000 employees and on average, you know, spending $30,000,000 or more just governing the existing data estate. And you can imagine that's a much bigger number for a larger organization. So as we look at this, a very strong recommendation we can make that we've already executed with countless of our clients is to put tremendous focus and investment right now on creating the AI ready data that you need and and do it as a foundational investment.

In fact, we've been it's one area where, ironically, I guess, or or, in a very beneficial way, you can use AI to better manage the data that you're readying for AI. And that's what we've seen as we've been evolving our own, data platform here at EXL so that you can continuously monitor your data in motion and deliver improved accuracy and reduced manual effort. Okay. So let's get on to the more challenging topic, which is the talent and the organization. So, you know, you've probably heard if you work in strategy that culture eats strategy for lunch. Well, the deeper you look at this, I think the organization is gonna eat AI for dinner if we're not careful. There are you know, you can pretty much trust that the algorithms are going to continue to evolve.

And there are already you know, it may seem very expensive, but there are already, you know, various price points you can at which you can access, language models that are smaller or more purpose built or hosted locally. So not everything has to be, you know, the most expensive, most ideal frontier model. But that being said, the people and the organizational design is what some peep many organizations are now seeing as kind of the hidden challenge in trying to adopt AI and make the most of it. So, I mean, I think the siloed structures of the past just are not going to be able to keep up, and we need to be more agile and multidisciplinary. You have to start with, of course, reskilling people. Ironically, leaders tend to believe that the people are more AI literate than people feel confident that they are, and we have this lingering fear of job loss.

There's also confusion still. And what you need in the AI era more than anything is clarity, but still confusion about people who believe, AI strategy is being driven top down from the c suite. But a lot of employees, still think that the leadership is coming from the technology organization. So you've probably heard a lot of the consulting, analysis around what is the organization that's gonna work in AI. And it may not be the same for everyone. Right? It won't surprise me if if there's a whole host of organizations that emerge because enterprises have different business, you know, different business models, different strengths, and and different things may work. But what we do see, pretty conclusively is that the classic pyramid is not going to, survive because there will be more and more assistance from AI in doing some of the work that used to get done at the base level of the pyramid.

And some, you know, including Gartner, would predict a shift towards a diamond shaped workforce. And the implication there, similar as to the Pentagon and and some of the others, and and now you know what the Washington Monument shaped figure is called, an obelisk. But the idea is that there will be a thicker middle layer where experts or, domain specialists are leading teams that have fewer people, but together, they're managing an agentic AI system to get to a business outcome, on behalf of their group. So whatever the shape is, I do think smaller cross functional teams that are led by managers who are acting as conductors and orchestrators, will be the way of the future. So this will, as we talked about, AI is gonna continue to change. What it takes to get value from it and to make your organization succeed will continue to change.

And we believe strongly in agile learning culture is going to be, critical. So next challenge is on cost. And I hope, I I'm sure some of you have faced this already. You may be facing sticker shock. I loved the post from one of the startups, that I saw where the CEO of Startup eighty ninety, posted on x that they had reached a $10,000,000 a year run rate for token consumption and yet saw no improvement in revenue or profit. So, there we we do need to get a handle on costs and how we can connect it to business value. In the past, you know, you may have been looking at managing cost in terms of seats or licenses and subscriptions. Whereas now, more of your business logic is going to be embodied in systems that you pay for on consumption, token usage, compute consumption, API calls, which makes it more difficult to forecast and budget your costs.

On top of this, oftentimes, we we kinda revert to our old way of thinking. Not that it's bad, but it doesn't flex quite the way we want it to, where we want every individual effort to demonstrate an ROI rather than looking at a whole portfolio of initiatives and recognizing that we are continuing to learn and experiment at the same time that we're moving things that work into full scale production.

And then there are, the costs you don't think about when you're going through pilots. Right? You're not thinking about all the costs of procuring your AI tools. How are you going to measure? Are you gonna measure a single agent replacing multiple human functions as one unit of cost, or do you wanna keep measuring the equivalent work that could be done by one person as your unit of men measurement. And so that will, take some time to get good at. And, of course, there are other enterprise costs around data integration. And we have to recognize there will be continuous retraining as the models, tools, capabilities, and the job designs evolve. So, really, AI has become a financial question. Gartner's estimate is that the evolution from AI experimentation towards enterprise scale adoption is going to drive AI spending globally up to nearly $5,000,000,000,000 by 2029.

That is tremendous, and it's up from 2,000,000,000,000 or 1,800,000,000,000 last year. So, you know, we have seen that AI can create value. We've seen that. But the investments are substantial. So organizations have to be quite deliberate deliberate about where they place their bets, be disciplined in looking at the pace of scale, and relentlessly focused on value measurement. So if that all sounds painful, now we wanna talk about how to be successful. Alright? We are all, part of an organization of one one or another. And whether we're leading the change or we feel impacted by it, if we think about these keys to success, hopefully, that's going to help you each of us to shape our individual roles in adapting to and leading this change. So we need to articulate the vision, set up an inclusive business strategy, build a fluid learning organization, and create a thoughtful road map that lets us adapt as we go through our execution process.

So our North Star, you know, sometimes people think strategy, vision. That's this thing we do on the side. You know, it's a board presentation. But we don't really redirect our day to day based on those. But I I really think in the AI era, you have to get leadership together, and you have to turn that view around. We have to be thinking about how are we in a potentially dramatic way, transforming our own business, our own organization for the benefits of AI. And it won't come about by itself. So, now developing this bold vision, you kinda have to go through the normal things you would on a strategy, but through the lens of AI.

So what new value is going to come from AI in your industry? You know, what are your competitors going to do? What are your customers going to expect? And a lot of people talk about the efficiency, the savings. They use AI as an excuse for why they had to go through a a restructuring. But the new value from AI, the new work that can get done. Like, even in my own team, if I I have a small strategy team right now, but I am probably asking for about 10 x the amount of research and analysis as I did before because we're able to use, AI to really supercharge the reach and depth of the the research that we do. And I think that that that's going to happen across all kinds of, industries. There will be new new value you can create and new ways that your customers will look to consume greater volumes, than they do today.

And, traditional models of strategy and execution are just not going to move fast enough, So the adaptive road map has to, test the beliefs and assertions that you build into your vision and be ready to make changes as you see the market changing and you see, what's possible with AI changing in your space.

So when we talk about this inclusive strategy, one thing everyone is recognizing is the complexity and the pace of change and the scarcity of talent means you have to be, working with a team. You can't really go it alone. But, of course, having, an external ecosystem does increase some risk. And so maybe the most important piece of this is what you see on the bottom, which is the AI governance because we we've only touched on this so far, but the security, the, ethical use of AI, your governance process, the guardrails you put in place, these are all essential. And you have to recognize that the risk you take on by working in a full, external ecosystem is your risk. Right? It's not the partner risk. So your governance needs to extend further into your vendor partners than perhaps it has in the past. You have to look at everything that might be impacting your customers, your employees, right, and the data, that you protect and extend the same protections throughout the ecosystem.

So learning. We've talked about learning and literacy, maybe not enough. Learning, you know, continuous learning is always a value for most corporations, most enterprises. I think it becomes even more critical in the AI era because what you can get from AI can bubble up from all parts of your organization. It's, again, more fluid and more collaborative. So only if you are continually reinvesting in your people will they be able to continually adapt and identify the opportunities, for AI as it continues to improve? And it it needs recognition, and it needs to be a very clear expectation. You know? Something we have seen in some of the research as well is a lot of employees don't really have a a clear understanding of what's expected of them.

So if there is, you know, if this is a value you take into your organization, it helps for there to be, a really clear statement about the, the importance of individual colleagues being given the time and making the time to continue on their own learning journeys. Next, we'll talk about how you decide what you're going to put your money into. Sometimes sometimes whether an investment is working gets caught up in culture and performance and gets personalized. And as much as you can, try to separate individual performance evaluation from success or learnings from AI initiatives, you will be better off. Because I think there's a great deal yet to be learned. We want people to be motivated to experiment, to be ready to fail, and to be rewarded for the effort of experimentation and failure so that there's no hesitation to surface up what's working, what's not working, you know, diffuse the learnings from best practices, and, accept when something's not going to work out and redirect those resources as quickly as possible.

And I think putting letting the data do the talking, you know, building that kind of discipline into your investment process. And, again, rewarding the learning more so than the outcome can help you become more adaptive and continue that virtuous circle of learning. Okay. So oh, and this is exactly how this happens. Right? You use the outcome based metrics, recalibrate, review outcomes, and, keep the the circle moving so that you know what's working, and you can inspire your team to continue trying new things until they get to that outcome. Okay. So we thought it would be nice to share a couple of examples of, large corporations that have shared publicly, how AI is truly making a difference in their businesses.

And I do encourage you to to learn more about this. They've all shared, how they are becoming AI winners by adapting the organization, investing in, and spreading AI literacy, and using AI not just for efficiency, but to, drive new end customer value, which is leading to new revenue. Just as a couple of examples, you know, retailers like Walmart operate on very thin margins, and Walmart has reported that they have improved both their top and bottom line from the ways in which they're using AI. For example, they're using it to more accurately predict consumer spend at the neighborhood level for specific products, which is helping them manage inventory and boost sales despite the tariff headwinds that they have faced this year. They've also found that their new shopping assistant, which happens to be branded Sparky, which is one of their four super agents as they call them, it's actually leading to higher customer spend as well as they're finding that people who use it have a 35% higher order volume than average shoppers.

Starbucks is doing a lot of similar things using AI to improve the way that they serve their end customers. Now they have their in store point of sale system syncing with your the customer's mobile app. So when and this is in some places, maybe not every Starbucks, if you walk in there, after you're done with the conference today. But it will inform the barista what the the preferred order is for this individual, identify who they are, and and create a more seamless, friendly, and personalized customer experience. So moving forward, you know, as, having spent my entire career in technology, strategy, and corporate development, I have always been, always in environments that are heavily male. And I think everyone in this women in technology audience probably already knows that more more than half of the global workforce, is comprised of women, and yet women are only about a quarter of AI talent.

So, you know, we've all heard AI won't take your job, but someone using AI will. I just it is incumbent on all of us not just to embrace AI, but to get the women we know, our colleagues, our friends, and families to get comfortable with it because we have to embrace AI and not let it become the new glass ceiling, right, or the next broken rug. So the opportunity is there. Success depends on culture and trust. We really need leadership and employees working together on the change, agenda, and I hope that you build a resilient AI driven organization that will see you through a successful tomorrow. Okay. So I think I've just about used up my time. So I do hope this has been interesting and helpful, and would love to, you know, hear from anyone, you know, whether it be on LinkedIn or otherwise, if you'd like to discuss anything that I covered today.

Thank you.