Making Generative AI Real for Business with Vivek Jetley

Generative AI has swiftly transitioned from a futuristic concept to a tangible force reshaping the business landscape. Its application has expanded across industries, unlocking innovative potentials and revolutionizing operational paradigms. As businesses increasingly harness the power of Generative AI, the focus has shifted from theoretical possibilities to real-world implementations. This shift signifies a pivotal juncture where businesses are embracing this technology to drive tangible outcomes, innovate, and enhance operational efficiency. 

Vivek Jetley serves as the Executive Vice President and Head of Analytics at EXL, a renowned global powerhouse specializing in data, analytics, and AI solutions. With a vast client base spanning across the US, UK, Europe, Australia, and Asia Pacific, Vivek oversees one of the world’s most extensive teams of data scientists. His leadership spearheads enterprise data management, cutting-edge analytics, and AI deployment, driving tangible and transformative results across diverse industries. Vivek’s adept guidance has been instrumental in the analytics sector, witnessing a phenomenal 25x growth over the last decade under his stewardship.

In the upcoming interview, Vivek aims to debunk the hype around Generative AI by exploring its practical applications in today’s business landscape. He plans to highlight real-world scenarios where this technology has significantly improved operational efficiency, innovation, and customer experiences. Additionally, he’ll touch upon collaborations that have supported their Generative AI efforts and share valuable lessons learned from project setbacks, emphasizing adaptive learning. Finally, Vivek will outline his vision for Generative AI’s role in shaping businesses over the next decade, pinpointing key areas of substantial impact.

AIM: What’s the buzz about, especially studying the AI landscape for the past couple of years? With every industry diving in, what excites you about the fascinating combination of AI and New York?

Vivek Jetley: New York’s home for me. I’ve lived here for more than 20 years now. And it’s something that’s always thriving and always changing. Being such a big commercial capital, New York tends to emphasize ensuring it’s always transforming and keeping up with the latest changes in business, technology, and finance. You are on the cutting edge in all of those instances. In the last few years, it’s also been on the cutting edge of technology, with Silicon Alley being the big startup space. That’s grown here. Now, with AI in particular, it’s very interesting because while some of the foundational tech tools for AI were actually incubated on the West Coast, the application of how AI comes together with business, a lot of that thinking is being driven here. And that’s where it’s the perfect spot for us because you’re sitting down with your clients. You’re sitting down with some of the key executives who are asking the question of saying what does this mean for me? What does this mean for my customers? And solving that problem is core to us.

AIM: What makes New York such a ripe environment for data science? With the urgency for quick results in businesses here, let’s explore the practical value of AI’s journey. Why do you think AI’s hype has lasted longer than other technologies? What keeps the fascination with AI enduring?

Vivek Jetley: We get this question a lot of saying: is Gen AI all hype, or is it real? And the answer is that it’s mostly real. Now, there’s some amount of hype for sure. Some companies are coming out and saying AI will be a $20 trillion opportunity. You have got a lot of numbers with trillions thrown around in it. And you’re seeing just a tonne of money that has been put into AI startups; you’ve seen all the buzz that’s going on about it. And inevitably, when something like that happens, you tend to have some amount of money put in and things that are not real, you have some amount of hype accumulate, and there’s some of that happening. But when you say, “Okay, let me start looking at how people plan to use it?” That’s when the reality part of it starts to emerge. To me, three things stand out regarding the AI adoption story. 

Number one, if you think about it, follow the money. Following the money implies looking at how much is getting invested. If you look at what’s happening with the hyper scalers, the cloud providers are pumping in massive investments in putting together data centers. They’ve already started building our product strategies that are embedding AI into all aspects of what we do in productivity and in our day-to-day work. And that will have a transformational effect on how we do our work every day. Here at EXL, we are already testing some of those feature sets. We are in beta mode with some of the cloud providers. They’ve created some solutions for us, and we’re testing those out. It’s working well so far. 

Number two, there is an enormous amount of interest and use cases right now that are getting lined up in every single company. Where we stand today is companies are thinking about the 2024 budget, and I can tell you 100% of the clients that EXL has spoken to, and we’ve reached out to all of our clients. 100% of them are investing more in AI next year and probably more the year after, so they’re all scaling up their investments and taking AI to a place of saying this is not just a fad. This is more than experimentation; we will have real use cases. That starts making it real. 

The third part that makes it real is just the spread of applications that AI has, and that’s where I frequently look at the comparison between AI versus, say, robotics or AI versus metaverse. With AI, you have the benefit of use cases across every function across a number of different tasks, and the impact it’s starting to create in those tasks is transformational. So when you combine those three things, the investments, the adoption at large companies, and the scope of that adoption, it starts adding up to saying this is for sure real.

AIM: How does EXL navigate the surge in investment amidst technology hype while ensuring efficiency? What framework does EXL employ to discern viable use cases from the fluff, especially given its presence in New York, emphasizing low margin for error and the need for immediate results?

Vivek Jetley: What we are doing is we are working with our clients and helping them think through their investment decisions. We are helping them figure out what it is that they should invest in? And what is it that they are looking for in terms of returns next year? That’s for them. And in parallel we are actually doing that for ourselves. We are looking at how we can make AI work for us both in terms of savings and in terms of what we are taking to market, our product offering and how we can get benefits for ourselves? Let me touch on both of those things. 

On the first one for clients, I think most of our clients right now are in a little bit of the experimentation mode. AI, if you go back to the whole GPT release, it’s less than a year old at this point. And in that year, there’s been a lot of buzz about oh, let’s try this. Let’s try that. The first couple of things that companies started doing was, let me just get access to ChatGPT. Let me open that up. Let me see what happens. That’s now gone into a mode where they’ve actually created teams that are doing experiments, that are testing out what we call proof of concept. And now it’s getting to a point of saying we know which proof of concepts work, let’s now take it to a broader expansion and we start implementing it. What we’re doing is we’re working with companies through that staging, helping them build out that funnel, helping them understand what they should take to implement versus what they should not. And we’re actually most importantly, helping them work through the investment scenario of how you should think about these, so that five out of ten work, will you still be able to get your investment case done? What should you think about me putting money in a gating criteria, so that you actually put a little amount of money first, test it, see the return, then go to the next one, then go to the next one. And that bringing in a financial and investment thesis to it is actually what’s going to help make it real and drive the adoption further. Because the worst thing you can do for something which is a new technology is have the first couple of use cases fizzle out, in which case everyone says, Oh, that was fake. And it doesn’t work. I also want to address what EXL is doing ourselves. I think for ourselves, we are true believers in terms of the potential of what this can create. For us as a company in terms of productivity of all of our employees, number one. Number two, in actually modernizing all of our assets and our solutions and our technologies and taking that to market. And three just transformational changes in terms of using AI to do things for our clients that we weren’t able to do before. And for us, the set of all those three things makes this probably one of the most exciting new technology changes that’s happened. So we really believe in what it’s going to be able to do for us.

AIM: How does EXL determine the pivot point in its iterative framework, transitioning from small-scale initiatives to broader implementation? Have you established specific parameters or a framework to discern when to persist with a strategy and when to pivot, or is it currently a case-by-case approach?

Vivek Jetley: No. We have created an investment thesis. We have created thinking and criteria for our clients to take it from one stage to another. Some parameters are financial; others are around the regulatory framework, what allows regulatory approval versus what doesn’t, and what allows compliance versus what doesn’t. And there are elements to it where the technology component starts coming in. And is this something you can do at this point or not? So there are a number of different criteria, and cost is just one of them. We use all of those criteria to select the use cases and then ensure that the gating criteria are met from one to the other. The one thing I want to emphasize here is, by its very nature, the AI applications that everyone’s working on and the ones we’re working on with our clients will require iteration because the technology is changing so quickly. And the underlying criteria. Some time back, OpenAI launched its new capability sets, and they’ve changed pricing. They’ve dramatically changed the way they allow access. And it’s transformational enough that it will enable a whole set of new use cases. So we are telling our customers to do this iteratively; don’t think of this as a long build-out. Think of it as a series of short builds, and then keep having a viewpoint at the end of each short build and saying it’s a go or no-go decision to go to the next one. And that’s the process that we’re recommending that they follow. 

AIM: How are you building and scaling a successful generative AI ecosystem within your company? Regarding talent, are you hiring dedicated generative AI engineers or repurposing existing ML talent for these roles?

Vivek Jetley: At EXl, we have had a unique advantage in that we’ve been using AI internally and for our clients for a fairly long time. We are one of the largest data science companies. We have about 8000 odd data scientists, data engineers, and staff, and that capability and skill set give us a unique advantage in saying, look, we already know how AI works. Generative AI is a new toolkit, so we’re retooling and repurposing our workforce there. So we come to it with the fact that we already have the engineering talent. What we’ve been able to do on top of the engineering talent is because so much of AI use cases are going to be within clients’ workflow and client processes, we’ve taken the knowledge that the company has about workflows, the knowledge that we have about running our clients processes, and bringing that knowledge, embedding the AI with it and taking an end to end solution. EXL has a very unique advantage in terms of the fact that we already have those teams. And we’ve been able to retrain those teams and get to work with a pretty deep skill set. The foundation was strong enough; the skill set was there. So we have a stupendous advantage when it says, Okay, I don’t have to build a team. I just have to retrain.

AIM: How do you identify when an AI venture may not be viable? What cues or challenges signal the end of a business case, leading to challenging discussions with clients about the project’s feasibility?

Vivek Jetley: What you need to do is a lot of this happens in the setup. So you’ve got to set up the work that you’re doing correctly. If something is an experiment, you must be very clear upfront on saying this is an experiment; there’s a chance it won’t work. And here are the parameters for it. And here is how soon we’ll be able to find out. So, number one is the expectation setting. Number two, there’s a very important thing that you need to do in terms of setting up interim checkpoints. You need to know if something won’t work; you need to know that sooner rather than later. And you shouldn’t get to a point where you’re at the end of the project, and that’s when you find out. So, some of those things are setting up the expectations and getting the answer faster. But at this point, we’ve been pretty open with customers in terms of saying, “Look, this is a brand new technology; there are the odds that X won’t work or Y won’t work.” And that’s when I told you that our conversation with the client is always iterative. You must work on it in iterations rather than having one silver bullet for something.

AIM: Have you transitioned from Proof of Concepts (POCs) to full production in your AI projects? In recent discussions, many organizations mentioned being in the Proof of Concept stage—have you observed a similar trend in the industry?

Vivek Jetley: One of the things that differentiates us, and we are right now getting this feedback from customers and from third-party kinds of analysts, is that we have an advantage in terms of how fast we’ve been able to move with this. Some of that goes back to the fact that we had the experience earlier; we had the teams to mobilise much faster. So, we have a couple of different use cases already in production. We’re doing a tonne of work in the area of agent assistance. This is effectively where a human plus machine is interacting with the end consumer and customer, helping with the calls with their interactions and the ability for the machine to keep prompting the agent to say what comes next and ask the question. That’s something we’ve already taken into production; that’s life. We’re working on a large number of calls that are passing through that particular engine. We’ve done a couple of different use cases to help our customers improve their conversational BI and business intelligence applications. A couple of those have gone into production. And there are many things we are working on right now that are nearing that production stage. The focus from us has been since we’ve been doing the experiments and the POCs for a while; we think that while we’re already in production on a few, that number will be much bigger by the end of this year. By the end of H1, that number will be very steep. We’ve now started going to all our client meetings by saying I’m not going to show you a deck; I’m going to show you a demo and the ability to go there and point out to them that here’s something that’s already working. You don’t have to trust me on what my capabilities are; see what I have. That’s made the adoption curve a lot faster.

AIM: How do you envision the future of businesses with the integration of Generative AI? Looking back at the impactful shift observed with data analytics around 2009-2011, what transformations do you anticipate Generative AI will bring to the world of business?

Vivek Jetley: It’s a very broad question of what the future of work effectively looks like. And I will give you what I can conjure up as my best answer and take it with a pinch of salt. There will be a big transformation in how work gets done. The analogy that I want you to think about is that before Excel came around, people were still doing financial analysis, creating models and P&Ls, and analysing data and pivot tables. Still, it was done in a very different manner and was done in a very manual way. And when Excel was adopted, it led to a complete, universal adoption. And everyone then started creating their use cases out of it. You created financial analysts’ banks, stuff for analysis, and so on. But everyone got some productivity benefits out of it. So here was a core underlying tool, which everyone customised based on their use case and got productivity benefits. And those productive benefits were real. You could now start churning more data than you ever needed to and get better insights faster than you ever needed. That’s the analogy that I want to drive towards for AI. It’s going to have almost universal adoption. There will not be a company in the world that says I’m not using AI. There will not be anything we use in our own personal applications that won’t have a component of AI to it. So all of us, in a personal capacity and in the companies that we work with, will have this toolkit, which will create productivity benefits for us. The range of those productivity benefits and how they will be delivered will differ. Still, the productivity benefits are real, and they will happen and get almost universal adoption to it. There is a flip side to this as well. The flip side is what happens because of productivity. What happens if those jobs go away? What happens if somebody’s job gets replaced by AI? And I’m not going to tell you that that won’t happen. It will probably happen in certain very small areas. But I also believe in the power of when all of this productivity gets released; it will get put into something new, creating new business for the world. And that’s going to create more jobs. I don’t believe in the zero-sum game for jobs. So I think of it as a very big positive for the world where there’s going to be an enormous amount of productivity and then think of that productivity as creating something new, which will add to our overall employee pool, overall goodness in the society.