Latest / Make It Real / New Possibilities in the Age of Agentic AI with Samrah Khan | Google
Transcript
- Samrah Khan (00:00): The technology is the easiest part. It's the anthropology that's the hardest. When we talk about AI as a team sport, we have to treat AI agents not as a software upgrade, but rather as a new hire. Shirley Macbeth (00:14): You're listening to Make It Real, brought to you by EXL. I'm your host, Shirley Macbeth, and on this show, we're exploring how artificial intelligence is reshaping workflows, industries, and the way real work gets done. And yes, we're going to make it real. On this podcast, we talk about how to make AI real for business. And one key truth that we know is that organizations struggle to deliver AI transformation alone. It really takes a village to orchestrate success across complex data, workflows, platforms, and much, much more. And one great example of that village or a partnership, if you will, is with Google Cloud and EXL. EXL is very proud to be a Google Cloud strategic services partner, and together EXL and Google Cloud are working together with a number of great organizations to help accelerate their AI-led transformation and drive real business outcomes for clients. So I'm so excited to learn more and talk to a great leader, Samrah Khan. (01:18): She is the systems integrator partnership business lead at Google Cloud. Her focus is on driving strategic partnerships across cloud, AI and data. And a little bit about Samrah. She's not only a leader on the tech side, but she's also very committed to driving and advancing women's growth in tech as well. So welcome, Samrah. So glad to have you on the podcast. Samrah Khan (01:41): Thank you. Pleasure's all mine. Shirley Macbeth (01:43): EXLlent. Well, let's get right in. Let's start by talking about our relationship between EXL and Google. Obviously, Google is a major leader in AI in this space, and Google Cloud is a foundation for so many AI solutions that are being built on top. EXL is well known in the domain space and also as a leader in data and AI. How did our companies get together and why did Google choose to partner with EXL? Samrah Khan (02:07): Yes. Thank you, Shirley, for having me. And great question. Look, Google, as we are looking at partners more and more in the era of AI, as we call it, we're becoming very intentional on how we select our partners and which are the partners that are right partners for the AI era, especially the agentic AI era. And we intentionally selected EXL as a partner since you bring so much experience in helping organizations modernize and transform their customer engagement, their consumer experience platforms, and really, really focus on the industry specifics and the problems that exist within that industry, whether it be financial services, whether it is healthcare or life sciences or other areas. The EXL has the knowledge on how the work actually gets done within an organization, whether it's around a business process, whether it's around a specific consumer experience use case. So the challenges we are currently trying to address are all around business function problems. (03:17): Therefore, combining your expertise and your DNA and your customer trust with Google Cloud's AI platform and our data platform makes our partnership a natural fit in terms of helping our customers specifically address those problems. And the good news is that we're both focused on helping our customers turn AI ambitions into actually business outcomes, whether that's scaling AI across the enterprise or delivering better customer experience. Basically, I believe Google and EXL can really help companies get stuff done with AI. Shirley Macbeth (03:56): And I would say the feeling is mutual with EXL. A lot of our customers, of course, are running on the Google Cloud systems. And I think coming together, we've had such a shared vision and joint approach really to how to solve customer needs. And we'll talk a little bit about data, the data challenges and other areas. It's just been a really good working relationship across a number of different domains where we have expertise, but also sort of cross-functional use cases like customer experience. So it's a great partnership, so glad to be working together. Samrah Khan (04:27): Absolutely. Shirley Macbeth (04:28): Okay, great. Well, Samrah, let's set the stage with the AI landscape that's out there. And of course it's evolving so quickly. Just a couple years ago, we were talking about what is this potential of generative AI. Now we're into Agentik and the pace is just accelerating and we're trying to turn agentic AI into making a real impact. What are you seeing in the market? Can you talk to us about some of the trends that you're seeing? Samrah Khan (04:50): One of the biggest thing that we're noticing in the marketing and the data is out there to prove it. 90% of the enterprises over the next three years are going to adopt agentic AI in some shape and form. We can debate what the total addressable market for that looks like, but we definitely know that it's somewhere between one and three 10 trillion depending on who you talk to. And agentic AI is a technology that really will be adopted by the enterprises with the help of services company. Therefore, companies like EXL and Google working together becomes very important because services are delivered by EXL and our platform will only see the velocity and the adoption by our customers if you all bring your expertise to the table. With that being said, in 2026 compared to the last two years, we've really moved fast. AI just helping us write emails, summarize our docs, create podcasts, create slides. (05:56): We know that at this point in time, AI has proven that those employee productivity use cases are table stakes. Now in 2026, the trend is all around agentic AI, but in agentic AI that's not just suggesting and giving you insights, rather it's actually acting on your behalf. And what do I mean by that? What I mean by that is for a decade when we talk about AI as I've been talking about AI in terms of human in the loop, I strongly believe that now is the time where we have moved from human in the loop to human on the loop. AI agents are now orchestrating complex multi-step workflows like managing supply chain disruption, like processing an insurance claim from end to end without needing a human to prompt every single step. I would just fall in an agentic leap. So you're no longer required for somebody to drive the bus. (06:58): Rather, you need a conductor who's on the bus just supervising while the bus is being driven. So end-to-end business workflow optimization through intelligent agents is really a big focus, and that's the essence of the agentic era. Our customers are really looking for solutions that can enable an automatic task execution without constant human intervention or interaction using real-time context aware decision-making. So I think that's going to be a big one. The second thing, as I personally go around and talk to our customer and our partners, what I'm noticing and seeing is that there is a big democratization of agent creation, and you and I talked about it a little bit last week in a different setting in a different meeting, is with Google Cloud's AI platform, we are providing powerful, accessible building tools, including no code options for people like you and me who are not developers on day-to-day basis. (08:01): However, we are still able to use agents and build agents and create agents to do what we need to do to improve our daily productivity or make our interactions with our partners or our customers more meaningful, while at the same time giving the options for the engineers and developers to have full flexibility of using our ADK to develop any high-code agents that they do want to. So again, there is a strong demand for personalized context-driven customer experiences and that are also driven by personalized agent for each individual sitting within a specific function. Shirley Macbeth (08:51): I was just going to jump in and say that it's fantastic and I love the agents leap, but keep going. Samrah Khan (08:56): A lot of people compare AI to, "Hey, this is the next internet, this is the next mobile." I think in my mind, AI is bigger than internet. It's bigger than mobile. Perhaps it is very similar to the discovery of the moment where we found that the atom has a splitting capability, which of course, as we all know, led to the famous equation of Einstein E=MC², but that splitting of atom had deep profound impact on how we produced energy and at what rate, speed and capacity that had meaningful impact across all industries, whether it was healthcare, whether it was manufacturing, defense, et cetera. And I think the head of AI is in my mind compared to that time. Shirley Macbeth (09:44): Well, it's so exciting to hear you talk about that. And it's true, just the rapid acceleration, the transformation. I love what you said about democratization with the agents that really anybody can build an agent. We have an EXL, what we call an AI playground, and it's something that anybody can go to and create no matter if you're a tech person or not a tech person, and it's really helping to just rapidly transform industries and everything you just talked about. And I think it's very powerful when you think about what EXL brings of the context of the workflows and the work that needs to get done and the power of the Google platform as well. Bringing that together, it just is accelerating so many opportunities to transform and automate things that were very manual before that now can be agents on agents and automating entire system. (10:37): So that's amazing. We've talked about this promise and all these exciting things that are now possible, but there are some challenges and hurdles for organizations to get it right and to make it real, which is the title of this podcast. What are some of the challenges you're seeing? Is it the models? Is it something deeper in the data foundation? What is your point of view? Samrah Khan (10:57): A lot of people ask the same questions surely quite often. I think there is somehow a common misconception that for some reason the model is a bottleneck and the models are not sophisticated enough because maybe somebody gave it a try two years ago or three years ago when a model just came out. And what they're not realizing that the development and the advancement of the AI platform technology, including models are happening at a rapid pace, which means that we are making improvements to the product platform and shipping products on a weekly and monthly basis, not like on a quarterly or a annual game plan. So if somebody tried something two or three years ago and they think that the changes have not happened, I can assure you that the world has changed tremendously in the world of AI and products. So I think the hurdle isn't the intelligence of the model. (11:51): It is the executability of the data. You may wonder, what do you mean by the insecuritability of the data? In the generative AI phase, which is what we started with back in 2023, I believe, is when data was slightly messy, a human could still interpret the output, make some adjustments and fix the outcome. But with the agentic AI, the agent is taking the action on the human behalf as we talked about human on the loop versus in the loop. If the data is ambiguous or if it's murky or if it's muddy, the action that the agent is taking is actually wrong. Your agent is only as smart as what you feed the agent, right? Just like we always say, your health is as good as what you feed yourself. This is really no different. So agentic AI actually requires what I call executable data. That means the hygiene of the data has to be really, really good. (12:49): So in this phase, what we really need to do is focus on what is our data strategy, how are we thinking about it and not have a bolt-on, sort of forced AI solution or model in a silo somewhere, rather than holistically thinking about what kind of outcomes we are looking to why. Shirley Macbeth (13:10): I think that makes so much sense. And EXL shares that vision, certainly with what you're saying with your point of view, we always say an AI-led organization needs to start with getting that data foundation right. So making your data AI ready for what we do with a lot of clients and together with Google Cloud as well is to start with that data foundation. We have a solution called EXLdata.ai that really looks at both structured data and unstructured data, and that's the foundation. If you get that right, so many great applications and use cases can be built on top of that. EXL and Google Cloud have been working together with a major financial services company in that area of data migration and getting that data foundation in order. Why don't you talk a little bit about that? Samrah Khan (13:54): Yeah, totally. I think we've done it in a couple of different instances. I'll talk about maybe the financial services example or a use case. The important thing is that for the financial services company, the big issue generally tends to be they have massive data platforms, especially if they're in transaction space and they have to do big transactions. They have periods like Black Friday, Cyber Monday, Christmastime or holiday time, et cetera. And this particular financial institute was really having hard time processing their transaction because I was seeing their data platform. So at some point in time, they went on a journey to modernize their data platform and they were looking at several different hyperscaler options. EXL, knowing what you all know about the financial institutes and then the deep relationship that you had within the account and also the awareness of the businesses processes that you had of their business, obviously chose to work with Google Cloud. (14:57): And jointly together, we have done one of the largest data migration in the banking industry. I would just say a significant amount of terabytes of data from the poll system to the Google Cloud data platform, and the impact of it was not only just dramatic, but also immediate. The customer is now able to process thousand payments per second during major peak times like Black Friday, which is the one day of 365 days of the year where transactions are at the peak. So the important thing is that the task that used to take them three months of work on the old data platform now just takes few hours on Google Cloud platform. So your business process and domain expertise in defining their workflow combined with our technology and platforms really has tackled the key foundation business problem for the customer that has delivered massive operational savings as well as agility, which was very important for that. Shirley Macbeth (16:10): I think it's such a good example of EXL knowing that domain, knowing that industry so well where deep in financial services understand the intricacies of how that business works. And then with Google Cloud to really scale that and drive that, I think that's one of such an exciting story. You can really see the tangible business outcomes driven and driven so quickly after that data migration success. So incredible. Well, another area that Google Cloud and EXL are working closely together with a number of customers is in the area of customer experience. It's an area that's right for AI success and driving better outcomes through lots of different areas. Can you share some examples of our work together? Samrah Khan (16:52): Healthcare is obviously another big industry outside of financial services where EXL and Google are actually working together. The challenge that comes to mind is a specific example in the healthcare industry where a customer was basically looking to replace their call center tool that they were using. And that particular call center had around 60 million calls that they were processing annually. So it was pretty massive scale operation. So the customer, again, was evaluating various different options and ultimately decided to go with EXL as well as with Google Cloud and chose not only our consumer experience platform, but also decided to couple that with our conversational agents because of course this is all about customer experience. And our joint solution led this healthcare customer to migrate this massive call volume to Gemini Enterprise. But the good news is that even though we are in early stages and the results that we're seeing is already shown a significant impact on improvement around call containment, as well as how quickly we're able to get to the resolution several times without having the actual interaction or a need for a human to pick up the phone and answer the question. (18:15): So it's sort of twofold, which is not just handling the call volume at the rapid pace, but also actually increasing the containment. So what set Google and EXL solution apart was sort of twofold. Number one, by utilizing EXL's accelerator built on Google Cloud called TransformCX, which is obviously built natively on Gemini, we're helping the customer move beyond just the simple chatbot to a truly intelligent experience. The projected impact is massive. The potential of generating over 130 million in cost savings is actually tremendous. So for us to jointly, whether it's around addressing what's coming through chat, what's coming through emails, what lives in the docs, what's happening on the call, together be able to process and deliver a better customer experience or a consumer experience to the person on the other side of the call is very, very important. It's not just about delivering top line or bottom line for our enterprise customers, but it's also about in an industry like healthcare where empathy really matters, solving a problem of a consumer or a customer or a person calling into you, I believe is equally important and might be a non-tangible ROI, but definitely a very important and critical ROR. Shirley Macbeth (19:48): Yeah, I mean, I think cost savings and speed and accuracy, but when it's kind of the bottom line also of the experience that caller or they're trying, regardless of what industry calling in and trying to get to an outcome faster, we all know that that experience that you have, you're very quickly, as a consumer, trying to get to an outcome or a result or an answer or whatever it is much more quickly. And that's what's exciting, I think, to me with a number of these joint examples that we have in that area of CX. You can just see how AI is really transforming that ability to get to the answer faster and it's a win-win all around. Well, certainly with customer experience with such an amazing area of opportunity, the flip side of that is employee experience. So the people that are actually now working with AI within these organizations that have adopted AI. (20:36): And there's an important piece of change management that Samrah, you and I have talked about before. You've mentioned to me that AI is a team sport. And when you're deploying these AI agents into these CX systems, whether it's a nurse or claims adjuster or a customer service rep, there's a big important element about the change management required for how the people within that organization or how do they trust and adopt that new virtual teammates. Let's hear your thoughts on the change management that's required. Samrah Khan (21:08): Shirley, you're spot on. The technology is the easiest part. It's the anthropology that's the hardest. When we talk about AI as a team sport, we have to treat AI agents not as a software upgrade, but rather as a new hire. In my mind, the biggest issue is around replacement versus augmentation and the fear in people's mind that AI is going to take your job. I think we have to really focus on that if you are a claim adjuster or if you are a nurse, and if you can put an agent to work for you and gather all the data for you, so you can really focus on treating the patient or treating the person who's just had a car accident or something like that and they need help and lead with empathy, IT, you will deliver much better consumer or customer experience rather than focusing on data gathering. (21:59): So it's all about build a glass box rather than a black box. So my always recommendation to the enterprises and customers is it's not like a top down approach. It's like a bottoms up approach as well. Make your people part of the process. If the claim adjuster has to say in how they want to use the agent and what they want the agent to do and they're part of building the agent, then the adoption will go high. So to me, change management is very strongly tied to inclusion, and it's not just a tech initiative, it's a company culture initiative. Shirley Macbeth (22:37): You said so many good things there. I was thinking about an AI agent, not as the tech, but as a new hire, and instead of fearing that it's actually augmenting what the humans are doing and how the processes can work, I think that's such a big and maybe undervalued part of a deployment in some cases that everybody's focusing on the tech, but it's really also the adoption and the follow-on and getting the people involved in how these agents are going to work. So such good advice. Thank you so much. Well, Samrah, we could go on and on. This was such a great conversation. I really appreciate your insights from Google Cloud and from what you're seeing in the market. And for me, the key takeaways that I learned from this session were, number one, data foundation is key to AI. Getting that readiness on board is critical to launching your success. (23:26): Number two, from some of the case studies that you talked about, there's such a massive opportunity for really immediate benefits when you're looking at areas like customer experience and across a range of industries. We've talked about healthcare, we talked about financial services. The third thing I took away was really around the importance of change management, having that agent feel like a new hire and the importance of inclusion as part of that experience. And then I think lastly, through what we've been talking about, I think an ecosystem is so important for partnerships to help accelerate the success with some of the organizations that we're seeing. And that's why EXL and Google have been such great partners in driving outcomes together. So hopefully I summarized that pretty well. And I just wanted to thank you, Samrah, for your time today and for your great insights. It was a really a pleasure having you on the podcast. Samrah Khan (24:20): Same here, Shirley. It's always a pleasure to sit down with you. I think together our mission is that we're going to serve our customers in the way that when they don't have AI, they will feel underserved, and I think that's where things are going. Shirley Macbeth (24:33): Awesome. Well, thanks again. Have a great day. Samrah Khan (24:36): Thank you. Shirley Macbeth (24:40): Thanks for listening to Make It Real. We hope today's conversation gave you ideas, insights, and inspiration to help bring AI to life in your organization. Remember, big ideas don't drive change, people do. Keep learning, keep experimenting, and keep embedding AI where it matters most. Follow along so you never miss an episode.