Ashish Chaturvedi — Practice Leader, Disruptive Technologies, Retail and CPG, and Supply Chains, HFS Research[00:21]
Hello, everyone. Welcome to another session of HFS Unfiltered Stories. This is Ashish Chaturvedi, practice leader at HFS for disruptive technologies, retail and CPG industry, and supply chains. Recently, we released the HFS Horizons report on the AADA Quadfecta for the generative enterprise, where AADA stands for AI, Automation, Data platforms and Analytics. 27 providers participated in this study. And one of those 27 providers was WNS, which also emerged as a Horizon 3 player, which is the highest level of recognition possible in a Horizons study. So today I have leaders from WNS to get their views on some of the study findings and how they are approaching this exciting space. I’m also joined by my colleague Hridika. So let me hand over to each of the guests for a brief introduction and then we’ll dive right in. So let’s start with Gautam.
Gautam Singh — Head, WNS Analytics[01:17]
Hello, everybody. My name is Gautam Singh. I’m the head for WNS Analytics, covering data, analytics and AI. I’m based out of London, and I’ve been with the organization for a few years now, and came through a company called SmartCube, which was acquired by WNS two years ago.
Ashish Chaturvedi — Practice Leader, Disruptive Technologies, Retail and CPG, and Supply Chains, HFS Research[01:39]
Thanks, Gautam. Naren?
Narendran Thillaisthanam — CTO, WNS Vuram[01:41]
Yeah, hi, Ashish. Hello, everyone. Pleasure connecting with you. My name is Narendran Thillaisthanam, based out of Bangalore, India. Like Gautam, I also came through an acquisition. I represent the automation division of WNS — CTO of WNS Vuram — focusing on hyper-automation technologies, and I’ve been with the company for about five-plus years now. Pleasure connecting with you.
Ashish Chaturvedi — Practice Leader, Disruptive Technologies, Retail and CPG, and Supply Chains, HFS Research[02:05]
Thanks. Hridika?
Hridika Biswas — HFS Research[02:08]
Yeah, hi, everyone. I’ve been with HFS for about three years now, and I’m very happy to talk to Gautam and Naren, along with my colleague Ashish, regarding what worked for WNS for this assessment.
Ashish Chaturvedi — Practice Leader, Disruptive Technologies, Retail and CPG, and Supply Chains, HFS Research[02:23]
Perfect. So let’s start with Gautam. Gautam, the reason we thought of doing a study on this topic was because the AADA Quadfecta emerged as one of the prominent investment areas where enterprises are right now expecting to invest big in the next two years. And this came up in our HFS Pulse study. So essentially, in the HFS Pulse study, we interview a quarter of the Global 2000 firms, and there is a lot happening in this space as well, and the dynamics are changing, that we’ve got to know. So one thing is the spending on data platforms is increasing. That was one study finding. Second thing was, historically, analytics used to get a lot of spending when it came to the data space, but that pattern is slowly getting changed to, or passed on to, AI. So right now AI forms less than 20% of all the spending in the data space, but that’s going to increase to almost a third of all spending in the next two years. That’s what we heard from enterprises. Are you seeing a similar shift? And what changes are you making in your offerings to take advantage of this shift?
Gautam Singh — Head, WNS Analytics[03:32]
Sure, thanks, Ashish. So first of all, I would agree with your findings. AI is definitely making a big impact and is getting integrated across AADA. So you’ve got analytics and AI separate, but actually AI impacts analytics, AI impacts data platforms, and it impacts automation. First and foremost, AI is getting integrated across all of these areas. And therefore at WNS there are a number of things we’re doing to take that to market. First of all, we have built an AI utilities hub, which is basically our investment in IP leveraging AI that develops assets that can be used across the AADA value chain. And it’s really getting integrated into all our client work. This utility hub has these pre-built assets that are being leveraged under a concept called AI plus HI, which we’ll talk about a little bit later as well, hopefully, and that is what we are building to enable all of this. The second thing we’ve done is we’ve invested — and this came through the acquisitions of both Vuram and SmartCube as well — in the AI space under a concept called AI Labs, which is really an R&D space where we are experimenting with specialists, playing with use cases where AI can make a big difference. They have a free way to leverage this across whatever it is that they fancy, and that develops useful ideas and innovations that we then take to our clients. So that’s the AI Labs side. The third aspect, very quickly, is on the talent side. So AI, as you can see, is becoming more prevalent everywhere. Everybody needs to be aware and understand how and what AI can do to whatever they’re doing. So there’s a lot of investment happening at our end on talent uplift, leveraging AI. And the last thing I’d say is that AI then extends into GenAI as well, and GenAI again applies across AADA and all of the three areas that I mentioned. AI Labs, the utilities hub, and the talent uplift is also being extended to the GenAI subset of AI.
Ashish Chaturvedi — Practice Leader, Disruptive Technologies, Retail and CPG, and Supply Chains, HFS Research[05:49]
So in the reference interviews that we conducted with the clients, building a data spine was one of the core issues that they highlighted, and one of the impediments to reaching or leveraging the Quadfecta the most. So we wanted to understand what are your thoughts on that.
Gautam Singh — Head, WNS Analytics[06:12]
Yeah, absolutely. I think especially in the age of AI and GenAI, the data aspect, which is the foundational element, is becoming even more important to get right. So the data spine, as you call it, is critical before you can talk about analytics, automation or data platforms. And therefore, from both what we are doing and what we’re seeing wider in the market at our clients, there’s a magnifying lens on the data side, and therefore getting your data spine right becomes very critical. So first of all, we are working with our clients to make sure that there is a unified data architecture that we are helping them build, which lays out almost a design to make sure that data is being looked at holistically and comprehensively. The second thing is to look at data from a data fabric perspective, which is real-time integration, ensuring governance, ensuring overall orchestration of the data, or the ability to orchestrate the data, before we get into the analytics or any other aspect of leveraging that data. All of this is done through combinations of expertise in data engineering, in data governance, in data quality management, and so on, which all becomes part of this data spine concept that you’ve talked about as well. Finally, and most importantly, we talked about AI and GenAI. Again, AI is being leveraged across this data spine aspect as well. So when we talk about data orchestration, when we talk about building your data fabric or your unified data architecture, we are incorporating AI where relevant and GenAI where relevant as well. So it’s not done afterwards only, it’s at the data spine level itself. As a consequence of all of this, getting your data spine right, even without the analytics, we are able to deliver 30% saving on overall efficiency on the cost side. We are improving effectiveness, because when you get your data spine right, the quality of the data analytics or the hyper-automation, or even leveraging the data platforms, gets enhanced from an effectiveness perspective, and we are seeing that being delivered by getting the data spine right.
Ashish Chaturvedi — Practice Leader, Disruptive Technologies, Retail and CPG, and Supply Chains, HFS Research[08:28]
Thanks, Gautam. Moving from Gautam to Naren. So Naren, over the years, if we look at all the four areas, they have been continuously evolving and converging. In fact, Gautam just partially touched upon it in his previous responses as well. For instance, 10 years back, when we used to talk about automation, it was basically automation workflows and RPA. RPA used to be that pinnacle technology in automation, but now when we talk about automation, we talk about intelligent automation, where we are looking at ML, OCR, NLP, RPA, and so on and so forth. So my question is, how are you positioning yourself in this evolving space, which is also converging, and what’s your value proposition right now?
Narendran Thillaisthanam — CTO, WNS Vuram[09:11]
Sure, absolutely, Ashish. First of all, I agree with your proposition that 10 years back we could just walk into an organization, sell our BPM services as a silo, sell our RPA services as a silo, sell our data services as a silo, and you could all be within the same organization, but the buying centers were different. What we’re beginning to see is that this is no longer the case. Customers are not comfortable if you just go and tell an RPA story in isolation. What they’re looking for is an outcome. What they’re looking for is a transformation. So customers would come to us and give us a business problem, and then we should be able to tell them and articulate that this is the way I’m going to solve it. What it means is that, just like you talk about a data spine, we are beginning to see an action spine. Think of this as your business orchestration layer, where you bring multiple of these technologies together. And what’s more interesting is that this layer is getting more and more intelligent. What I mean by that is, especially with the birth of GenAI that Gautam talked about, for the first time we are beginning to see machines being able to interpret unstructured data. So they can not only understand, they can contextualize, they can go back to the existing information pool about your past history, past consumption, they can blend it, and then help us take an action. So what we are beginning to see in the BPM industry, with the birth of GenAI and AI in particular, is this is going to be very driven towards what we call agentic business processes. So it’s not just business process alone. It is going to be a very high-cognitive kind of automation, and that’s the position we are beginning to see and take, and that is actually seeing some results. It’s early days for us. Your study also confirms the fact that there’s going to be more investments in AI, so the shift is very palpable.
Ashish Chaturvedi — Practice Leader, Disruptive Technologies, Retail and CPG, and Supply Chains, HFS Research[10:47]
Got it. And on that topic of AI, agentic AI also surfaced as a big theme in this study, so you can command a higher rate card while promising an overall lower cost of ownership. So that was the value proposition of agentic AI. There were multiple examples we came across of agentic platforms like Google Gemini, GitHub Copilot, Ghostwriter, and Amazon Q, for example. So what does it mean for a business or digital services provider like WNS? That is my question number one. Second thing is, what kind of delivery model are you envisaging for the future when agentic AI takes its place? Maybe Gautam can go in first, and then Naren can go in next for this one.
Gautam Singh — Head, WNS Analytics[11:44]
Yeah, thanks, Ashish. So I think this is very interesting. The whole agentic AI concept is actually relatively new. It started happening six months ago or whatever, but already it’s taking this area by storm, and rightly so. So first of all, the point about commanding higher rate cards — I think higher rate cards are driven by higher value. And absolutely agentic AI, but a whole bunch of other things happening in our industry, are allowing us to focus on being able to deliver higher rate cards. But specifically when you come to agentic AI, the key thing that I would point out is that agentic AI now allows us to break complex problems up when you look at it from an end-to-end workflow perspective. It allows, at least from a WNS Analytics perspective or WNS perspective, to be able to go further up the value chain in terms of what we actually can influence and deliver, and B, go wider. And I’ll explain what I mean by going up and by going wider. So what agentic AI can allow us to do is, in breaking down complex workflows into smaller components and letting there be a supervisory agent sitting on top, you can really allow the AI, and the agentic AI in this case, to be able to do a lot more of the work. So at WNS we talk about the concept of AI plus HI. And we strongly believe that we can make workflows far more powerful when you combine the two. What agentic AI is allowing us to do is to look at the AI component of the HI plus AI, really turbocharge it, take significant cost out, but increase effectiveness as well — and I’ll explain that in a second — and thereby enable the human bit, the human intelligence, which couples with the AI, to take on a more complex workflow. And WNS comes from understanding and having deep domain strength and workflow understanding. So it’s really in our sweet spot to leverage agentic AI. So for example, claims management or revenue growth management, which are not typical areas that we would address in the past. Typically, a lot of the analytics and automation focuses on the cost side, but now we can take on things on the revenue side as well, which is very exciting and very differentiated when it comes to what we can do for our clients. So if I take claims management or revenue growth management, by looking at it end to end, agentic AI can break these complex workflows and bring value to the table, and therefore go up the value chain. But in the spirit of going horizontal, we can also take on additional data sets that traditional AI couldn’t address. So now we can take on a lot more unstructured data, which may be image, video, even audio, and other sensory data that can be brought in to further turbocharge that end-to-end workflow. So let’s take claims management as an example. There’s a lot of video data that we can now incorporate into the claims management process, and that video data doesn’t need humans to interpret, using agentic AI propositions that can be taken out as a piece that the overall supervisory agent is able to combine and address the overall claims process. And Naren is going to take another case example as well, but I just wanted to highlight that from a WNS perspective, A, it allows us to take the end-to-end workflow and expand it into more complex workflows, including the revenue side, and B, bring in horizontal components like further unstructured data that further turbocharges the human use of the AI in combination. So Naren, over to you to provide a case study.
Narendran Thillaisthanam — CTO, WNS Vuram[15:55]
Sure, yeah, thanks, Gautam. I want to build upon what Gautam mentioned in terms of how it helps us with our domain knowledge, and I’ll take an example from shipping and logistics. Now, if you take shipping and logistics as an industry, for example, it is characterized by two big propositions. One is it’s very, very business-rules-centric. Let’s say there’s a container that’s going from Tokyo to Europe passing via the Suez Canal. There are different countries, different geographies, different regulations, different government mandates that come into the picture, so it means it’s a very, very deep business-rules-centric problem. And the second thing is it’s also very document-centric. If you look at it, say for example we’re shipping a hazardous material, you have to comply, so it’s very highly compliance and regulatory in nature. And by that what we mean is there’s a lot of unstructured data that is flowing into your process. Traditionally, there’s no 800-pound-gorilla software in the industry. This is where, using our domain knowledge, we started building in — and we talked about the RPAs and the BPMs and workflows of the world — we started implementing a solution which is actually in production and used by multiple of our customers, and we brought this workflow layer, if you will. Now, where we began to see a major shift and an arbitrage towards the revenue side that Gautam was talking about is with the power of agents. So we brought in three propositions when we moved into an agentic architecture. Proposition number one is that what we noticed is agents were phenomenal learning machines. Combined with LLMs, combined with long memory, and the actions that humans were actually taking, agents could actually learn sitting by their side. And you actually had these specialized agents, one agent for booking, one agent for cancellation, one agent for addressing queries. And so they were getting specialized by the day, and that elevated us in terms of how we are bringing the HI plus AI part of it, with the AI taking more and more of it and HI doing more specialized services, which brought us better price points. That’s one effect that we are beginning to see. Number two, very interestingly, what we also realized is that the multi-agent architecture mimics the way — your question was also about deployment. One of the fundamental things we noticed is we could mimic agents the way a business operations floor is performing. We have agents who are managers, who supervise certain things, and you have individuals working underneath. So when you deploy these agents, think of them as your digital twin, if you will. So one agent, or a few agents, specializing in booking, and you may have a human as a supervisor. And in your deployment architecture, you don’t have to redeploy the way your production floor is. You can actually compartmentalize it. You can make them your buddy, and this kind of deployment architecture makes it really interesting for us, because you don’t have to take a big-bang approach. You can take a small silo, you can automate it. Agents are deeply reflective. They actually bring better price points, better accuracy and better evaluation. And then you overlay with the supervisor, they learn over time, and we are seeing this in an incremental and cyclic fashion, which I think is what Gautam was also alluding to. So it gets us to a better proposition overall, both on the cost side and also helps us reach on the revenue side.
Hridika Biswas — HFS Research[18:59]
Yeah, that was very interesting, actually. We were also planning — in our services vision for 2030, if you have seen it — this is a big part of how we are looking at it in the services-as-software model. So this was very interesting, to listen to what WNS is doing in that perspective, bringing together a balanced mix of AI and human intelligence to actually productize your offerings. And we are aware that WNS has done a lot of things well, that’s why Horizon 3. So this is something that struck a chord for us specifically, because it aligns well with our services vision for 2030. I think this is fantastic. Today’s conversation highlighted how AI, automation, data platforms and analytics are not just evolving technology areas, but powerful enablers of smarter and more agile future operations. I hope this discussion sparks new ideas. Until next time, let’s keep pushing the boundaries of what’s possible. Thank you, everyone.
Gautam Singh — Head, WNS Analytics[20:19]
Thank you, Hridika. Thank you, Ashish.
Narendran Thillaisthanam — CTO, WNS Vuram[20:22]
Thank you, thank you, thank you, Ashish, thank you Hridika.
