Transcript Disclaimer: This transcript has been generated using automated tools and reviewed by a human. However, some errors may still be present. For complete accuracy, please refer to the original audio.
Opening (0:00)
There was an article which said that this is going to generate about 100 million jobs in the next two to three years. If that is the case, how is the BPM industry really gearing itself to meet those requirements of human-in-the-loop services? For us to have the right intervention for human-in-the-loop, that is where we need to ensure, one, we have AI-enabled workforce. My spectrum tells me three to five years there is going to be an uptick on AI-enabled workers till the time the models stabilize.
What is happening now in the industry is, you know, the core model is changing. Next five years, I think we'll be more focused on measuring the outcomes and measuring the value-add that we are doing as an industry compared to, you know, how many people we are putting in. With the use of AI, we have the risk of, you know, creating things which can soon change.You have to have some balancing to ensure that there is continuous growth. We must keep in mind why we need to upskill people. We also need to upskill the AI. Integration of technology is the biggest challenge right now and we term it as, is your organization AI-ready? It's a big opportunity for all the organizations in the world to change our narrative, to say that we are the AI bed for the world.
Introduction (1:32)
Hello and welcome to this very special edition of our webinar series brought to you by Business Insider India and Nasscom. I'm Ridhima Bhatnagar. As we continue our journey exploring the various (1:45) fascinating facets of the BPM industry in India, today we focus on AI-ready industry as far as the readiness of the BPM industry in India is concerned. And to do that we have a stellar panel. But before I introduce our panel, what are we going to talk about today? You would all agree when I say that we are at the cusp of something great, something new.It is a new era where artificial intelligence is redefining how we approach, how we understand and eventually a solution to customers as well. Today you have businesses integrating AI in organizations which is helping them to make things far more agile, far more effective and far more productive as well. But the larger question now is, what is the future? Where do we go from here on? And that's exactly what we'll discuss with our panel. So let me not take any more time and introduce our stellar panel.
Introduction of Panellists 2:43
Mythily Ramesh, Co-Founder and Director at NextWealth joins us. Hi Mythily, good to have you with us. Hi, good morning.
We have Nitasha Atreya, Head of Transformation, Digital Operations and Platforms at Wipro also joining us. Hi Nitasha, good to have you. Good morning and thanks for having me here
We have Suhrid Brahma, Chief Technology Officer at WNS Global Services. Hi Suhrid Brahma, good to have you. Hi everyone
Completing the panel, Sundar Hariharan, Vice President, Digital Business Experience Center, Alorica IQ as well. Hi Sundar, good to have you. Hi, good afternoon to everyone and thank you for having me here.
So at the onset I would like to firstly thank all the panellists for being kind enough to take out this time. I hope this will be an insightful conversation and as I always say, hopefully we'll all have some learnings by the end of it.
Understanding where India stands with the integration of BPM & AI (3:30)
So let me kick start the conversation with Mythily. Mythily, before we really get down to the brass tacks to try and understand what is the work that is being done by organizations and where India can move on, let's really begin from the fact as to where we are currently. Now we've spoken about the integration of AI in the BPM sector for a while now, but if you could just set context for people who are watching us and even for the other panellists as to where you think currently India is as far as AI integration is concerned. Ridhima, I would first of all thank you and really nice being here and meeting the fellow panellists as well. I think as far as the AI integration that I would like to split it into two. The first part being AI integration into the BPM industry to deliver services to clients and number two, what are clients doing from the AI aspect of model building and therefore what does it mean for the BPM industry.
So let me take the second part first. We do talk a lot about AI and Gen AI especially in the last one or two years. What has been happening is a lot of clients especially the big tech companies have started, while they've been on the AI journey for a long time, they have started with Gen AI models in the last two years and therefore AI and Gen AI models amongst our clients especially is the POCs and the model building is actually mushrooming. With that, what is needed is the human-in-the-loop services because we have to actually distinguish automation and AI. Automation is 100% what it's supposed to deliver. AI is a probabilistic model and therefore the human-in-the-loop part and the human eval part is always required.
So with these models mushrooming across the client industry, there is a need for the human-in-the-loop services as far as the industry is concerned and industry players are concerned. So therefore, if that has to happen, the industry itself has to gear itself to be able to meet those requirements and I just think you know there was an article which said that this is going to generate about 100 million jobs in the next two to three years. If that is the case, how is the BPM industry really gearing itself to meet those requirements of human-in-the-loop services and I'll talk a little bit about that a little later on the conversation.
On the first part which is integrating AI back in the industry itself, I think there are a lot of use cases today. One is the contract management, there is conversational AI from chatbot to conversational AI, contract generation that is happening, there is AI assistance for notes and minutes, SIEM security incident and event management applications are also there. For example, in our organization we generate about half a million security events in a day and in a year it's about half a billion.
How do you really sort through that to see which one is really an incident? So we have an AI assistant which picks out the top 100 cases or top 100 events for us and then we work on it. There are a lot more use cases, probably I'll talk about that a little later. Okay, okay.
Mythily, thank you for really setting context and I really like the big number that Mythily gave for us. She said there are estimates that this could create 100 million jobs. Now if that is the kind of potential that this sector has and this is the kind of end result that we can see, the next question of course is how do we get there and how ready are we? And Natasha, this is where I want to bring you in, right?
Infrastructure Ready (7:08)
Because we're still setting context to try and understand where we stand to understand where we can go, right? If Mythily is saying the potential and the untapped opportunity is 100 million jobs, help me understand how is that 100 million going to be created? Of course, one of them is infrastructure and other is the human capital that you get in, right? So let's really begin from the infrastructure.
From an infrastructure point of view, where do we stand currently? And actually, and thanks for that question, Mythily touched upon a very amazing point when she said that there are going to be new jobs in this industry and that's a very contrarian view because what we're hearing actually is AI will take jobs. Take jobs, exactly, yeah. But we who are in this field actually have a very opposite view saying that we think it is going to create more jobs and Mythily touched upon this human in the loop conversation.
What's really happening in this field is everybody has jumped onto this bandwagon of creating models, of ensuring how do you get AI into different business units which we look at, whether you talk about a CFO organization, you talk about CHRO organization and the way we look at it is whether you're talking about horizontals or verticals within an organization, independent of the type of industry you are in, whether you are in retail, whether you are in energy, whether you are in aviation, independent of that what's really happening is there is a need for AI enabled worker. What that basically means is because everybody is, and Mythily touched upon a point saying that it is a probabilistic model, what that basically means is it is going to, the models are going to work basis what has happened in the past and is throwing probabilities at it. To ensure that those probabilities are always kept in check and there is mistake proofing happening at every different section of the spectrum, there is going to be a requirement of human in the loop.
But for us to actually have the right intervention for human in the loop, that is where we need to ensure one, we have AI enabled workforce. What that basically means is if today I'm doing a repeatable job, which we basically call in the BPM industry, if I'm doing a repeatable job day in day out, I have to actually ensure that the organizations are ensuring good amount of training at the base level for everybody to be trained on what AI actually means. And hence how do you enable enhancing of the models which are playing up every day.
And that is where the need of new jobs is going to, actually we will see an upswing in this whole area. My spectrum tells me three to five years there is going to be an uptech on AI enabled workers till the time the models stabilize. And honestly speaking, God forbid, if there is another black swan event, then the whole models will go for a change and hence AI skilled workers.
So AI enabled workers and AI skilled workers. That is how we see this panning out. Okay, so where I want to get you in is to try and understand, we're all speaking about a very idealistic state where we feel those many jobs will come and we feel that integration will happen at a very, very rapid pace.
Like as Natasha is saying, her estimate is three to five years, we'll see this landscape change completely. But what I was essentially trying to understand is whenever there's a new technology, you need to understand what is the kind of tools that are needed to aid that technology for its full potential. So today, from an organization perspective, give me an example today, and say five years ago, how much has changed in terms of AI readiness from an infrastructure perspective? That's a great question.
And again, I want to start with the qualifier since you introduced the topic as a new, you know, AI has been around, it's definitely not new. I'm sure my co-panellists will agree on that point. I think the extent of change we can drive now, and with the kind of easy availability of you know, all these tools and platforms has probably made it a very different thing.
So if I look at, you know, five years back and now, I think what is happening now in the industry is, you know, the core model is changing. Earlier, it used to be more like a people driven model, right? The best measure was, you know, how many people you're putting in for running an operation, right? There is a shift, which is moving away from people to, you know, outcomes, right? What are we really covering here? And I'm sure, you know, as we go forward from here, next five years, I think we'll be more focused on measuring the outcomes and measuring the value-add that we are doing as an industry compared to, you know, how many people we are putting in. That's number one.
Number two is, you know, in terms of what type of work people are doing out here. And I think both Mythily and Natasha covered this in a brief. And we have a view of, you know, looking at it in three different lenses.
You know, one is what we call as AI for operations, you know. So, when you're running operations, how do you use AI to enable the workforce so that they can be the human in the loop? And it's mostly, you know, I would say AI skilled workforce and how they can do that. So, that's AI for operations.
Second block, what we are doing, and this was probably not as much there earlier, is AI for applications or platforms. Now, it's very common across the VPN industry now that, you know, we can vary in terms of size and scale, but usually with every VPN deal, there is a technology component that goes in. It can be an enabling tool that sits on top of the client ecosystem.
It can be a platform also that's provided as a service model, and it can process something end-to-end and then go back and update the client systems, right? So, and there's a huge leverage of AI coming into applications. In fact, I have a personal view that what we see is more of a gen AI, you know, it's probably a human brain which is understanding what is written, but then you need the hands and legs to execute, right? So, AI in isolation probably won't be as successful. It has to be embedded in the enterprise systems to make it work, right? So, that's AI for apps.
And the last block that we are working on is, you know, AI for what I call as analytics, you know, the data and insights, right? You know, analytics earlier was more of a rear view, you know, something that has already happened in the past, and you're trying to figure out from there, it has completely switched to a predictive mode. So, if you look at it holistically, you know, I see there has been a constant evolution of the model compared to what we had five years back. And if I, I mean, none of the crystal balls, I'm sure nobody talked about AI till November 2022.
But if we still do a crystal ball gazing, my view is that this whole industry will grow for sure. But the nature of work that we do, the kind of value we do for the clients, and the kind of outcomes we deliver as an industry in India, or overall across the globe, would definitely change and it's going to grow. Okay, I like that optimistic tone of how, you know, optimistic the players are, because the true potential is understood by people like you who use it day in and day out, and who must be seeing a change already in terms of the solutions given to customers, the way technology is being adapted for efficiency and productivity, which I'll talk in just a bit.
But Sundar, I want to bring you also in the conversation, and I really like what Suhrid said, right, you can't see it in isolation, you do need the hands and legs for it as well, right? So help me understand that integration of not working in silos, has that been easy? And have you seen that adding to efficiency and productivity? So, you know, a lot of great inputs from the earlier speakers already into this topic. And, you know, AI as an enabler, or as augmenting the human being is, is probably the way this industry is, you know, going to progress, because with the risks that are associated in completely leaving AI to replace humans is just not going to probably cut. And in a service industry, like the BPM industry, where clients, you know, are at the other end, and then there is an end customer, whose work, whose conversation we are actually taking or whose instruction we are, you know, doing in some form or the other, at this end, you know, in terms of the work that we are performing.
So AI cannot be isolated, it has to be augmentative. Integration challenges will always be there, you know, client systems, legacy systems can, can cause impediments in terms of how you want to put in the full nature of AI and how it can be, you know, working alongside some of these legacy systems. Now, you know, mainframe used to be the legacy systems earlier.
I sometimes feel RPA, which probably tarted 10 years back has kind of become legacy, in many ways, because processes have changed. And even the RPA organizations have kind of, you know, adopted a lot of AI into their systems to, you know, better enable the processes that are there. The robots on its own, requires frequent changes, etc.
And these integration challenges lie. The other things that are happening is... Sorry, Sundar, I apologize for interrupting. Sundar, you were saying that, you know, the robots themselves need changes from time to time. If you could just help us with some of the changes. And I just wanted to understand because that will give us an offshoot to understand what are the kind of challenges later. So I just wanted to understand what are these changes that you're referring to? So you have created a robot to do a particular service, let's say in a banking or a financial services, it has to do ABCD in a set of processes.
Now, with the changes that are happening in terms of digitization, etc. These robots usually are built with the understanding that they are non-touch means they work externally to an existing CRM system or something of that sort. Now, if the banking CRM system itself changes, and there is inbuilt automation built into some of those processes, which earlier the robot was probably performing, you will have to, you know, either sunset the robot because those functions are now enabled natively within the legacy system, which earlier was there on which your robotic system or automation system was built.
And that has, you know, that has augmented by itself. And there is AI within those systems as well. So you take any, you know, CRM, you take any ERP systems or any kind of systems that are there, all of them have inbuilt automation, most of the services comes out of the box with capability to do those transactions, which earlier was relied on by humans or by robots or a combination of things.
So there is a change happening within the systems that we are using as a BPM services provider, the system applications platforms that we work on. And with the use of AI, we have the risk of, you know, creating things which can soon change. And therefore, the cost of that change is something which at this point is very difficult to predict.
So you have to have some balancing to ensure that there is continuous growth, as well as, you know, the integration challenges are better adopted and change is more easier to manage. Okay, I have I have a lot of thoughts. A lot of very interesting points have come from what Sundhar said.
Mythily, I want to bring you back in the conversation. And this is something that I want to actually delve a little deeper, right? All the panellists agree. And we're speaking about the integration of AI going forward in terms of what the potential is.
The other aspect while there's consensus that there will be job creation. But you know what one of our panellists said that none of this can work in silos, right? So when we talk about change, it's not just change in terms of an infrastructure or just technology integration, but it's also mindset change, right? And for that, you need to go a few years back, right? How we've seen a change from the BPO to the BPM industry. So from a human capital perspective, Mikey, I want to understand what is the kind of change that is required? Okay, so, so if you take the entire AI lifecycle management, right, of models getting developed by the AI or Gen AI models getting developed in client organizations, there is something that happens initially in the AI development model of the AI or Gen AI model development, where when the model is getting developed, and then you know, it gets developed, (20:48) and then it goes into deployment.
But the reality is for every 100 or so models getting developed, only one goes into deployment. And these 100 models have to get trained with data, either synthetic data or human data, and they have to be fine tuned. And they have to, based on the data that is getting developed, the AI model functions, you have to validate the output of the model to see whether it's functioning well or not. Because you have 100 models, and only one gets into deployment, there's a lot of requirements for data training and validation. And therefore, data annotators, and labellers, image annotators, video annotators, and so on. And therefore, the skills of the people who are actually doing this work, they need to be retrained on understanding. And remember, this is not standard SOP, like we would have in the BPO industry, I have a set of rules, standard operating procedures, and then I execute. Here, the client themselves are building AI models who they don't, I mean, they themselves don't know what that AI model is supposed to deliver. I mean, they have a broad idea, but whether it is, you know, it's not crystal clear, like an SOP.
So, when it comes to the people who are actually doing this work, they must understand the context with which that data training needs to be done and what that AI model has to deliver. Therefore, skilling at that stage, skilling in terms of understanding what the model is supposed to deliver, ability to do this data annotation and labelling. Third is also being able to take certain judgment calls.
So for example, we do for a client, a retail client, they want to be the cheapest online retailer, they have something like 5 billion SKUs, they have an AI ML algorithm, which runs and compares their products and pricing with that of competition. So it's for 5 billion SKUs and category managers have to take decisions on pricing based on this. So how do you know if you have something like this, which is the mouse, there are 20 specifications, the mouse, 10 are matching, 10 are not matching. Is it the same product or not? The algorithm will not know. So it will come to the annotator or the associate. Associate has to take a judgment call to say, okay, these 10 specifications are matching, but looking at the image and various things, it is the same product or it is not the same product, and they go to train that AI model back.
That's what I mean by saying that they not only need to be able to do the annotation and labelling, but they must understand the context with which that AI model is operating, and therefore are able to take that judgment call and feed that information back to the AI model so that it can learn. I mean, what we call is reinforced learning with human feedback. So that's how the people have to be, the associates have to be trained.
This is on this side of the spectrum. When it goes into production, you are now touching the client directly, because now it is into production. So give you again an example, we have a client where if you go to Airbnb in the US, your ID card is taken, your picture is taken, and then it is compared to see whether it's the same person or not. And typically you will give your passport. So this client's platform actually does that comparison. The client's platform is an AI model. It does a comparison of the pictures. And in not all cases, it can say it is the same person. For example, my Aadhaar card and my face is totally different.
So not always it can make that comparison. And in some type of countries, the face, facial itself, points itself is very different. So the AI model is able to do to the extent of about 50%. But 50% comes to us here, and it's a 24 by 7 by 365. We have to look at the picture face, take a judgment call based on certain parameters, and then revert back to say, yes, it's the same person. No, it is not the same person. And these are the reasons why it's not the same person. So again, the training for the associate, the human capital management that you rightly mentioned, has to have this sort of skilling, has to have those judgment skills, judgment capabilities, must understand what thatAI model is trying to do. And in this case, it's very dangerous if you have false positives.
So, you need to be all the more careful. And I'll talk about the delivery part of it. But therefore, the skilling, the knowledge, the context, the fact that I don't have SOPs every day, which is a standard, I'll have instruction sets, which will come to me, which will change every two, three days.
All that has to be part of the training that goes into the So I think it's a, I think I mentioned that we are at a cusp, as far as the industry is concerned, whether in terms of skilling the people, the type of work that is going to come to us, the type of delivery that we will have to do, the methodology of delivery we'll have to do, and hence the sort of automation that we might have to bring in to bring in the productivity and efficiencies is definitely going to change and in the three to five years like Natasha mentioned. Sure. So this is really fascinating.
And that's exactly what I wanted to understand, right? We keep seeing that, you know, AI is going to make it far more effective, far more agile. But what we need to understand also, and this is very beautifully explained in the example given by Mythily is that they can't be a blind dependency as well, right? You need both the technology and the human to work simultaneously. And for what technology cannot, the humans need to have that spur of the moment thinking as well.
Implementation of upskilling programs in organizations (26:23)
And Nitasha, this is where I want to bring you in. This is a very crucial aspect of not just skilling, but upskilling as well, right? Because you're not going to get new talent every time. You will have existing talent that you will retrain, you will reinforce, you will ask them to upskill as well.
So again, from an organization perspective, help me understand what are the kind of upskilling initiatives or programs that you've implemented with the advent of the new technology and what are the kind of changes that you've seen? Sorry, Riddhima. Sorry, Nitasha, I just wanted to add to what you said rightly. It is collaborative intelligence, which is equal to human intelligence plus artificial intelligence, which is what you're telling me.
Yeah. Sorry, Natisha. Thanks for that, Mythily.
It's interesting, given that how much upskilling, because honestly speaking, Riddhima, as you rightly said, we are not in a situation to go out and hire this talent all of a sudden from the market, because even if we were to do that, the talent is not ready right now. And hence the need for upskilling. From a Wipro organization standpoint, what we have done is we have actually divided this into four major categories. One is basic general awareness level. Do we even know what are we talking about? And that is our ground zero, which is for all employees. And we have already, we are staring at 200,000 odd employees already trained on general awareness of what AI and Gen AI is all about. We are actually terming it as L0 or AI 101. Then we have a very separate intervention for our sales teams. Because when you're actually in the market, when you're actually selling solutions to the clients, general basic level of AI awareness will not help.
You need a very focused intervention of training from a standpoint of what really is going into that space. And then that space changes the sales team. Actually, that space changes from industry to industry and market to market, region to region. Europe behaves differently than America's. And hence we are staring at some 30,000, our sales team, 30,000 in number for them to have a foundation level, which we are terming as L1. For them to be aware of what goes on into the geographical levels, the compliance, the regulatory things, all of that is what our sales and business teams need to be aware of.
Then we have a separate intervention as Mythily was touching upon, the developers and engineers. That is where a high level, advanced level of training is required. And that is what we're turning as L2. And this is somewhere around 10,000 of our developers and engineers who are actually trained on that. And the last but not the least, but the most important one is our architects. These are the people who are actually working on the models.
And we have a very separate level of, and we are determining this as a professional certification. This is a very different level of certification required for people who are actually working on the models. So if you actually sum it up, we have a four-pronged methodology or an approach of ensuring how do we train our workforce or how do we upskill our workforce. And then, as Mythily was saying, business-to-business rules and the methodologies which we are adopting, client-by-client, that is a different perspective altogether. But this is what we're doing at a Wipro level. Lovely, lovely. That really helps us understand the kind of work that goes behind, right? Because it's not just, as you were saying, Nitasha, it's not just even if the organization makes a decision that, okay, I'm going to hire this much talent from outside. One, I don't know if that talent is available, as you rightly pointed out. Plus, there is a cost at every element for all of this. Right? And every organization has designated budgets for this, which, of course, is a different conversation that we'll do separately.
Customer Demands- Challenges and Solutions (30:17)
But, Suhrid, I want to bring you back in the conversation. Now one end, of course, was, you know, upscaling the labor that you have, integrating it with technology. Now the other end of this patenting spectrum is customer demands. Right? At the end of it, all the organizations are working with the same goal, which is how do I simplify this entire process?
Can I solve these problems faster? Can I understand them better? So first, help me understand two steps backward. How are you seeing the kind of nature of customers evolve? And then we'll understand how technology is changing to their demands. No. No. Absolutely. We'll do that. But I'll let me start with one point on a lighter note that Natasha mentioned earlier and you talked about upskilling. We must keep in mind while we need to upskill people. We also need to upskill AI. There is not a point in time because every time as Mythily mentioned, you know, you can start with, say, an AI rule engine, which is 50% accurate. But as you process more data and as the rules change, you can get from 50 to 60 to 70 to 80. I'm sure all of us have done that journey. So since we heard about upskilling, I was thinking so coming to your question, Radhima, see Yeah. There are there are, like, few patterns, and let me just elaborate those. Right? Yeah. See, one part is when we are already doing something for a customer. Right? All of us are existing customer, existing processes that are running. Right? So one set of customer demands that are coming our way is saying that, look, I mean, there's a lot to do. Everyone and, again, while we have a AI generated strategy, there is a, customer CTO organization or a CD organization, whatever you call it.
They have a strategy also. Right? I think the first task is, which is very common, is whatever is your existing base. Right? Right. How can you really apply, AI and AI enabled models combined with, you know, upscaling people? And how do you drive a better outcome or a better overall, you know, which is more like a win win for both the customer and the provider. You know, that's the first one. Right? The second big ask is where which which I say is, you know, which we call as a new customers or, existing customer new process, which is not being run already.
Right? But probably in the back of transmission. What has changed and as you said again, if I take a two year back view and, view now, I think everyone is telling that, you know, it's a transformation first. Right? We cannot afford to say that we'll let a process run the way it is running today, you know, have the liberty of running it for another two years, and then pass on the benefits.
Right? So Yeah. So it's like, you know, changing the wheels of a card that's running. Right? So the ask is and I'm and I'm talking about the net new, which is not to be with any of the BPM providers, but everything new coming in. So the how do we really embed all these AI data analytics technology tools inside whenever we get those in? And what kind of benefits of course, you know, there's a commercial conversation. How do we do this? So the second block. Right?
And the third block is very interesting. Right? Which I'm sure you'll not love to have problems. Also, what they're seeing. See, there are clients who are actually looking at us and saying that, look, I am actually not interested in BPM services. Okay? Because they would still want to run the process. But Yeah. Fact that they see we have and there's a value play I was talking about earlier. Just that we run so many processes.
Right? Say, finance and accounting, we probably run it for more than 400 customers. Right? So Sure. We have a view in terms of, you know, what we can do in, say, particular again, it's after all finance and accounting, like a profit to pay or to cash. Or say insurance, you know, we run it for clients across the globe. We have a best practice in terms of, you know, how we would do our claims or a policy admin or an actual. Right? So then clients, you know, and that's only the net new segment that we can talk about after all this Sure. Related to the coming in. They're saying that, look, since you're doing it already for your existing customers Yeah. We may or may not be ready to do a full blown business process management conversation. But can you also bring in those nuggets of AI, of those tools, of those platforms and allow us to use that. Right? Correct. That kind of, you know, is changing the game. Because one other thing is earlier, you know, if I go back and ask them, we've been counseled seven, eight years back. RPO was the poster boy. Right? And Yeah. Debating whether RPO will take the job survey or not. If you understand, you know, RPO was at best task automation. Right? Yeah. With AI coming in, we can actually do an end to end process automation. And when you do an end to end process automation, there'll be a part that is executed by the BPM providers, but there's also is a part that is what we classically call as a written organization or the client written process. Right? Okay. The ecosystem for application of AI and, you know, the broader set of tools. Change from the earlier set of only the BPM process to then the next process got to see the kind of technology, the kind of platforms we have more. And that's Sure. Big change. And my view is that, you know, we probably have a lot more happening on that side. Sure. So if you know the way forward here too.
Okay. So then I want to bring you in and we are also slightly running out of time and I still want to discuss the challenges and way ahead. So then but I quickly want to understand again from a customer perspective, can you give me some case, you know, examples of certain AI tools or strategies that you've implemented from an organization perspective to more personalized the customer experience? So we have lot of examples. At the moment, you know, one of the big launches we have done in partnership is, real time voice, translation. Okay. Multiple languages. So customer speaks in, let's say, German. The agent, in India can hear it in English and can speak back in English, and the customer will hear it in German. So, we we've integrated, knowledge management systems, with, GenAI Okay.
Into a lot of, customer processes. So, loads of data, documents, you know, specifically in tech support and those kind of areas, you will have different model numbers and, lot of literature associated with the product. Gen AI can easily kind of, generate, you know, what could be the troubleshooting steps, depending on the customer query, etcetera, very easily. And it can, also generate it in ways you want. You can generate it as an email. You can generate it as bullets in a to be you know, put it into a chat. So, the options are many. Okay. Even also using AI, which is not really generative AI, but, AI in, you know, a lot of voice modulation and Sure. Accent neutralization, neutralization kind of, work.
So mother tongue influences, etcetera, can be removed while, customer is in conversation with our agent, avoids repeating of certain words or terms, on either side, for easier understanding makes the call better, from a customer success perspective. So opportunities are many. Yeah. The there is a push, to put in a lot of these kind of technologies to a better effect, in our customer spaces. And like, you know, some of the panelists spoke earlier, the opportunity now is, around how you can, you know, use your specialization in a particular vertical or horizon, and play into the client space while the client retains much of the, BPM, activity.
That is start augmenting that using tech tools and technologies that you have developed Yeah. Within the client space. So, moving from a BPM service provider, pure play. To more a tech organization, and a consultative organization Yeah. Seems to have a lot of traction, going forward. Okay. Lovely. So, you know, we don't have much time left. So I want to quickly now, go across the panelist. And I quickly want to understand some challenges, some solutions.
So I'll try and divide this between the panel. Natasha, if I could come to you. We've spoken about the kind of potential, the kind of work that is being done, but none of this comes without its set of challenges. Right? So if I had to ask you briefly the top three challenges that are ailing the integration of AI, where AI is placed in the BPM sector, what would you say?
The topmost and if I have to talk in terms of the clients Yeah. Clients, The siloed technology, and that is what I mean by that is if you have, say, for example, 10 different systems, and those systems do not speak to each other, then the art of possible of what AI can drive for you is very, very limited. Okay. So integration of technology is is the biggest challenge right now, and we term it as, is your organization AI ready? That's that's point number one.
The point number two, because AI survives and thrives on availability of data. How mature is your data infrastructure? How mature is your data availability? Is it for example, what we mean by that is if if we are trying to figure out whether my invoices get processed on time, is it as simple as just the way I just said it, are my invoices getting processed on time? Or I'm actually speaking to five different systems, and my data lake maturity is there that it can provide me that data with real time inferences. Yeah. Is your data lakes mature? That's the second point. The third most important and and they're not necessarily in that rating order, but the most important point is, am I doing it responsibly and ethically? Correct. I think that particular piece is for most of the client organizations and our organizations also very well. It's something which we are still dabbling into. Okay. Taking into euro laws. Every country is coming up with their AI, laws.
Yeah. Are we doing it responsibly? Are we doing it ethically? I think these are the questions which if we are the AI providers or because we are in the BPM industry and we are the solution providers for all of that, then we are the piece, especially the third one, is something which we really, really need to take into account before we even embark on this journey. What did so we are in that in that situation of guiding our clients what it means to do AI responsibly and ethically.
Integration of AI In the BPM Sector –(41-39)
Okay. Lovely. Suhrid I want to ask you a different question. Right? There are companies that are as established as the ones that we're talking about, and there are organizations that are still beginning.
So what will be that one advice that you won't want to give businesses and organizations that are just beginning the integration of AI in the BPM sector? Okay. And let me forget your question. Right? You're talking about the client organizations. Right? Client organizations. We're already matured and we are probably starting. Yes. I actually, you know, see that as an opportunity because and if I if I look at it, people would agree that if I look at BFSI as a as a block business, like, banking, financial services, they are probably way ahead in terms of investing in technology. Right? Yeah. Now there are other industries and, for example, we have a lot of clients in shipping and logistics and travel who are probably not who do not spend as much or haven't spent as much money. We are actually advising them to see it as an opportunity and to live strong. So you don't need to go through a to b to c to d.
You can jump from a to d directly given what you have to do. Right? That is one advantage you are talking about. The second thing, Ridhima, is that, look, earlier, we are probably constrained by a fragmented or siloed system, and the shared doctor wrote it. Now instead of trying to integrate so many back end system, which is a very difficult proposition.
Right? We can bring something in the and I call it more of system of engagement, a system of records. Right? So for the industries, you know, who have probably not invested and still wants to operate on the legacy system of imports. And, obviously, you cannot go and tell them, okay. Change your everything. Change upgrade. Because, you know, you won't have that budget. You won't have that kind of Correct. Luxury. So we now have the ability to run a process end to end at the system of engagement layer. Say, take an example. I'll stay with shipping. Right? We have built a platform that can process the shipping document end to end.
You know? Okay. All the routing, all the rules, all the logistics, all the commercials, and all the in a different port of interest, different country regulations. Right? Having done that, then we can actually push it back to and that's where the integration plays a major role. Yeah. So you know what is what should go into the financial systems, what should go into the whole logistic system, what should go into the, the regulatory systems and all that stuff. Got it.. That is what people are doing that instead of feeling that, okay, you know, I'm not interested so much and I'm lagging backwards. Yeah. We are trying to do a leap frog. Okay. To the last point, is end up the not in the challenge, we're end up with opportunity to see. No one can say that, you know, I'm fully integrated. Right? And one of the Obviously. Reasons for AI to be successful is, you know, once you have the integrated view and a clean dataset. Right? Yeah. They can only work if you have the clean dataset if you don't have that. So I think that is the other thing that almost Okay. There are some higher than I should recall, some lower is Okay. How do you bring that integration and, common view of data Sure. To benefit from the AI models that we have.
Conclusion Emerging Tech in the BPM Sector – (44:45)
Okay. Lovely. We have completely run out of time, but I'll give Mythily the final word. Mythily, let's end it on a positive note. So briefly and quickly, I want to understand one trend and emerging tech that you think will dominate the BPM sector going forward.
I think, first of all, I think it's a huge opportunity for the BPM industry. While the segments, retail segment is an ecommerce segment is changing dramatically, automotive segment is changing dramatically because of AI and GenAI coming in. I think it's a big opportunity for all the organizations in the BPM industry in India itself to change our narrative, to say that we are the, AI bed for the world. And probably calling it AI ops or AI enabled services, That is an opportunity going forward. And, therefore, the tech will be the different type of tech or platforms that I need to have to be able to deliver to that service, a different type of delivery methodology to deliver to that service for the clients. And if we are able to do that, I think our narrative can change, and it can open up huge opportunities. And I talked about, hundred million jobs for us next in three to five years. I see that as an opportunity. And that's something for us on the table to leverage as, the BPM organizations in India.
Lovely, Ladies and gentlemen, unfortunately, I've completely run out of time. I know if we had couple of more hours, we would have been talking about the other aspect as well, but it's slightly difficult to crunch all of it in the given time. But it has been an absolute pleasure hosting you all and to hear such great insights with case studies from all of you. So big thank you to all of you. We hope you enjoyed the conversation as much as we did, and we hope the viewers will enjoy and, you know, eventually have some learnings from it as well. So thank you so much. Thank you. Thank you, everyone. So this is to our viewers. We hope you enjoyed this conversation as much as we did. We hope you understood the journey of AI as far as India is concerned and, of course, that bright optimistic future that we are all looking forward.
With that, it's a wrap from my side. Thanks a lot for watching.
