#111, President, Global Head of Consulting @ Hexaware, Arun “Rak” Ramchandran: ''Disrupting Ourselves with GenAI''

26:06Productivity, workflows, knowledge management, handling customer queries, you know, so on and so forth. The big impact is going to come soon, maybe triggered through the agentic architecture, maybe something else comes after that and we can talk about it. which is going to be business model changes. Fundamental business model changes, new ways of doing things and coming up with products and servicing markets and servicing customers, which currently are not even being thought of. I'll try to provide an analogy, right? Internet happened in the late 90s, right? 95, 96 is when the email, you know, the Yahoos of the world and then some of the dot-com started happening, which is the dot-com boom from 99 to 2000.

26:51But Googles of the world, Facebooks of the world, they all happened three, four years later. Yeah, of course. And they actually took this in a very different direction. I mean, nobody knew there could be social media to connect people when internet actually started. Nobody talked about the power of Google search when you started internet. And I feel that something similar and maybe much more powerful would emerge out of the Gen AI ecosystem and the AI ecosystem. I don't want to just limit it to just Gen AI because Gen AI is a stratum on top of AI. And there are many other ways of doing Gen AI.

27:30I think this is going to create fundamental business model changes and the way products are developed and the way consumers consume. and what is needed to actually make some of these industries function in a manner which is going to be very different than what we have done. I mean, let me just give you an example, right? Even something like self-driving cars, we think about them as having a particular internal structure. There is a driver's seat, there is a passenger seat, and so on and so forth. In the future, with AI, with advanced IoT and whatever the new developments are, the internal structures of how a car is built itself will change.

28:10You may not have the same configurations. Same thing happens for urban planning. Same thing will happen for how healthcare is administered. So I think you'll start seeing a lot of innovation, a lot of disruption, business model changes, which we currently are just starting to get glimpses of. Yeah. No, I didn't think about manufacturing. I suppose AI will find ways to make manufacturing more efficient for products to be put together in a more efficient way, maybe more efficient from an energy standpoint, environmental standpoint or whatever, but it'll push everything forward, won't it? It'll accelerate the evolution of pretty much everything, surely.

28:49Yeah, one is evolution. The second is revolution. The main essence of what intelligent or close to AGI kind of intelligence would be is that it can take some decisions and it can figure out a more optimal way of doing something which human beings may not have thought of. A lot of our structures, like a lot of our business structures, a lot of our consuming patterns are built on human-centric design. Yeah, yeah. And an AGI will not necessarily always think from that perspective. I mean, you know, we all heard about this epochal move called Move 37, right? And you probably heard about it, which is the Go game, which Lee Sedol played with DeepMind way back several years ago, which created a paradigm shift in terms of what people thought of AI.

29:41This was the Go game and, you know, the DeepMind beat Lee Sedol. And what happened in that particular game, I think it was game two or something. And, you know, all the training that it was being given, this Move 37 was something which none of the experts saw. It was not something which any human being would have played. Even the creators of that algorithm were scrambling to figure out why that move actually got played by the deep mind. And then it ultimately won the game. And, you know, suddenly that becomes like a totally unexplained thing, but it helped sort of optimize its winning probability.

30:14I think something similar would happen when all the data sets have been incorporated, all the different kinds of algorithms are, you know, parameters have been put in. a new way of doing certain processes, a very different way of optimizing supply chains, a very different way of configuring transportation networks, very different way of manufacturing products. I think these are things which AI would be able to bring in, which currently I think human beings are even not even considering. Yeah, no, it's a good point. And that will be all for the better, all for the better. So you mentioned at the beginning revolution.

30:50We were talking about evolution and revolution. So speaking of revolution, jobs displacement. It's a controversial topic, Rack. Obviously, we talked about history earlier on. And as you and I both know, whenever there's technology innovation through history, it typically is where there's been a net gain in jobs normally. Just the jobs have changed. So I suppose it's a similar question to what I asked you. Is this different? Do you have concerns about companies automating so much that there's going to be certain swathes of people out of work? We've faced this before, right? We've faced this before with automation.

31:22We faced this before going back to electricity and steam engines and locomotives and so on. One example I think about is the use of automation in telephone switching. You know, when that automation came in, automated switching and PBXs and so on, there still used to be operating stations with multiple people actually doing the individual switching of telephone lines and routing calls. After the automation came, that pool of labor didn't just disappear right away. It took almost, and I was reading it somewhere, it took almost 15, 20 years for the last of telephone operators, manual telephone operators, to actually start disappearing.

32:01Right. So that is the curve through which automation actually worked its way. I mean, you know, we can go back further in terms of agricultural automation and so on. I think what the point I'm trying to make, there will be displacement, but I think it will be different kinds of jobs. Yeah. Which will get created. Of course. And this time, the displacement is coming for knowledge workers. Automation struck at the blue workers, the blue-collar workers. That's the first time ever, though, isn't it? All technology innovation in the past has really displaced blue-collar workers. Now it's the white-collar workers who are getting displaced.

32:37That's a first. So I think one way to look at it is AI will augment human capabilities. So people who can work with AI will be better placed than people who refuse to adopt AI. So think about it as in the future, there will be people who have a mastery of AI tools and can use that to augment their work, who probably will be more successful than people who don't use AI. That's the way probably to look at it in the current stage. But then there will be new roles created. I mean, we are already seeing new roles which are getting created, right? I mean, we are seeing that some of the basic coding is getting done by AI.

33:15So maybe the junior developers, they are finding that maybe their skills are not necessarily required. But as you become senior, as you understand features, as you understand product roadmaps, your utility can come in different ways. Interfacing with the business teams, you know, helping create, validate the results and outcome and so on. So I think there are different jobs which are getting created and different fields will open up, which we didn't have in the past. How do you think generative AI will impact things like the SaaS market? You know, obviously it's an enormous part of most economies.

33:48this SaaS business is everywhere. If you can go onto a model and essentially build a piece of software for your business like that for free, and extract it from the model and run it locally, you don't need SaaS products anymore, do you? So, I mean, obviously, there'll probably always be a place for SaaS products, but do you envisage an enormous chunk out of the SaaS market over the next 10 years because of people creating their own software using these models? Absolutely. Absolutely. I think the SaaS market will be inverted. Software as a service is the original definition and service as a software is the new definition.

34:29Right? So you're basically saying that some of these services will be delivered through software, which in today's language, it could be AI agents or co-pilots or queues and so on and so forth. So they will actually do some of the individual pieces of work. So as long as you have data foundation, ready. As long as you have access to the right kind of sources, data sources and data pipelines, what does actual SaaS package, current like the ERP packages and all do? They basically specialize in certain kinds of enterprise data, system of record, system of transactions, pull that stuff out and make sure that it's available in a particular format.

35:08Now, AI agents, if they do work as intended because they're currently still facing peathing issues, in terms of being able to access that kind of data, understand the categorization, understand the domain context, right? So there is a business logic, there is a domain aspect to it, and are able to connect the different parts of the workflow in a manner which obeys the need for an enterprise backend the way we are currently used to, right? And that is one stepping stone to having software as a service, right? And service as a software. That's the other way to look at it. Some of the services will be delivered as software agents.

35:48Has that already caught on, service as a software? I've not heard that before, but it's a concept I've been reading a lot about recently. I hadn't really thought about the impact on the SaaS world. It was Chamath on the All In podcast, which I'm sure you're familiar with, living in Silicon Valley. And they did talk about the SaaS, SaaS being this phenomena between the birth of the internet and the birth of AI, this phenomena which happened in between, which is kind of going to get, it was born because of the internet, it's going to die because of AI, which I thought was, that sort of blew my mind a little bit.

36:19And I just thought, wow, because we all just thought SaaS was here to stay and that was just the future. But I think AI is such a game changer. What is going to happen is, you know, the SaaS was really powering the large companies, the large corporations, because they were so complex and they needed all these rules and all these interfaces and there's ways to keep records. If AI can help manage that data and the workflow, which is still a big, big challenge right now. It's not that it's easy to solve. Just like, you know, AGI. I mean, you know, just enterprise intelligence, right? I mean, I'm talking about, you know, moving away from just human affairs and so on, just focusing on large enterprises, JP Morgan or Goldman Sachs or Delta Airlines and so on.

36:59They have their own enterprise entity and identity and DNA. and there is a lot of intelligence which is around. If AI can use that to create the services, you really don't need big backends for that. Yeah, sure. How are ExaWay leveraging generative AI across the business then? In terms of your own utilization of AI to deliver services, can you tell us a little bit about that? Yeah, absolutely. In fact, one of the pillars of our vision for AI and Gen AI, first of all, our stated intent as a company is to be AI first. Now, that by itself is not differentiated in the way that a lot of companies are talking about.

37:38But we're actually living through it. And when I say living through it, one and a half, two years ago when we formed this unit, one of the goals was not just to take services to the outside market, like, hey, we can do prompt engineering or we can do this. It was to actually adopt Gen AI in our own services, which is to disrupt ourselves. Now, what does Hexaware really do? It does three things very well. We build software which powers platforms and products which our clients want to take to market. We help modernize and transform the technology's landscape and tech stack, whether it's data warehouses, whether it's infrastructure, whether it's applications.

38:17And we help create the right tools and platforms for them to run this and operate this on an ongoing basis. And we use data and AI to power all of this as a foundation. So in every one of these services, which is sort of bread and butter for a lot of technology services, Gen AI is actually going to impact the way the service is being delivered. And we said, look, we know this is going to fundamentally change the economics of what being a technology services company means. When you're talking about coding or when you're talking about testing, you currently do it manually. A lot of it is done manually and a little bit of automation.

38:57It's typically time into material, P into Q. Now you create an AI-driven model which can help obviate the need for so many people, which can reuse what you already have, which can create an automated test framework, which can test the code which has been generated, gather the requirements, so on and so forth, or even service desk and help desk, right? Proactively solve problems and so on. You actually are creating a different economic model for delivering these services. So we invested, we have invested in certain tools. We call it RapidX and Amaze and Tenzai. These are platforms and accelerators which can actually deliver some of these services like coding, like requirements gathering, like testing, like cloud-based transformation and migrations using AI and Gen AI, powering them.

39:44So we said we will disrupt ourselves. So that's been one of the big pillars of our Gen AI strategy. Of course, the second one is to disrupt and work with our clients in their transformation journey. Right. So this is where our service catalog and solution accelerators and the fact that we have tech trained people comes into play. And we are actually taking this whole thing right from getting your data ready, helping create the AI models and fine tune and integrate the AI models, which are already there, helping create the UX layer for use of these AI apps, manage these on an ongoing basis through ops and so on.

40:19So that's the other part. And then the third one, the third and final part of our vision is essentially identifying some big bets for certain key industry segments. We have identified a few. We believe that these will actually be industry game-changing solutions and go-to-market IPs that we will have. We will work with some partners, whether it's Microsoft or AWS or Cohere and so on. And we have some relationships which are now happening with NVIDIA and so on. we will actually be able to take a few key slivers of these industries and transform them using Gen.AI. Wow, okay. Well, what real-world AI applications are making the biggest difference for your customers at the moment then?

41:00We are working with customers primarily on the productivity bucket, as I was telling you, right? And a lot of customers want to get started with making sure they have a fundamental underpinning of the right secure implementations of whatever model they are using. whether it's a GPT-based use case or it's knowledge management or document summarization. First of all, they want to have security. Make sure enterprise data doesn't leak out. Make sure there is no external data which is poisoning their database and so on. I think so that's one, right? So a lot of people are experimenting with it. The interesting use cases we are seeing, customer service for agents, right?

41:38So an insurance company using it for real-time transcription, cross-selling, understanding customer context, the policy information, what is missing, what kind of question is being asked and being able to provide those answers in an intelligent manner. And using voice to text as part of some of the interface, which is needed. Hospitality company is using us to actually transform their event management. Hardware manufacturers used us for product listing and helping create better product descriptions which can stand out on the website when compared to some of their competitors. and, you know, it's written out in such a manner.

42:17Those are becoming more and more prevalent as we go along. Yeah, no, absolutely. So AI regulation then is tightening globally. How are companies reacting to that, implementing AI capability in a fair, explainable, reliable way? What sort of considerations are you having with your customers in terms of implementing AI in compliance with regulation? Are you having any problems with that? Are you having any headwinds, RAC, or is it quite a seamless implementation in your experience? Some of our customers are very wary of adopting it in a big manner because they feel there is going to be certain issues with regulation.

42:58So they're getting their legal team involved, their ethics teams involved and so on. These are big corporations, big financial institutions with a lot of compliance and regulatory requirements. Technology-wise, there are ways of handling it which have started emerging, whether it is use of RAG, so just use your own data sets and use LLM only for the language part of it. There are smaller models which are coming into play, which are just focused on specific slivers or functions, which will become a little more prevalent in the future because of lower costs, both on the training side and on the inferencing side.

43:34So there are customers who are beginning to figure out lower cost options, more secure options which are contained. And that is probably going to be the future. I mean, what DeepSeek and so on have really shown is the art of the possible of trying to do AI implementations at a lower cost without having to spend like$60 or$200 a user and so on that was seen as the sort of the established payment benchmark. And that's being questioned, right? But that's why people are doing POCs. that's why enterprises are still taking some baby steps because one of the things they really want to be sure, especially if you're a financial institution, a healthcare provider, somebody who manages that kind of data, right, which is private and has to be regulated and so on, is whether there will be leakage, whether there will be injections which can disrupt some of their practices and privacy requirements and compliance requirements.

44:32So that is being really looked on seriously right now. Yeah, sure. AI talent. I've heard you talking about the AI leadership vacuum before. There's clearly going to be, and there is, a shortage of AI talent. How should businesses, enterprise organizations perhaps, or SMEs even, how should they respond to this lack of talent available within the AI space right now? And what are your customers doing to upskill their employees in light of this new innovation? I think two key ways to look at it. One is catch them young, right? I mean, there is more AI talent in the younger set, folks who are just coming out of college and just graduating because they are used to some of these tools and technologies and they've got more recent exposure.

45:14So I would say, one, catch them young. And two, focus on retraining. Some of this is not complicated stuff. I mean, we're basically saying the new interface programming language for the future will be English, right? So we are talking about tools which can be used by non-engineers as well. So provide more training to your existing set of people, even business users and leadership and functional operators, rather than just focusing on your engineering and tech team. Yeah, of course. Some of these tools will actually make their lives easier. What type of training are we talking, Rak? Are we talking sort of AI ethics?

45:50Are we talking prompting? Yeah, and this is what we did, right, in our organization. So we have 32 ,000 people, and we have been able to train around 90 to 95 % of our tech workforce into the basics. And we curated a course for ourselves. We went to Coursera, we went to Udemy, we looked at different kinds of courses which are available and we came up with a list of 30 to 40 hours of training program. We said, this is like fundamental. Everybody should spend this to get level one proficiency and we tracked it. We had a team which was tracking the training and the compliance for that. and even we had our management team members trained on it.

46:27I personally conducted some training for our leadership and for some of the clients that we have done. We created three layers of this. We said, this is the basic, this is the intermediate and this is the advanced. Then we created function-specific trainings and we got an outside company to help us with, you know, sort of rolling it out and they had a platform for doing it virtually as well. And we said, okay, all the delivery heads, you need to get trained on these kinds of things. All the client partners and account managers, you need to get trained on this. And it's been an effort over the last one and a half, two years, but it's been very successful because people want to learn.

47:01Believe me, I think if you get the right set of people, tell them why this is important, I think people will take time to figure out new technologies. And that's the hallmark of a growing organization, of an organization which knows that AI first is where the future is. Yeah, no, absolutely. So what are you most excited about, Rack, in terms of the future of generative AI within business? And also, what are you most fearful of? I'm really excited about the evolution and the revolution that AI can bring about, right? From automation to augmentation to amplification of human capabilities to ultimately being autonomous in many ways in the beneficial sense.

47:40I mean, whether it's the agentic thing, which will go on further, the AGI, or the embodied AI where you get the robots and the AI together and are able to do some of these tasks in a more intelligent manner. That's Skynet, isn't it, of Terminator? When you've got the robots and AI together. And that's really where it could go wrong as well, right? If not controlled properly, if not supervised, if we don't have the right ethical guidelines, the right regulation, the right testing, the right safety guardrails, then things could go off track. I mean, I'm not just talking about enterprises and business.

48:14It could go off track in many other walks of life for us. So I think there is a danger if we don't manage it properly, if we don't control it, or if you're not sure how it's going to pan out, we might let it go out of control. Yeah, of course. You've just done an IPO, a 1 billion IPO, biggest IPO in India in 10 years for a services company, which is incredible. And congratulations to you and your team and everyone at Hexaway. That's incredible news. Looking back on your career now, if you could talk to your 21-year-old self, What would you say to that guy? What advice would you give to that person?

48:50You know, one of the things which I tell people, because I also run our future leaders program at Hexaware, which is where Fresh MBA graduates, and I'm a chief mentor for them. And I do it as one of my other hats I wear. I tell them, I think, essentially, it's that life and career is a marathon. It's not a sprint. Yeah. Right? Things will change as well. Prepare for the long haul. Be patient. Everybody, right? irrespective of their origin, status, whatever, ultimately over a period of time, 10, 20 years, will get to their potential if you apply yourself. If you have the right attitude, if you apply yourself well, your own talents and your merit will come through.

49:30You may find something starting off, oh, I have a wrong boss or I didn't get the right role to start with or whatever. But those are temporary aspects in one's career. Yeah. Right. You will find your niche, you will find your track. And over a period of time, your talents will get recognized and so don't fret, right? I mean, but I think very important, people think that they need to have an attitude and sometimes they just burn themselves out sometimes because they think it's a sprint. And I would advise that save it for the long run. It balances out over the long run and focus on your strengths, right?

50:05Don't fret too much about your weaknesses. Have a good attitude to people, the people you work with, your bosses, your circumstance. And I think finally, most important, is life goals can change, right? Your situation could change. So be a little nimble and look at opportunities as they come rather than having a very fixated sense of where you want to be and what you want to be. Sometimes new opportunities can come which can change your trajectory and that's fine. Absolutely. And that's what it's all about, isn't it? It's taking the opportunities that come your way. Look, I've really enjoyed chatting, Rak.

50:39Thank you so much. Some incredible content we've got here. Let's finish on this one. You've just gone through the IPO. what next for Hexaway? I'm focused on the party tonight. Why not, eh? It's going to be a big party. It is going to be a big party. But I think we had a very fun, festive event today morning. I think the whole management team was there. Our stock shares popped a little bit, which is great. We are focused on the future, right? So, you know, in short term, stock and share valuations can go up and down. That's fine. Our entire leadership, our CEO or board is focused on making us, I mean, currently our valuation is close to$6 billion.

51:17We want to be$10 billion in valuation, which means that we want to be$3 billion and above in terms of revenues. And we are currently around$1.5 billion in revenues. So our focus is next three to four years, we want to double our revenues and get to a$10 billion valuation. And we want to obviously be an AI-first company, which is actually disrupting this industry using AI. Rakh, thank you so much for your time. It's been a pleasure. Thanks for coming on the Tech Leaders Podcast. Likewise, Gareth, thanks for this opportunity.

51:51Oh, that was fab. I love that. So one of the most eye-opening parts of that conversation was about the impact of generative AI on the SaaS world. This idea of shifting from software as a service to service as a software, as Raksa, you know, it sounds unusual, but it really makes sense, doesn't it? You know, imagine being able to create custom software with hardly any technical skill, just by inputting the right prompt, this would be truly transformative and would transform a massive industry. And that is inevitably, it looks like that's where we are. So frankly, I don't feel like this is getting enough attention.

52:29If I were working in the SaaS world right now, I'd probably be looking over my shoulder and looking to get on the right side of this innovation. But I'd love to hear your thoughts about it. So please drop a comment or a review. Let's start a conversation about it. Thank you so much for tuning in. I really hope you enjoyed the conversation. And please don't forget to subscribe and engage.

52:56This episode was brought to you by Be Digital. Be Digital support leadership teams to optimize cost and get more out of technology investments. Be Digital and the team have unrivaled expertise with technology license management and data remediation and are therefore perfectly positioned to help prepare organizations for AI technology capability. And on the last point, BDigital have developed a cutting edge AI readiness assessment, which provides tech leaders with a platform they need to make well-informed decisions about AI adoption strategy in 2024 and beyond. Go to BDigital UK to find out more and get in touch.

53:41Thank you.