LIVE: Infosys Investor AI Day 2026 | Infosys Unveils AI Growth Roadmap in Bengaluru

At our Bengaluru campus. A special hello to everyone who's joining us via the investor relations webcast. Today's proceedings are being recorded and the audio transcript and the presentations will be made available soon on our website. So, we request you not to take pictures or record the sessions while they're going on. Before we begin, I have some important housekeeping announcements. On your tables, you will find the agenda, an important information sheet, and the feedback forms. Please follow the timings in the agenda to help us keep the day running smoothly. The important information sheet also has the Wi-Fi details. We request you to fill in the feedback forms after every session. If you need any assistance, we have volunteers around wearing Infosys ID cards. Please reach out. Please take note of the nearest emergency exits on both sides of this room. They're clearly marked. In case of an unlikely emergency, please follow the signage and volunteer instructions. Restrooms are also outside this hall on both sides. Signage and volunteers will guide you. Kindly access charging stations on both sides of the room for charging your electronic devices. Please note, we will not be taking questions at the end of each session. There will be a question and answer session at the very end of this event. So, please hold on to your questions till then. For departures in the evening, airport coaches have been arranged for the ones who wish to avail them. Please note that the first coach for airport transfers leaves at 4:45 p.m. A second coach will leave at 5:00 p.m. Those who have flights at 8:00 p.m. are requested to take the 4:45 p.m. coach considering famous Bangalore traffic. Please inform the help desk just outside this hall if you need an airport transfer earlier than 4:45 p.m. And lastly, I request you all to put your devices on silent mode. Thank you. With that, let's start today's program. For our first session on tech transitions, why is the AI transition different? Please welcome Nandan Nilekani, chairman of the board Infosys.

Thank you and great to have you all here in these tumultuous times. Today I'll talk about tech transitions. I have had the fortune or misfortune of being in this industry for more than 40 years and I've seen a lot of transitions. So I thought I'll talk less about that and more about why this time it's different and what are the implications of this transition. Now, safe harbor clause. Now we have seen technology shifts for centuries, whether printing press or telegraph, but over the last 60-70 years we have seen a much faster change and PCs, cloud, gen AI, agentic AI and so on. So change of technology and the speed of change has been a constant for many decades now and each time there's a change, the way we address that change has been different. So we went from mainframes to mini computers to PCs, client server, LAN, web computing, mobile, enterprise apps, big data and each time we had to think of it in different ways. How do you think of it in terms of making it globally available through internet or how do you do enterprise IT? So each time there was a tech transition, it had certain implications for us and firms like Infosys had to deal with what was new. So we are used to the fact that each time there's something different. This time the AI transition has been much faster than earlier transitions. If you look at the number of years it took to reach 1 billion users, internet took more than 10 years. Smartphones took 5 years. AI is taking a couple of years. Now you have to realize that the AI speed is because of the first two things. The internet was already ubiquitous. Smartphones were already ubiquitous. It therefore allowed people to distribute a ChatGPT or Gemini or Claude very easily. So in some sense the speed of AI is also because of the infrastructure of the previous era. Now what has happened this time is that this is a much more fundamental change to the way businesses will operate. So this is not a layer of technology. When smartphones came we could build applications where instead of having a PC you did it on the phone. It was like putting a front end to an existing application. When cloud came we could do a lift and shift, you could take the app from your in-prem and move it to the cloud. So you could do a lot of things to get going. But this time it's not that. This is a fundamental change in the way we do things. Obviously there's a technology dimension and it's all about having AI native architecture but there's a whole business dimension to this. We cannot run business the old way and businesses have to change, the customer journeys have to change, all those things have to change. It's a huge challenge for talent. Talent will have to deal with a world where writing code will not be the goal. It'll be actually making AI work, orchestration and those kind of things. So the jobs will change and operating model. How do we make this at scale? How do you get a firm with hundreds of employees to change all the things and make it work? And of course our mental models have to change because we worked, you know, technology was always deterministic. You said a plus b equal to c. So no matter how many times you said a plus b, the answer was c. In this AI world, you know, every time you give a prompt, you'll probably get a different answer. And therefore how do you deal with this non-deterministic world but how do you make sure what you build has the robustness, reliability and resilience of the deterministic world. That's what the challenge is for everybody. So this is a fundamental root and branch surgery of the way business is done which is why this technology transition is so dramatically different from anything else that we have seen. Now one clear learning we have is modernization of legacy systems cannot be deferred anymore. What happened over the last 60-70 years is people would not replace the legacy systems. They just added to it. So if you go and look under the hood of a large enterprise they will have mainframes from 1960. They'll have mini computers from 1980. They'll have LAN from 2000. They'll have all kinds of things and all coexisting in silos. That is over. If you really want a firm to take advantage of AI, you have to fundamentally clean this up. So this is a massive, massive cleanup job which everybody's dealing with. There are reasons for that. One is the financial drain. Many large companies are spending 60 to 80% of their IT spend on maintaining systems. There's no business value out of that. They want to go from 60% or 70% maintenance and 30% new systems to 30% or 40% maintenance and 60-70% new system. They want to flip the way they spend money. But they can't do that without that fundamental cleanup they need. Moreover, many of these systems were designed in an era before you could have online attacks and so on. So security breaches which you see every day are just going up everywhere and there are more state and non-state actors who are getting better at it using AI. So security is a huge problem for everyone. We have seen so many cases in the last few months and because the data is all in silos, you can't even innovate fast. So there are fundamental structural issues today we have. So demand side is absolutely demanding modernization. But the good news is for the first time because of AI we have the tools now to do modernization fast and very quickly and at a much more economic way. So we have a huge demand and we have the ability now to do it and perhaps our team will talk about that. So fundamentally accumulated tech debt over decades must be paid. You no longer have the option to defer this and this is a huge, huge requirement and obviously it's a huge opportunity for us. Now the other thing which is there is as AI becomes a bigger part of the spend, the balance of advantage is moving towards build rather than buy and that is actually what is, if you see some of the concerns about what will happen to SaaS companies and all that, it's because of this that building applications has become so simple that very often you may just build or you may replace something that you have which you bought and with something to be built. And so we see, and that again actually benefits folks like us because about building, who's going to build it for them? It's going to be us only who will build it for them. So fundamentally it's good for us. And the other thing which is there is that our view is that foundational systems will increasingly become systems of record but the interface will be agentic because agentic interface makes a lot of sense. Agentic interface allows people to produce something which is designed pro-consumer or pro-user and agentic interface enables you to take out the complexity and hide it behind the agent. So the agent is simple to use. It's a very simple idea. Now enterprises will therefore want to put agentic layers on top of all their applications even if they leave the system of record the same and that is something which will be a combination of bought out agents as well as building their own agents because finally the agents have to be composable in a customer journey which is seamless, which is a mix of agents which are your own or from somebody else. Again that requires orchestration and work which somebody has to do. So there's a huge amount of work required once they go towards build rather than buy. Now the other thing is the pace of change is something which obviously we have not seen. We all know about the trillions of dollars being spent and all that but even the technological change, I mean, you know, 2023 foundation frontier model had 100 billion parameters. Today it has 1 trillion parameters. There are only 10 to 12 agent networks. There are 60 agent networks. So this is only going to go up. In the US alone, there are at least five frontier models. In China, there are big four or big five. So this is only going to go up. In India, we have seen so much action and you'll see some big announcement this week on Indian-based sovereign models. So I think this is something. Now there are certain implications of this because if I'm a businessman and I have to choose my technology, how do I make sure I don't make the wrong choice because something which I invest in today may have fallen behind tomorrow. Already people are facing this reality and therefore how do you architect your technology so that you can deal with this rapid change is a very fundamental and structural need for enterprises and again they need help on that from somebody who has done this in 2,000 locations and understands the pros and cons of every approach. But the main thing is that the technology is far ahead of its deployment because of this race and spending billions and some AGI and all that. The technology is moving faster than the ability of enterprises to deploy it. If you look at this chart, you can see that the model performance is going up, but the progress in implementing is not really because implementing this is hard stuff. Fundamentally, it's about organizational change, business change, retraining your people, thinking about non-deterministic approaches, changing your data. So, it's no longer in silos. So, fundamentally, we have a situation where there's a deployment gap between the power of the technology and the capacity of businesses to use this. So, if you guys think that some better product has come, nothing's going to happen because the problem is here, not there. You get it? It's about how fast companies can implement. So you have to look at that. And this is, we call this the deployment gap. But this is actually a concept by Professor Clayton Christensen at Harvard 25 years back. He called it technology overshoot where technology gets ahead of the need. And in fact he argues that that's how newcomers come because newcomers can then launch new products that are not as sophisticated but good enough for customers. And Satya in his recent blog talked about model overhang which is the same idea. Fundamentally the tech will keep getting better and better because billions are going to be poured into it. There's massive competition but enterprise deployment is not going to go up and this deployment gap is what we can help to address. So again it's a very important point. Now I think talent transformation is huge. It's not that you will not need talent, you will need talent but it'll go from QA testing or development, we have all kinds of new roles: AI engineers, forward deployment engineers, AI leads, forensic analysts, data and so on. Fundamentally the challenge will be how do you take your workforce and make sure that they are reskilled and ready for the new business and that's really the challenge that all the firms will face. And this is something, so there will be roles, now the way you hire will change, the way you train will change, the way you deploy will change, all that is going to happen and I think we'll have sessions on that but fundamentally there will be a need for people but they'll be doing different things. Also a lot of the talk of productivity is greenfield. Writing greenfield is not a big deal. I can take a tool and give it to a kid and he'll generate a million lines of code. But that's not the real world. The real world is the fact that companies have trillions of dollars invested in their systems. They have technical debt. They have data silos. They don't have documents. Somebody was telling me the other day that there are some old systems and on contract they have guys as old as me, 70-75 year old guys because nobody else knows what the hell is going on. And then when there's a crisis to be sorted out they're pulled in from Phoenix or Florida or wherever they are and they have to solve a problem and nobody else knows how to solve. So we have that kind of situation out there, undocumented dependencies. So taking brownfield systems and modernizing them is a hell of a lot more difficult than doing greenfield development. And a lot of us get biased because all the guys who talk about productivity are talking about greenfield development. And therefore getting the large enterprise organizations productivity going. Also AI implementation requires laser focus. The very fact that you can generate stuff means you can generate slop. You know, in fact five years from now there'll be more AI legacy system than any other legacy system. All kind of stuff will have been generated and we'll have to clean that up also. So and if an organization is not, or you can have this fake productivity, let's say there are two guys and they are having a fight, one guy will draft an email which will be one paragraph, he will give it to AI to make it into a 10 paragraph email because he wants to impress the other guy, the other guy will take the 10 paragraph email and summarize it to one paragraph so both have used AI but what have we achieved? Nothing. So how do we make sure AI is used and therefore you need usage guidelines, you need quality gates, you need explainability. So how do you make sure that AI investments lead to real performance in productivity and not just some make-believe stuff? This is something which is very important. So what still matters? First principle thinking. One of the things when we train people is they have to learn to do this without tools because all of us learn to do this without tools. So when we got the tools we knew how to use those tools better. But if you start by teaching them tools then everything is a black box. Then you know it's like the guy who never knows how to calculate because he was born with a calculator. So first principle thinking is very important. All the more important as you think about strategic transformation of large enterprises. First principle thinking is very important. Second, understanding enterprise context. Every company is different. Every company has a different legacy. Every company has different systems. Some of them have come from acquisition. Some have come because they have five business units all buying their own version of technology. All kinds of reasons. Everybody has a complex estate of systems and context is essential to being successful at AI deployment and each context is different and it's dealing with these contexts that is the hard part which is where again we believe we have a way of doing that. I'll give you an example. The self-driving cars. The first DARPA challenge was 2004. The first time they rolled out was 2007, 20 years back. Then everybody said oh yeah by next year we'll have self-driving car. It's 20 years later and all we have is a few cities in America where there's self-driving cars because the context is different, every city is different, every road is different and by the time they come to Bangalore it'll be 2047 because they're dealing with Bangalore traffic will be a different level of context. So enterprise context is so important and that is something which cannot be done by a tool, it has to be done by capturing implicit knowledge and making it explicit. Agnostic design, again what I mean here is don't get locked into a tool because that tool may no longer be obsolete in two years. So how do you design for agnostic that can choose any system. Getting the house in order, we talked about removing technical debt, we have to go make the house in order for this whole thing. And then massive change management. You're changing organization, business, sector, people. I mean this is like unbelievable change management. So unless you have leaders who can do effective change, nothing is going to happen. It's also about strong collaboration across firms. So firms have to, because all the knowledge is implicit in the heads of different people. How do you make that explicit in one customer journey? Focus on productivity. It's not about using AI tools. It's about productivity out of those tools. Otherwise, you'll get false productivity which leads to more complications. And then this is an engineering game. AI engineering is a whole way of doing things and that is part of your change management and transition that you have to do and that's a big thing. So my view is there is no opportunity gap. If anything the opportunity is bigger than ever before. So don't get distracted by that. You can, you should still ask the question, what is the firm doing to take advantage of this? What is the firm doing to transform its talent for this new world? What is the firm doing to design the services and products for this new world? What are they doing to tell the customer in a way that it resonates? What are they doing to make sure that the front-end conversations with clients are done properly. These are all the issues and I'm sure everybody will not execute the same way. So there is an execution risk in doing that. So it is not an opportunity risk, it's an execution risk. You get it? And therefore the balance of assessment is how do we know that each firm has the execution plan ready to get to where they have to get. Are they able to do it well? Are they able to do it with speed? Are they able to do it with scale? Are they able to do it with new mindsets? That is really the question of the day that all of you need to ask. So I'm hoping that today you will hear from our team and it'll give you some reassurance that we are on the right track. Thank you very much.

Thank you Nandan. For our next session on the AI services opportunity, please welcome Salil Parekh, chief executive officer and managing director Infosys.

Thank you Salil. For our next session on the AI services playbook which is by three leaders, I would like to welcome Satish HS, chief delivery officer for the first part.

Let me set the context before we deep dive into our Infosys AI playbook. Every tech shift whether it is PCs, internet, cloud, digital, each one of them led to rewiring of enterprise work and workflows and AI is the next rewrite. Here's a typical enterprise landscape. Sounds complex. This is one of the simpler ones. A typical enterprise will be far more complex because of the scale, because of the fragmentation internal and external, the heterogeneity and the divergence of the operating model and the regulation with which it operates in which is by design by the way and of course technology debt. So AI in enterprise is not just about low-hanging fruits like localized efficiency or user productivity. Integrating AI in an enterprise is not just a software upgrade or a plug-in. If it was that simple in such a huge complexity, no surprise that AI projects fail. Then what is it about? It's about harnessing the full potential. It's about reimagination of systems, processes, and deeply embedded ways of working and rewiring legacy power structures. So, Infosys has done this enterprise rewrite with each of the past tech shifts and we're now doing the next enterprise rewrite with AI. So let's look at how is the enterprise tech stack changing. Typically it consists of the systems of record which deliver deterministic programmatic capabilities which are codified, structured processes and enforce policies and this delivers governance, accountability and compliance. We have systems of intelligence which facilitate engagement, collaboration, transactions. But usually it is the humans who engage with data and the workflows. Above the enterprise stack lies a vast non-deterministic or non-programmed flows. These are unique, unstructured. They need novel problem solving. It needs experience, gut feel and is usually handled by humans. This is the layer which is underserved today and this is ripe for AI transformation. So it's a myth that enterprises need just two layers which is AI and data. It is enterprise and algorithm. How should we harness AI? Then AI is not the endgame of tech transformations, it's just another one but it's a giant leap in raw capability and it's not system complete. So we need a multi-layer transformational approach and purposeful orchestration to harness the potential of AI. This is why AI diffusion in an enterprise lacks the rate of AI adoption and it needs time. So how is AI transforming the enterprise stack? What is this enterprise rewrite about? The co-intelligence of humans plus machines will be seamlessly shared across the layers so that every layer gets reinforced within the enterprise stack. The systems of record need acceleration so that the business and operational processes can become more efficient. The systems of intelligence needs a seamless integration of structured and unstructured data so that intelligence can be wired into user journeys, business processes, transactions and decision making. It is estimated that 60% of effort in an AI project goes into doing this. The models will come in later and this requires deep industry knowledge and context of the enterprise. Encoding intelligence and AI in the flows leads to more automation and autonomy and this is leading to the development of a new layer within an enterprise stack which is what we call as the systems of cognitive work within an enterprise as an outcome. Humans will shift from acting on data directly to a governance and oversight role on flows and decisions. All the new AI tools that keep coming at us at fast pace will get plugged into the systems of new cognitive work and it'll accelerate the reimagination of an enterprise or its flows and decisions. Infosys will unlock tech debt and complexity and harness the power of AI to be enterprise-grade and bridge the evolution adoption gap and expand our addressable market. So how do we monetize this? Our playbook reflects the structural changes necessary in our industry so that we drive value at the intersection of intelligence, engineering, and domain. With AI, capability is a commodity because it's available to all. Sustainable moat for an enterprise can be created by deep integration in specialized workflows and unlocking unique organization knowledge and context. So every enterprise has got unique data, processes, risks and complexity and AI will not unlock uniformly across the enterprises because of this variance. So our client intimacy and deep understanding of our client context will help us mitigate this and unlock value and also drive culture and change. So we have built an engineering approach in our delivery where we can codify enterprise context which will help accelerating scaling of AI and this also enables enterprises to retain and protect this unique enterprise context within their four walls of the enterprise so that they can keep their competitive differentiation and this is not diffused into the AI models. Our depth in engineering and frameworks on IP and patents will accelerate AI readiness and adoption with our Topaz Fabric and our specialized talent in the form of full stack and FDEs. For a financial services client, they were looking for an AI partner. When we started talking to them, we realized that they have a very strong enterprise AI platform that they have built. But then what we realized was they had an adoption gap and we pulled out our agent control framework as we call it which would address the quality of code that is being generated which would address the AI slop which led to a poor adoption within their organization. So now we are working with them on taking our framework and fortifying their enterprise AI platform so that we can accelerate that journey. We have embarked on a talent transformation journey to build an ambidextrous workforce which is deep in engineering and creative in reimagining work and workflows from first principles. We see new opportunities with domain stack. You know we have over 25 years of industry focus experience and when co-intelligence connects with agentic economy the play of AI elevates.

from how work is done to new outcomes that are possible. So we are invested in building the domain stack powered by our depth and domain and knowledge of how to deeply integrate AI tools and plugins into the flow. We have created strong differentiation in our services track. AI is now integral to how we deliver every service. Our services stack is powered by Topaz fabric. We now have an approach to productize and reimagine work and workflows that will lead to a human plus agent model.

We're also seeing momentum with new deal archetypes: legacy modernization with reduced risk, higher predictability and cost, and accelerated timeline; large deals with integrated ops and tech and transformation wired in; organization transformation encompassing enterprise stack and people. This includes the AI-first GCC approach which we have pioneered in the market. We have also elevated our play to take end-to-end accountability from strategy to actionable roadmap to execution and eventually outcomes. We have expertise in delivering both above the line, which is business value, and below the line, which is efficiency. Infosys is best equipped to deliver enterprise AI ambition with the power of our client intimacy and our AI playbook.

A quick example here is a client, a CPG, who had an ambition of growing the revenue to 7 billion. They came to us to bridge their ambition and deliver an AI operating model so that we'll have an actionable roadmap and execute to it with executive AI value office along with managing risk, governance, and assurance. We used our Infosys IP and built their unified data foundation. We built their enterprise agentic AI platform with the requisite guardrails. This enabled them to rapidly innovate and diffuse AI across their functions.

Today they have 10 agentic AI products in the business across different functions from R&D, sales, marketing. Above the line, with the agent that we developed in research and R&D for product formulation, they now have line of sight to 50 million revenue which they didn't have before. Below the line, we've been able to unlock 25 million cost savings through just optimizing operational work. And then beyond this, we were also able to deliver 40% of business productivity improvement in functions like procurement and marketing. Thank you.

Thank you Satish. For the next part of this session, I would like to invite DH, chief delivery officer.

Thank you, Siman. I'm audible. Thank you, Siman. And good morning, ladies and gentlemen. Thank you very much for sparing your time here today.

As we navigate the landscape changes with respect to AI, I think our priority in the last 6 months to a year has been how do we accelerate our customers from experimentation to really looking at an AI scalable at industry level. Now AI as a technology is as good as what it can really understand from orchestration across systems of record which Satish alluded to, ability to really understand the complex business processes and navigate, and most importantly how does it even get to understand deep-seated legacy systems that are there. And frankly looking at all the complexities that we had with respect to some of the estates that we've been working as well as with the customers, we decided to codify our entire services across six strategic pillars and these six strategic pillars are all integrated in a way. A customer who wants to start the journey of AI adoption has to look at each one of them and see how does it, and then our intent is to take this strategic pillar to walk them through this entire journey of scaling the AI.

Now let me start with the pillar number one, the AI strategy and engineering. Nandan alluded to it. So as organizations are extremely complex, one needs to really look at a top-down looking at how do I really create an AI blueprint, a need of really setting up an AI transformation office. What would that mean? I need to first understand which business units, which business processes that I would be able to unlock the AI value. You can't really be spraying across the AI in multiple different business units. Now that is a very key thing because that's where the value realization framework of Infosys comes to bear to look at how do I map the process to the value that it actually arrives with. Now the second key thing is how do I change the ways of working because you have so many models, so many platforms that have come by. One needs to really put together a very technology-centric AI platform with the models that it needs to go with, the governance that it has to happen, and how do we make sure that we diffuse this particular platform across multiple layers of the organization so that we have one standard way of really looking at how do I scale the AI. The third layer obviously is to really look at the entire governance model that you need to put together in a way this AI is gone very strongly. The last one obviously is to really look at a purposeful selection of an AI infrastructure where you also need to continuously keep an eye on what would happen to my AI ops.

Very recently we've been working with the CEO as well as at the board of Danske Bank where we really started working with them to looking at how do I enable this bank towards the journey of being AI first and digital first. So we set up an AI office along with the board working along with the CEO to really put together a strategy all the way to engineering, augmented that with innovation labs as well as identification of core processes like KYC, the fraud detection, credit and whatnot. So in a way the journey is just not that I would start doing software development but it needs to have a top-down view of really setting up the strategy for an organization.

Now the second pillar, my friends, data is everything that AI needs today. Organizations have multiple pools of data sitting. It is just not structured data. It is in fact even unstructured data, multiple modes in terms of videos, in terms of speeches and whatnot to an extent of about 60 to 80%. And today most of the time is spent on data in all our projects. Data is one which accelerates your AI journey or potentially could also decelerate your entire velocity of how you going to do it. So our frameworks here today is to really look at help our customers try and harness the data, transform the data, bring them all under one uniform data fabric and it just doesn't sit there. At the end of the day data also have to drive intelligence. The intelligence can be driven only by connecting the semantics and the ontology on top of it. And that's a very elaborate process one needs to go through because the processes are different by region, by business units and it has been all codified in systems of records as well as in systems of experience and that's a humongous job. The third layer there is to really look at how do I govern the data because the data fingerprinting is extremely critical as you really look at who would get the access of data. So that in some sense there is a framework if you really look at so in one of the large industrial manufacturers where it was 10 petabyte of data that we had to really bring in all together, harness it as well as create the semantic ontology, today help them actually drive the supply chain optimization by over 20 to 30%.

Now the third pillar, Nandan alluded to this, this is all of re-imagining the entire business processes. Most of the large enterprises today have point-to-point solution. One need to really look at in the context of domain and reimagine the entire business processes and these processes have to be reimagined with respect to how a human intuition works along with the AI and agent and it is extremely critical that every workflow by persona has to be reimagined and has to be codified the way AI would actually come by. Now this also has to be contextual to the business and the regions where you really work by. You just can't take a sourcing and a procurement process and say that it is a domain that it could be applied to every industry. It has to be really be contextualized to the industry and the business that you are really looking at. And most importantly, since it's going to change and touch every one of us, it has to also be looked at how do I operationalize the entire workflow, the technology, and the change management all put together to really help realize the end-to-end value.

We've been now with Toyota Motors Europe working through a supply chain transformation process where we took our industry asset of automotive, on top of it the entire agentic playbook, really looked at by persona of a buyer, a planner, or a customer service agent and really double click to look at what does this transformation really mean. And just to look at one critical process of drop ship which is so critical for an automobile, is to really bring in agent and orchestration there to really take away all the manual work and bring in much better inventory visibility.

Now the fourth pillar obviously is modernization and everybody has talked about it. It's actually the seal of our large organizations today. There is so much of tech debt sitting there. The code is obsolete. There is no written documentations. There are no availability of memes today. And it's not that the customers did not really try to do the legacy modernization, but the ROI and the time it took never stacked up in the past technologies. So what AI models as well as powered by our Topaz fabric today enables is to make sure that we transition some of these large legacy both on the data as well as on process side into the most modern cloud-based microservices architecture. And today it does stack up and you would hear one of the examples very soon.

Now physical AI is something we believe is at the cusp of really accelerating the AI journey. Whatever intelligence that we all thought and built as a part of the digital workflow is finally now moving towards the actual physical objects and this helps the acceleration of AI in lot of the products that we really look at. Some of the key cases that we really look at is in terms of data first, the new product introduction wherein the entire process would be reimagined as well as infused with AI. The products have to be defined or designed with AI in the front and with more and more products with the software bomb being larger than actually the physical and the mechanical, we have a huge play in terms of embedding the AI as we look ahead. The second case is the intelligence today, the real-time intelligence is moving from cloud to the edge. Now this will help accelerate the decision making at the edge which means vehicles, the industry operations, running infrastructure, all of this would actually increase the advancement and usage of AI. And lastly the autonomous systems today, the prevalence in lot of industries as well as areas is continuously increasing. Towards that we see that we accelerate the journey of actually infusing the AI in the physical. And here I also want to draw your attention to the two acquisitions that we did. One in semiconductor which meant the silicon design as well as validation. The other one in automotive directly fits exactly into this particular pillar where it would help us bring more context as well as acceleration in terms of enabling the physical AI.

The last, ladies and gentlemen, is not the least and the most important is the trust and the governance. If I today ask everybody here who has used the AI, I'm sure that all of you would raise your hand. But how many of you really trust the output that came by? I'm not sure whether everybody would raise your hand because there are hallucinations, there are model breaches and there are also governance issues with respect to the new AI act and etc coming in. So in a way the trust has to permeate through all the other five layers for us to really make sure that we have the output that comes in an enterprise which is trustworthy. To me the trustworthy AI in every enterprise would be a huge differentiator. So we want to build the trust to our customers and we want to monetize the trust for our customers.

Now having looked at all these six important strategic pillars, I just want to dwell on one of the cases, Nova Chemicals. They are a large petrochemical manufacturer based out of Canada and US and it's a very asset intensive as you know the industrial operations is extremely complex. If an asset goes down they would have an impact on the top line and the bottom line. In the current context most of what they were doing with respect to maintenance was all manual logs and etc. So we were invited to a program in smart maintenance where we actually brought in the data across their machinery, their OEM manuals, their maintenance manual, the historical data, the log data and etc. to help a planner to really with a simple NLP on a chatbot would be able to guide them on what part of the industrial operations have to really go through a maintenance. And the most importantly we were also able to actually bring in orchestration with agent AI where the OT systems and the IT systems come together. So seamlessly we were able to really move towards actually creating the entire work order process, the planning process which actually moves over from OT systems to the SAP or the ERP that we have and we see the impact of bringing huge planning efficiency, asset utilization and etc. Of course here we partner with Microsoft and we used all the stack of the entire Azure to really make it come to life.

The last case I want to really dwell is about Hertz. I'm sure that all of you know it's a very large mobility organization where we are today as we speak embarking on really modernizing their entire reservation, their fleet management, their pricing and the whole thing which today is approximately 3 million lines of code actually sitting on a Tandem computer. And I would like to hear play the video, let's hear from the customer on what their experience has been and what we've done with respect to this journey.

So where are we right now? We are in a journey to modernize legacy platforms and let me tell you a story about 6 months back. My CIO at that time and we traveled about four cities in 5 days in three different locations. Each and every we actually met four partners and each and every partner was given the same problem: take this 10 pieces of COBOL program and tell me how are you going to convert it. These COBOL programs were running on the NonStop, right? And every I think you might have heard about HP NonStop or it's also called a Tandem. These programs were running on the NonStop.

Three partners, three vendors gave me the typical, oh, we will do analysis, we will do design, we'll do unit testing, we'll do... I'm like, no, that's not of interest. The day we walked into Infosys office, in the first one hour, they showed me a working prototype, working model of the converted code, right? No presentations. I was not interested in PPTs. I wanted to see a working model, right? And that is what actually impressed us to partner with Infosys. So if you think about what we did, they have a platform called IELD. IELD helped us do the documentation.

These programs have been existing for the last 20-25 years, right? As you can imagine, a NonStop has been in operation for the last 25. Documentation doesn't exist. Nobody knows why a program was written. Nobody knows how to test a migrated COBOL program, right? Nobody knows what scale to test it at. So the documentation actually helped me big time to say program A talks to program B talks to program C. It also showed me the database dependencies, the table dependencies. Right? So in one sheet of paper I could now see that a given program what it was dependent on as well as the tables. So obviously we had a part of a journey where we are migrating, we are about 6 months into a journey. Without the platform I think we were taken about four years to convert. With the IELD platform I think we are targeting about 18 months.

So the important point to note is models are there. I think workflows are there but our context of Hertz in terms of what their processes are, how the code has been written, how the existing architecture is and what is the new modular architecture that we need to really help them migrate to is the context that this IELD brings to bear and that is very critical as we really look at this entire legacy modernization. So that ladies and gentlemen summarizes the six value pools that we're talking about and thank you so much.

Thank you so much DH. For the last part of this session I would like to invite Balakrishna, Head of Global Services.

Thank you Simran. So my colleagues DH and Satish talked about AI first services which is new value pool that is being created out of AI. What I'm going to be talking about is AI augmented services. What we mean by that is taking our traditional services and inducing AI in that and we want to be a leader in this space as well and I will talk about how we are doing that. So we have taken each of our services that we actually traditionally provide whether it is application development, testing, modernization, migration, engineering services, operations, business operations. So 20 plus services that we actually provide and we have created detailed playbooks of how we can use AI in transforming the way we deliver these services. When we are doing this, we are partnering with the best of the technology that is out there. We are working with models, leading models from Anthropic, from OpenAI, from Gemini, from Amazon Nova. And in fact, we have also working with open source models, right? Open source models like DeepSeek, Llama, and we have even created our own coding model in this space. Perhaps the only GSI that has actually created a coding model. So we use a combination of these models in our engagements.

So in the Hertz example that you actually saw, we in fact used two models. We used a cloud model that actually generates the code and one of the things about LLMs are they are better in critiquing output than generating output. So we used an OpenAI model to critique that output and that's how we actually improved the accuracy. So one is the models, the second is the tools and in the tools we are working again with the leading tools. GitHub Copilot was the first and it had almost 100% of the market share. Two years back we set up a GitHub COE that was inaugurated by the CEO of GitHub and then we are still ranked the number one GSI in terms of GitHub adoption. Just few months back we got this award as the leading GSI for working on GitHub. But just not GitHub, we are working with Anthropic, we are working with Gemini, we are working on the models from Anthropic Claude, we are working on the new age models, right? Whether Devon from Cognition or also we are working with Cursor AI that you would have recently seen. So we're working with the leading models, we are working with the leading tools.

But then again as my colleagues also talked about, a lot of these models don't understand the enterprise context. They don't understand the standards that are there in the enterprise. They don't know other libraries that are there, other programs that are there. So we have to do a lot to actually bring that context into the tools that we are actually using. Right? So we do that by creating MCP registries. We create a knowledge graph of the enterprise context and we combine that. In the recently the release that we just did with Anthropic, you hear Dario talking about that they need Infosys to bring the enterprise context onto the models and that is what we actually do. In addition to that we have actually created agents specifically for each kind of services. So in application development we have taken the life cycle and said we need agents for requirements, design, architecture etc. And we have created 100 such agents that we are using in each of our service.

In addition to that, we have to create other tools. You saw in the Hertz example, the LLM models can't digest huge pieces of code at one time, right? Some of these enterprises that we work with have millions of lines of code. If you give that to the LLM, they start to hallucinate. So, you have to chunk the code. You have to create graphs, call graphs on how these are actually associated. And that's when you get better output from LLM. So we have created all of these assets that are part of our Topaz fabric which my colleague Ravi will actually talk about in more detail in the next section.

In spite of doing all this we need talent, right? And sometimes people ask me if LLMs are generating code why do you need people, why do you need developers anymore, right? And I think they asked the same question to Boris Cherny from Anthropic because Anthropic is continuing to hire developers. So somebody on X asked him why does Anthropic, if Claude code can actually generate code, why are you still hiring software developers? And his response was that engineering is changing but great engineers are a requirement still and in fact the most important requirement going ahead. And that is something that we also actually believe. Right? The way that we actually deliver code or the way we support applications may change but still you need great talent. And so what we are doing is to take each of our developers and training them on AI. So we have 90% of our developers that have been trained on AI. It'll never become 100% since we are always hiring new people into the stream. But our intent is to actually have everybody be able to use AI in their daily work. In addition, we need specialized roles, right? Forward deployment engineers that will create the platforms that the teams will actually use and then we have created COEs for each of the partnership that we have. So it is a combination of all of these that actually helps us deliver our AI induced services. It is the blueprints. It is the technology, our own technology plus the leading technology that is out there and the people that we actually create.

The way we are going to market is also as you saw in the Hertz example, it is not about PPTs anymore. It is about actual demos and that is what we see creates the impact and that's how we actually go to market in all of our large deals. I will talk about a couple of examples of how we are using it in actual programs that we are executing. Right? What better example than Microsoft who is in the leading edge of this kind of technology adoption, creating the technologies and also adopting this technology. So in Microsoft we have a 360 degrees partnership. What we mean by that is that we go to market with Microsoft. We are one of the big customers of Microsoft. We also are, Microsoft is our customer. We provided services to Microsoft and we do multiple engagements with them. But I'll give you a couple of examples of what we are doing. Microsoft themselves are going through a big transformation. They are going from enterprise agreements, they're going to what they call MCAS, Master Customer Agreements. What this means is that through the master customer agreement they want to eliminate all the paperwork that they had to deal with in the enterprise agreement. They also are talking about evergreening the licenses. So enterprise agreements had only two years time frame. This is perpetual agreement that you can use for multiple years. Enterprise agreements had a minimum seat count of 250. This actually has no minimum seat count. In addition in MCAs you are able to monitor your usage, adapt your usage. You also bill based on your usage. You get a very flexible billing. So multiple advantages for the customers of Microsoft by using this agreement and also for Microsoft itself it actually eliminates because you are not having papers and documents anymore. It eliminates and accelerates the way they go to market and also the operations that they have. Right? On this engagement they have to build the IT system to manage all of this. So, Infosys is actually working with them to build that. We used all the technologies that I actually talked about and we are getting 2x developer velocity and 35% improvement in the time to market.

The other engagement that we are working with Microsoft is on their intelligent cloud. Right? So as you know this includes Azure and Microsoft Office. They carry a lot of mission critical systems of Microsoft customers on these clouds, right? And it is important for Microsoft that for this mission critical applications that there is no downtime and then it is actually trustworthy. Right? So Infosys is again providing support for Microsoft on this and the way that we have used AI is that AI agents today monitor the logs and predict issues before they actually happen and they give all of this intelligence and which we call triaging and then we route it to a specific support engineer with the information so that they can actually work on it before the issue actually happens. So this is for both reactive and proactive issues and so you can see that we get 40% improvement in the incident response and 10x improvement in the RCA turnaround.

So there are several such examples. Right? The other example that I have here is Danske. Danske has a Forward 28 strategy where they're looking at modernizing their entire landscape. They want to bring process efficiency and they want to completely create it as a digital bank. So you must have seen the press release where they've chosen Infosys for this transformation and we are helping them on the AI strategy and transformation as well and we are doing everything from AI strategy to implementation. We have set up innovation lab for them and we are creating multiple multiple AI solution work streams.

So we are using agentic AI in the code development, more than 2 million lines of code that we have generated. But then again these lines of code that we generate has to be validated and that's what our engineers have been trained to actually do and 97% of the engineers are using that. In addition we have created multiple AI solutions. They wanted to use ChatGPT but they wanted the guardrails so we created enterprise ChatGPT which has over 16,000 users. They created other solutions on risk and HR which have been quite successful. We'll hear from the customer itself. If you can play the video.

GenAI is already live today in Danske. We have 20 use cases from the most basic assistance to the more advanced agentic solutions. We've invested a lot in adoption by cultivating the right behaviors to deliver new ways of working so that people use GenAI both effectively and responsibly. At an enterprise level, tools such as our internal chatbot, Danske GPT, is used widely across the bank. Working with partners like Infosys, we're not just adopting this technology. We're actually helping to define what good responsible GenAI looks like at scale in banking.

With this I'll hand it back to Simran.

Thank you so much Bali. For our next session on Infosys Topaz Fabric and AI Next AI platform suite, please welcome Ravi Tarafdar, our chief technology officer.

Over the next 10 minutes, I'll cover how we are going to power AI first services and AI augmented services using our IP and platforms. Now earlier during the day we heard about the complexity with enterprise landscapes and when I think about enterprise landscapes I think about city maps. You know every city is different. The map for each of these cities are very different. Now if I bring any AI model, any tool, any platform, the only way to accelerate is by creating runways within an enterprise that can help us accelerate AI adoption from pilots to hundreds of projects and that's where Infosys Topaz comes in and we do it in five different ways.

First, we have created a rapid experimentation and innovation infrastructure where our teams working with our clients evaluate the latest developments that are happening in the AI space. They look at all the noise that is happening and identify tech that is relevant. They then build proof of values that are very relevant for their business and then they demonstrate the art of possible in very very rapid manner and today there are about 39 such innovation labs that we are running with our clients across the globe.

Second, we take a very value-centric approach in how we look at the end-to-end process because over the last few years we have realized that use cases cannot deliver significant value and this is where we are bringing the 25 plus industry blueprints we have in order to come with already re-imagined business and IT workflows that we can use with our clients to accelerate. So today when our consulting teams what they do is they sit with our clients, they understand the problems, then they use the product discovery and vibing tool from our Topaz fabric to very quickly identify good solutions, build a prototype and then using exponential engineering they actually create a production scale application by end of the day and then using this they're able to demonstrate how we can reimagine the complete workflow.

The third, Nandan talked about creating an architecture that is very evolvable. Now what we have done is the way we have designed our IP platforms is to make sure that we give optionality to our clients. They can pick any model they want. They can pick any agent framework they want. They can run on any AI platform. They can run on any AI cloud. And we can integrate with any AI native tool that they have partnered with and today in most of our production deployments we have a number of varieties that today we are already supporting.

The fourth runway is to build that enterprise context. Now here we are doing two things. One based on the work that we are already doing with most of our clients we have built an enterprise context. Think of it like a map. Whenever I want to navigate in a city, you need a map which tells you how to go faster. So we have built this enterprise context or map that tells me how the systems work, what the infrastructure looks like, where the apps are running, where the data resides and how they all connect. And on top of it, we're building an industry context. The industry context tells us what happens within a retail, within a bank, within a CPG context. And these are the models that we are bringing out of the box through a graph technology and we are building these enterprise twins and this is what will enable us to accelerate eventually in most enterprises to drive projects at scale. They need multi-speed IT governance projects so that they can onboard these AI tools at speed. They need to put these guardrails and that's where we are building a lot of tooling that enables them to deliver these AI solutions in a very trusted manner and all of this comes through Infosys Topaz.

Now in the IP and platforms that we have been building at Infosys, we have always kept our customer needs in mind. So if a customer comes and says look I want an end-to-end vertically integrated AI and agentic platform then we use AI Next as a platform to accelerate value. Or if the customer says look I've already made some investments I want a composable modular agentic and AI platform which can help accelerate my own AI journey at speed then we essentially bring the Infosys Topaz fabric. So with a combination of these we are able to meet most of the demands of what our enterprise customers have.

Now let me talk about Topaz fabric itself because over the last few months we are starting to integrate all the different IP that we have at Infosys into one common way through which we can deliver our services. Now Topaz fabric enables five key capabilities. The first is this builds on our customers existing investments. So this is not about replacing what they have. So this works above their model layer, above their platforms and above their enterprise systems and fabric can integrate with any model, any framework, anything that they have. So that's the abstraction that we have already built within this.

Second, it provides close to 600 agents which have been purpose-built for different AI first services, AI augmented services and also for industry specific flows. So this is something that we bring out of the box to accelerate the journey for our customers.

Third, what we have also done is we said we'll create out-of-box integration. So we have out-of-box integration with most of the coding tools. We have out-of-box integration with different models. We have out-of-box integration with business platforms like SAP, Oracle, Salesforce. We have out-of-box integration with data platforms like Snowflake and Databricks and with enterprise platforms with ServiceNow. Now what this enables is it enables us to deliver value to our customers quickly.

Now in all of these it is also about bringing the enterprise context and hybrid intelligence. So the way we are doing is we are starting to build a number of different ontologies and models that comes prepackaged as part of our Topaz fabric and that's something that we bring out of the box and as we deploy it we learn from the data, we learn from the processes and this is how we create a closed feedback loop where the context keeps improving as it gets used over a period of time. Now a lot of our clients also want to use a lot more predictability in the way AI is deployed. This is where we bring a lot of deterministic rules, couple it with AI models and we also bring our own small language models in order to create a right value proposition from cost as well as from a time-to-market standpoint.

Now all of these is backed by a lot of deep research and patents that we have filed over the years. Now while we are doing a lot of innovation internally within Infosys we also acknowledge that there is a lot of innovation happening outside. So we are working with AI native partners in three different ways. One, if they have a platform that is really good at doing something, then we are leveraging it for the tasks that are relevant to enterprise. For example, Nandan talked about brownfield. So, our Cognition partnership is largely to use Devon for a lot of brownfield engineering because we find that that is really good at it. Second, we are building embedded agents that can work within our partner tooling. So today we have built agents in fabric that can run within Claude Code, that can run within GitHub Copilot, that can run within ServiceNow. So whichever platform the client has these agents work within that environment. And third we have integrated with their tooling so that we can cover the end-to-end value chain that is required to accelerate the journey.

The next is also we are focused on where the industry is heading on AI and this is where we are working with universities to do joint research. We do today on agentic technology, on scaling and on trust with Cambridge, with Columbia and Cornell and we are also doing this a lot more with the research centers that we have set within Infosys. This is to make sure that we continue to build on what will come next.

Now let me bring all of this to life with an example where for one of our logistics clients they were finding that the customers were able to process the orders and bookings in a lot more accelerated manner and this was creating an issue for them and they said we want to be able to process these customer care services in a much more accelerated manner. This is where we took AI Next as a platform because they had already tried with multiple platforms and it didn't work and we said we'll use AI Next to uncover the existing knowledge because they had rules that are very specific to each customer. So we pulled out close to about 8,000 different rules that exist there. The second innovation that we brought here is that we automated the entire workflow. So when we started the extent of automation was about 0 to 10%, we took them to about 70% automation in their entire workflow. What this meant is the turnaround time reduced from 24 hours to about 30 minutes. The next is they also were very concerned with the sovereignty of the stack. So what we did in this case is we used the Mistral and Pixtral models to make sure that it addresses the sovereignty needs that they had and this today supports orders of bookings from 116 different countries and it supports 15 different languages. That's the power of what this could do. And eventually as we started scaling cost became an important driver. So we had to bring in a lot of optimizations to reduce the cost significantly for that customer. And this is something that we did over the last one year and today this is live. Now you can see a lot of these IP platforms as you come to our living labs and I encourage all of you to please spend some time and experience these technologies that we are talking about. Thank you very much.

Thank you Ravi. Now for our next session on unlocking AI value in communication, media and technology. Please welcome Anand Swaminathan, segment head, communication, media and technology.

So let me share a few things in the next 10 minutes particularly around what is driving AI demand in the communications, media and the tech business and what is the value that we

At Liberty Global, AI is no longer an experiment. It is becoming foundational to how we run our business and how we serve our customers across Europe. And Infosys has been a critical partner in helping us turn that ambition into reality. Together, we operate and continuously evolve the connectivity and entertainment platforms that support tens of millions of customers across our footprint. We innovate faster and we now have the engineering capacity to move ideas into production quickly and reliably. Over the past year alone, Infosys supported more than a thousand platform deliveries. These included major launches like Super Search, which serves around 8 million TV customers using advanced large language models to make content discovery conversational and intuitive across linear on demand and streaming. Now, we're also using AI to fix and transform how we operate through AI programs like agents assist and customer assist, which are now deployed in multiple markets. We're enabling more self-service journeys, improving satisfaction, and reducing pressure on our care teams. As a result, we've seen things like outages reduced by 50% year-on-year, and 60% fewer customers impacted. Our relationship spans more than two decades, including over a decade as a formal strategic partner, and it underpins major programs across our company. If I had to narrow it down to one thing that sets Infosys apart, it's trust. We have great mutual respect for one another and I always know I can call them on any issue and get a straight and honest answer and that matters to me more than anything.