Leading Through AI Disruption

Ashiss Kumar Dash:

Good morning. Hope we keep you awake with the coffee and tea in the back.

So this reminds me of an Indian wedding. We have employees, we have our partners, we have clients, and we have would-be clients. My wedding had only a small group of 1,500 people. So by the end of the reception, my cheeks were in deep pain. So we'll try to keep smiling, and there are people standing in the back, as well, that remains of how well received this topic is, and we are very grateful to the sponsors for really sponsoring this event.

The topic today that I wanted to touch upon is leading through AI disruption. People standing in the back. This is not lending through AI disruption because there's a lot of lending going on in the AI business. So this is leading through AI disruption. I wanted to share our perspective, our point of view, and what we are observing with our work with clients.

So there is a threat framing in AI, obviously. People are saying that there's going to be a lot of loss of jobs, there's going to be a lot of disruption in the way we work. And there is the opportunity framing, which is around what do we do to scale up AI at the enterprise level? How do we unlock and do things that we are not thought of doing in the past? I will focus more on the opportunity framing today.

We've all seen this. How many mainframe programmers do we have in the room? Okay, quite a few. So we moved from mainframes to PCs, to the whole digital revolution of cloud, then smartphones, and now AI. This chart shows you how fast the adoption of AI has been to reach a billion users. Now, the reason it is happening is twofold, right? One, obviously, AI has the advantage of everything we got through this multiple transitions. We got the data, we got the network, we got the cloud, and therefore, the speed of change is rapid, and it's going to be even faster in the future.

And that really gives us massive opportunity to rethink our IT estate, our technology estate, right? So the legacy modernization of looking at how do we become more agile? How do we become more resilient? How do we reimagine our business processes and deliver value that we had not thought of in the past has become more important now than it has ever been in the past.

Having done work in mainframes, I know so many clients today who run on TPF and mainframes and tandem systems. This opportunity was not there five, ten years back. Today, if you look at it, there are no reasons to delay modernization. In fact, the risks of delaying modernization - number one is obviously financial drain. On an average, when I talk to CEOs and CFOs, they say that about 60 to 70% of their tech budget goes into just keeping the lights on, right? So that means another 30 to 40% is available only for transformation work. But when you are able to modernize the systems with AI - and some of our partners are here, we had very good discussions yesterday on how to do that - you actually reduce that financial drain.

The second is security vulnerabilities. I think SURE as an industry, whether it is utilities, oil, and gas, mining, all of us are always very, very finicky about security, right? Threats to our assets, because we are an asset heavy industry. And therefore, I think making it secure, there has never been a better way to do that than modernizing the legacy. And the third is, this is a time to innovate, right? These are times - I'll give you some examples. These are the times to innovate. So innovation cycles are becoming faster and faster. But if you stay in the legacy, then there is this thing called innovation paralysis, which is you're not able to do things because the systems don't allow you to do things.

Obviously, there is demand, tremendous demand for modernization, which is through the low agility tech debt, like I talked about, this slow rate of change, cost of security, all of that. But good news is, the supply side is here. So we have high rate of change. Every model is changing. Every tool set is changing. Now we walked into the agentic AI framework, right? So speed of change is very, very rapid now. The second is, we are able to get very secure compliance, sustainable, and responsible AI deployment at scale. That's possible today. Third, obviously, the efficiency in productivity and then fourth is it can easily scale at the enterprise level. So, net net, I think this is a great time to rethink how the legacy estate can be modernized, how can it be done responsibly, and how can it be done at a scale that we had not imagined in the past?

This is an example of Infosys. So we launched Topaz Fabric in November last year. Just see the level of adoption that we have seen for this, right? This is our employees doing work for our clients. 3,941 applications using vibing. People are doing vibe coding. I think all of my colleagues here are doing vibe coding, including me. We have deployed about 600 plus agents using the agentic control framework of Topaz Fabric. We have built five small language models. These are purpose driven models that we built for our clients, for certain industries, like, for example, IT operations, we built a model. We built a model for the banking operations, and so on. For the energy industry as well.

We also got 90% of our talent AI aware, which means people are getting into Topaz Fabric, learning things, going and taking classes, coming back, trying out new ideas. And then we have got about 44,000 AI builders. These are people who are not really proficient in doing AI - to 44,000 now. Last year, this time, we were somewhere around 6,000 or 5,000. That's the speed at which the adoption is happening is phenomenal. We have already done 25 plus implementations of Topaz Fabric. It will bring in work with our partners to extend and create the ecosystem of AI deployment. And then we are working on about 80 plus implementations of Topaz Fabric. So this is the speed at which things are moving at us.

However, here is a bigger opportunity, and the opportunity is, there is an execution gap, right? The models, the foundation models are moving rapidly up, every week or every two weeks, you hear about the new release, which is far superior than the earlier one. There's a massive competition going on. But the enterprise realization of value is yet to catch up. There's multiple reasons for it. One is obviously clients want to make it safe, secure, reliable. They want to make sure that they are not the first ones trying it out. Second is, there is a productivity gap, right? There is a developer productivity issue. And therefore, I think the opportunity to do a lot more is here. This execution gap throws up so many things to do in different industry sectors.

But how are we approaching this? We have a three-pronged approach here. We want to be the AI brain, the strategic brain for AI, so we are enhancing our consulting practice. The Infosys Consulting, as well as the AI Consulting teams. We are ramping them up. We are focusing a lot on operations - like I said, if keeping the lights on is causing a financial drain, that's not the right thing. I think we need to move every dollar that we can towards transformation. So we're focusing a lot on making sure there is ROI-led, AI-driven, agentic framework driven operations that our clients have, and we've seen tremendous success in this today - during the panels people will share some of the data.

And third, which is very important, is how do we make sure that we design cloud native AI-first modern architecture, that is scalable, because you don't want to design it for today. You don't want to design it the way things have been here. You want to design it for future proofing your systems, future proofing your data, so that when things change, your orchestration layer stays the same. You can plug and play multiple models, you can build things for yourself, you can take advantage of the changes that are happening in the system.

And the most important part is talent. There is a gap in finding the right talent. So we have invested in building platforms that can train thousands of people. The example I gave of Infosys, but there are many clients who are taking advantage of our talent and planning, and the last thing is, how do you do it in a responsible way? So we've launched a responsible AI playbook available on the GitHub, and it is available on our website as well. So that adoption of responsible AI, having a clear site of value, and doing it at an enterprise level scale, is what's going to take winners from where we are today. Thank you very much.