Hello, everyone. This is Shripia. It's great to be here for this topic on generative AI for enterprises adoption and scale. Let me just check the chat.There are a few people who've joined. Thank you for joining. In the interest of time, since the session was scheduled to start, at this, slot, we'll get started. A little bit about me, I'm Shripriya, a director and chief architect. I work for Capgemini. I'm doing a lot of, work around for customers at scale. And it's great to be here for, the women tech, conference. A lot of interesting sessions, a lot of interesting lineup of speakers. It's an absolute honor. So what is Gen AI? All of us know what Gen AI is. There is a lot of hype around it. In fact, we have this running joke that you cannot have a meeting these days without talking about Gen AI. 

Every meeting, every discussion that we do with customers, that we do with our partners, invariably, we end up talking about generative AI. So, essentially, to keep it very simple, Gen AI is artificial intelligence that's capable of generating text, images, or creating other media and content. And before I get into the slides, just, a word of advice for those who've joined in. This is meant as an introductory session on how organizations and enterprises can scale on generative AI. For some of you, this may seem, introductory because a lot of us today have moved to doing a lot of projects at scale. However, I also wanted to cover some introductory concepts here because that can give, an encompassing view for all the folks who are joining in on how Gen AI can help organizations and how organizations can scale on Gen AI. 

Gen AI could be the key to solving a lot of world's biggest problems, and these are quotes that I present to you by, some of the famous folks who are in this space. And Gen AI models, they surprise and also scare us. And I talk a little bit about the, pitfalls of Gen AI in the subsequent slides. Moving on, AI is not new. It has been around since the nineteen fifties. And today, we see a lot of systems, a lot of, banks, a lot of retail outlets, a lot of, media outlets, pretty much anywhere and everywhere, including medical field, education that uses AI in and out. And we currently see a huge demand. So if you look at how Google searches are happening around AI and chat GPT and some of the newer tools that have made, Jenny I so popular. 

We observe that there is a lot of interest, a lot of searches that are happening in Google Trends in the past few years. There is a lot of demand for Jenny I lately. There's a huge amount of data that's created by IoT, medical devices, by social media, a large number of algorithms that are available today. And also with chat GPD getting, released a couple of years ago, there is a huge race for journey and chat GPD then, you know, deep seek. There are regional variance. There are some players that are very big in US, some players that are very big in Asia, in countries like China and so on, some players that are very big in Europe. 

So there is a lot that's happening in the space, lot of new tools coming up, lot of new models coming up, newer versions, data being available, and that has led to an unprecedented growth of AI and Gen AI in particular. Unprecedented growth of AI and GenAI in particular. And if we look at enablers of AI, just to give you a sense of the amount of data that is generated these days, 500,000,000 tweets sent per day on Twitter, billions of searches happening every day, millions of, Instagram posts, millions of TikToks, millions of videos, Snapchats, you name it. 

That's the amount of content, photos, images, audio, videos that are created every single day, every second, every minute, every day, and it adds up. So that's the data part. And then we have cloud, which has made it very easy for us to fine tune these models, for us to, create models that can have an output that can be fine tuned, that can be trained, that can be pretrained. So there is a lot of computing power that is available in terms of GPUs. You're not just restricted by the sizes that were available as well. And then there are plethora of algorithms. So we've also as the community advances over a period of decades, there are multiple and you go to the that are available that have been, tried, tested, proven that can enable use quickly and, you know, give output as expected. And then, of course, this is widely supported by talent. There are people who are constantly learning on AI. They're they're getting themselves trained. 

There are a lot of public, tutorials available, lot of public repositories on AI publications, and industry support that's available. And all of this, in my view, comes together in how AI has become so popular. Jenny, I has become so popular. So a little bit to go back. I don't know how many of you know about Eliza. Many, many years ago, there was this first chatbot that was created called Eliza. And by the way, it's still live. So those of you who are interested, you can actually go to Google, look up Eliza, and try to chat with it. And what you see in the screenshot here is where I was trying to talk to Eliza, and I was easily able to confuse the chatbot. And it it just shows how far we have come from q and a kind of bots many, many years ago to very, very smart bots, very contextual bots today that can help you with very, very deep, meaningful conversations that can solve a lot of, repetitive, questions, repetitive tasks that organizations today face. 

And today, you have chatbots that can learn every day, that can, easily be trained, that are good at task. And, that's just to give you a sense of how far we've come from the olden days. And those of you who are Harry Potter fans like me, you know, if you remember there was the part where Harry Potter is with Tom Riddle's diary, and he sends my name as Harry Potter, and then the diary responds back. So, you know, very similar to that. Today, we have the chart GPTs and the Geminis, and the beings of the world that are able to do that, but much, much more intelligent that can pretty much answer anything and everything that, you would like to ask. But I also want to show you one aspect. 

So in around 2023, which is the screenshot that you see on the screen, when I was interacting with chatbot, I was able to confuse it a little bit. And this was early on. And then what I observed is over a period of time, a lot of these bots also have learned based on what worked well, what didn't work well, the feedbacks that came in. And today, a lot of these Gen AI tools also have become really, really mature. And when organizations are trying to use these tools, it's also important that they put in all the guardrails and the checks in place to ensure that the bots and the Gen AI tools only answer in the context of what the organization would like them to, do. So those guardrails and checks and balances, should be in place. But this is just to give you a sense of, again, how far these tools have come. 

And a lot of the content that you see in my slides today have been created by Gen AI. So for example, you know, I I early on, I was trying DALL E on how, you could have, for example, these funky images of bear on Times Square or, a panda that's doing a scientific experiment. Now beyond, you know, the quirky stuff that you see on the slide here, there's also a lot of practical implications of using a lot of these Gen AI tools. It can create logos. It can create high end images. It can create videos. It can possibly even write, a new best seller. It can create a movies, short movies, advertisements. So there's a lot of creative side, which, a lot of these Gen AI tools are able to do, and these are really, really useful for our organizations today. 

Now switching gears a little bit on how organizations can scale on Gen AI usage. I think a lot of us as technologists today are at a point where we've been talking to a lot of our customers, partners, or we're working in product organizations that are creating Gen AI products for, very, very large scale implementations. So we are no longer talking to customers and accounts about, you know, creating a small POC or creating a point of value. It's also about how organization can scale and benefit and have business outcomes that result from the usage of AI engineering. And I will what I'm gonna do in the subsequent slides is I'm going to break down this slide that I have here into smaller chunks so that it becomes a little easier for us to understand. And and like I said, this is an introductory point of view. 

Each of these elements has a lot of sub, steps and blocks, but just to give a sense of how organizations can get started. So in our my experience, personal experience in the organization that I work for, the first step would be a strategic alignment and compliance. So you want to start any large NEI implementation. The first thing you need to do is understand the legal implications. What are the ethics? How do you want to use NII? How do you want to put the guardrails? What are the controls, the compliance? That's really, really important because you don't want 10,000 people using NII in a way that's not appropriate for the organization. That doesn't bode well. So that's gonna be the first, you know, step in terms of governance that you would like to put in. And the second thing is to do an assessment. 

You have large landscapes, today, hundreds of applications, lot of redundancies, lot of different tools, developers having different, IDEs, different ways of working, different technologies. So understand what the current landscape is and identify some tools that you would like to focus on. It could be specific hyperscalers, some open source tool tools, some paid tools, depending on the context. Identify a few tools and then run a few assessments around it. What is it that could benefit the organization at scale? The third thing and very, very important is also to take your teams along. So if you need to have hundred, two hundred, thousand people in your teams to start using Gen AI, there is that change management that needs to come in where they need to now start prompting before they code. They need to start prompting before they create a design. They need to prompt before they create test cases or test data. 

Or there are users who will need to prompt effectively to create some requirement documents or to create design specs, whatever the case may be. And it's important because that's a change in the way they're going to do their day to day work. So the upscaling and the training, the ways of prompting, understanding journey and hands on. And this is not very difficult even for non techies. Techies, of course, they get really excited with any AI. They want to try out these tools. But if you look at even non techies today, they are really, really interested in seeing what, Jenny I can do. And English is the hottest programming language now with Jenny I coming into play. 

So they are able to pick up prompting, new ways to prompt, and they are able to, make use of these Gen AI tools just with plain English. With the right prompting, they're able to figure out how to, get an output that meets their requirements. Number four. Now you did an assessment. You figured out a set of tools. You've trained your folks. The fourth thing that you need to do is to do a tool evaluation. I have four tools. Let's say you want to go with Copilot. You want to go with something that's paid or licensed. You want to go with something open source. You want to go something, with something that's, you know, more privately hosted versus something that's not privately hosted depending on the needs, depending on your specific task, you do a tool evaluation. And when you do tool evaluation, it goes hand in hand with the fifth aspect, which is running a few pilots and measurement. 

And this is really important because a lot of these tools, they look good. All tools are good. A lot of hyperscalers today have competing tools. But sometimes what happens is there may be specific tools that would be a little more apt for your scenario because every organization has preferences in terms of IDs, in terms of technologies, tech stack, ways of working. So there could be some tools that, you know, have an edge that may be a little more better suited, and that will come out when you do these pilots and measurement. What is measurement? So measurement is typically when you, for instance, you run, a tool and you run a pilot and you want to see how efficient it it was. So that's where you can scientifically measure measure how good or how productive the tool was, the generic tool was, in helping you do what you would have otherwise done manually. Now once that is done, go back and, you know, sit with the teams and do a strategic road map. 

So do a road map planning, understand which tools to use, what are the first movers, which application will move first, how to move, when to move, what are the caveats, what do you need to watch for, what is it that you need to, do differently. And sorry. I was just checking the chat if there were any questions. What is it that you need to do differently for your strategic roadmap planning? And this is again, you know, it it's going back to your board seeing all your applications, how they stack up, and then, you know, using the findings from the pilots to have an assessment and a tool evaluation. And then you start scaling. So, for example, you start scaling from zero to 30, people using zero to 10 applications using. Use for all the Java projects first, then go for, you know, your mainstream modernization, whatever. So, you know, you scale in an orderly fashion and then slowly expand, build upon what you did, what is it that you found, you have a feedback loop, what worked well, what didn't work well, come back and see how to improve, iterate, have developer feedback, what worked well for the developers, what worked well for the BAs, what worked well for the, engagement managers, what worked well for the CXOs. 

So every layer, Jenny, I will have different feedback, different outputs, different usage, and then slowly scale there on. So we started with an assessment, upscale. You do a pilot. You measure. You evaluate, and then you scale, and that'll help you, expand and then iteratively repeat the whole process. That's really important to do that in a structured scientific fashion. Now is Gen AI a one, solution for all problems? Is it a magic wand that can solve all your problems? Probably not. We have to be conscious as technologists, as architects, as a sports, as leaders on what AI can and cannot do. There are some aspects around sustainability. A lot of these AI models have a lot of carbon emission because, you know, they take a lot of processing power to be tuned. 

A lot of these LLMs also, we have to keep that in mind. In some cases, a simple automation may be the solution. In a lot of cases, Jenny, I may be the solution. There are concerns with privacy, bias, trust, ethics, data leaks. There's also the AI ray race. I spoke a little bit about at the start, you know, in terms of how the tools have been evolving. So when you use any tools, keep these in mind. We we don't, have to go into specifics for each of the tools, but these are concerns. These are mentioned for each of the tools and ensure that you look into descriptions of how privacy is handled. Is the data moving outside of your environment? Is the tool using the data to train its own core LLM. How are the ethics handled? Is there a data leak? So all of those have to be addressed. 

And as leaders, please ensure you keep these aspects in mind when you're choosing, with the Gen AI strategy and your tools. Now future of Gen AI could mean, like I said, Gen AI could be writing code. It could be writing writing movies. We could have these presentations. A lot of the, images in my slides have been created by Jenny and I. You could have coding done by AI, movies written, maybe flying cars, pop songs, presentations, you name it. Even in the medical industry, there is, you know, education. There is a lot that, Jenny, I can do. But, again, going back to what I spoke in my previous slides, it's important to have the guardrails in place to know how you're going to use those tools and, you know, how the tools use your data. 

It's important to keep those aspects in mind. Gen AI and AI is a very, very exciting journey. We need to tread carefully. And I leave you with, I don't know how many of you, know Peter Parker. So, I mean, Spider Man with great power comes great responsibility. For those of us who are leaders who are using Genya, it's important to ensure we use it responsibly and carefully.