As a way of introduction, my name is Vikash Rungta. I'm based out here in the Silicon Valley.And, I have, I've built multiple generative AI products in the last couple of years. And before that, I've built, some of the largest AI products at Meta, and my own startups before. Recently, I have worked on Llama, where I am responsible for building out a lot of the LaMaa products out there, as well as I teach generative AI at Stanford. So when you think about generative AI in the c suite, it it's really changing the way people have to kinda think about what your role is in the c suite has to be. Before, technology would be more of an enabler that leaders would use.

Broadly in the where they will rely on their product teams and others to kinda, look at it, but now things have changed. First of all, if you wanna get access to the slides, you can take a picture of this and kinda go to the QR code and just request access. After the call, I will provide access to that. Also, let's let's define what generative AI is. So generative AI really is it refers to the intelligence that can create new content. This is different, than, analyzing or sorting information which a general AI would do. So in a typical AI, which we call is, like, a traditional AI, you would be actually classifying things or you are predicting a particular bit about the characteristics of the data, with generative AI, you are able to generate the data. Right? It can generate things like text, images, music, code, and, which can then leverage it to even those it seems like you're just generating text.

What you really are generating is intelligence as a text. So what it means is that not only can you generate text, it can bring a lot of the intelligence around strategy, the things that you're doing, and bring it into a way which is consumable to you right away. It can generate images, but not only can it generate images, it can generate charts for you. It can generate emotions for you. A lot of this content that I'm presenting today, you can see is also generated by generative AI. For example, all the images are generated here, so the formatting has been done by that. Right? And, these are generally powered by advanced models, versus, like, the typical AI model that was used for, you know, a couple of decades before.

Things like GPD, ChargeGPD, LLMs, Lima, Gemini. And Gen AI learns from massive datasets. So there is this theory about, like, attention is all you need. And give you a story about, like, where did we actually start preparing or, like, how how do we generate the AI models even get started? Right? So Google was working on this problem of translating languages, one language to another, and they needed something that can do the job. And they wanted to invent a new technology and use that as a use case because it's not impacting their, it it does not impact their revenue. What they realize is that if they converted one word at a time, AI could do a pretty good job of translating from one, one language to the another. But what they also realize is that what if they actually, looked at the word before, then the translation becomes so not only the word that you're translating, if you look at the word before or the sentence before, like the entire sentence, the translator started to become better.

Then, one of the tasks was, like, what if you paid attention to not only the sentence before or or the entire para? Would it be better? The para was better than the sentence. They started paying attention to, everything that was written before in in the entire document. So what if you looked at the entire document before they translate that one word? And it turns out that it matters, and the quality was much better. And this is where the fact, the paper called attention is all you need got, got ridden up. And GPD model started knowing that by paying attention to everything else that's going on in the world, it is possible to kinda generate so much better text, so much smarter, better images. Right? So this is where generative AI is. However, as a c suite for you, the strategic imperative of generative AI is huge. Right?

It's, by 2025, it is expected that 25% of all enterprises will enterprise generative AI in one way or or the other. And if that is true, what we are seeing is that a RI adoption, cycle, which used to take ten years to do before, is getting compressed into two years. So what you would have done, like, in historically the the amount of deployment that you would do in ten years would has to be done in two years. And there are clear strategic winners emerging from this. You know, Lagasse will pay later, but, you know, Lagasse will have to will really suffer. So if you are a seashore executive here, you really have to kinda think, June do you wanna be in the strategic winner? And the answer should always be a yes. And how should we kinda, like, go ahead and adopt it?

So what we are seeing is AI is reshaping what it means to lead. Right? Executives are no longer just a decision maker. They have become the orchestrator of technology. Right? What it means is that, you know, as a leader, you're evolving beyond the traditional roles. Right? You're not just focused on the typical traditional roles of decision maker. You're, you know, you're also responsible for capital allocation and the culture of the company. You are expecting or you would expect Gen AI to handle a lot of the routine operations for you. So what it means is that you really have to kind of set what does it mean as a value system? What is the culture of the company? And where should you spend money? Because generative AI is not cheap. Right?

Like, you have to spend a lot of money to the compute, and where should your capital allocation go? Right? It is also, so that leaders can focus on high impact decisions, shaping the organization's future, which is, through cross functional automation. So you have to kinda think about, like, what are the cross functional things that can impact the way you do the business, where you're not deploying just humans to do the job. You're deploying a mix of humans and agents to streamline a lot of the operations. And to do this, if humans and, agents are gonna be peers in many ways, And in in in many, many cases, you will see that every human will be paired with, like, multiple agents. You can't just lead humans. You have to lead with AI and, like, you have to lead this AI, set of, agents that you we're gonna be building.

This can change the way how you kinda power your boardroom. Right? So imagine a board packet that's auto generated overnight. You wake up to the risk heat map, right, cash flow, anomaly detection, and strategy options ranked by ROI. Today, there is a large team that is kind of presenting and preparing this for you. What if you can wake up in the morning and you have all of this packet ready for you? Not only that, you you can have you can start making data driven, decisions. So when you get, like, you get the premarket analysis scenario forecast and rapid data driven decision making before the day even starts. Right?

So not not you, not any of your employees are working. The executive's job shifts from information gathering to strategic interpretation and high impact, forward looking planning. Right? So you really have to kinda think about what do you how do you interpret with so much data and so much insight? How do you kinda make those strategic shift and, plans? So couple of things here. Make sure that you are the leader who has this board packet ready for you, who has the data driven like speed so that ultimately you you change your focus to what it is. So I wanted to give, like, some sample, things that you can do. Like, these are some of the areas that AI can actually accelerate, and I've given some metric lift.

Of course, these are, you know, sample or, like, what I what I expect would happen. So what could happen is Gen AI isn't just hype. It delivers real business value. For example, in the cost leadership, it can slash the cycle time by 40% by using things like RPA, processing efficiency. You can bring differentiation boosting the NPS because you have a differentiated offering or solution, and you build hyper personalized interfaces or hyper personalized products. Your NPS Net Promoter Score would increase. You can create new market by building AI native product lines, which can definitely increase your revenue because now you have new product lines. On the contrary, if you don't do it, startups and other companies are gonna build that and take away this business from you.

You, you build the risk arbitrage, right, by providing real time compliance. So you can actually monitor any violation or fraud that's happening real time, and you are able to kinda go ahead and kinda mitigate that. And then, of course, you are leveraging the talent. So every, there is a say currently the values, like, everything is 10 x, and it may be 100 x in a few in a couple of years. What it means is that how can you take every function and give them 10 x tools that they are able to do so much more but with very little effort? So that your employees are not just working at one x or one and a half or two x.

They have these tools, which you can, for example, a Copilot from Microsoft, or you are able to kinda use tools like, ChargeGPD and others. And, there are products like Glean, which can actually go after your corporate, information and provide people a way to write better emails, better documents, process better things. And this productivity boost has happens across the board, across all functions. There are, of course, products like Cursor, which a lot of the engineers are using to write 10 x code. And if you are not actually doing this, you are leaving money on the table. You can see an immediate lift off like, double digit growth, in in terms of the productivity.

But in the next couple of years, it can be multitude of folds of that, as well. I wanna take, like, a couple of, enterprise use cases. And what it means is, if you look at Morgan Stanley, they really cut down their research time by using the AskResearch GPD, which generally would take a task that would generally take the thirty minutes was cut down by less than two minutes. Right? This is something that, they were able to kinda do. And through that, you can imagine the of course, this is one task, so 93% productivity gain. But overall, if a person has to do this multiple times a day, they would get a significant productivity uplift. Similarly, if you are in a company like Coca Cola where the user generated ads were generated for fourteen days, you know, or in fourteen days, and they had this campaign called create real magic. Google it up or look on YouTube. And it really brought, like, AI native, brand uplift.

Another example that I have, is and, one of my friends told me that he had, he was at he was at, you know, he had some kind of a fine charge on their American Express card, $40. So when he started chatting with American Express, he's they said, why are you here? And he said, for a late fees. And instead of passing on the buck to, like, a customer support agent, American Express was able to quickly identify what the late fees was. So it was able to kinda go and relate that the person must be talking about the this late fees because that was one of the only late fees that that he had in the account. It was $40. They made the decision, and they were to waive it off immediately. And they came the chatbot came back and say your waive late fee have been waived. Thank you so much. It saved time for the customer support agent.

This was a this would have been the decision probably even though it went to the customer support agent anyway. So they didn't hopefully, they didn't lose money on that. But the customer delight was massive where they didn't have to provide all of this informations like which day did you use it, why was the reason to do it. It's just providing this kind of a delight, would be an expectation going forward. One of the, one of the things that you would have to, think about when you are here is how do you make the build versus buy, versus integrate choices? And depending upon the size of the company, you are, how should an enterprise approach that GenAI adoption? There are mainly three choices. Build is for high differential, but slow, and, resource incentives.

So if you are a larger organization where you have a lot of, where you have a lot of the content available to you, you have the resources and the compute power, the engineering power, it may make sense for you to kinda build. It is slower, but it provides a long term ROI. Right? But there's also the risk is, you can actually spend a lot of time trying to build and not be able to kinda ship anything, to your own customer to your to your user base. Buy on the other side is, is it it's not a strategic differentiator. Right? Because every customer of your sorry. Every competitor of you would also have that. But it's super fast. Like, some of these things, for example, if you, allow Gemini or Copilot on your organization, it can be something that allows you to kinda build, right away.

And, but there is a vendor lock in that is possible. So you may wanna look at more vendor agnostic, tools, and that allows you to kinda, like, over time, transition over. Integrate, it balances the differentiation and speed, but requires strong integration architecture and careful handling of data. This approach depends upon the goals and existing capabilities. Right? So if you have, you may wanna actually do a little bit of a a buy plus integrate or build plus integrate or buy build and integrate. So the reality is you will probably have all three of them, but you wanna define what are the aspects that you wanna buy, build, and integrate. Always build your core aspect of it. If your business really depends on this, you wanna build it.

If your, if you're these are functions that you wanna, like, accelerate performance, you wanna buy it. And if you are, like if you think that there is a lot of custom, data or the solutions that may work for that may help, you wanna integrate. So when prioritizing the JNEI initiatives, focus on quick wins, high impact, high feasibility projects, like, finance co pilots or customer service tools. Right? These are the things that has already been proven. Find out which tools have been proven versus not because there are a lot of products which are not ready for disruption with Gen AI. Customer service is a great example of where there's a lot of disruption has already happened. There's a lot of training data around that.

So your organization should be able to kinda scale on those things. Scaling plays, and long bets, like like autonomous supply chains, can follow once you have built the momentum. Right? So because if you have you start with the hard problems, you're not gonna be able to kinda deliver that properly. The key, always measures both impact and feasibility, and don't be afraid to kill the low value effort. Right? So, you know, if there is something that you think there's not gonna be a lot of value. So killing you should be killing, like, 60 to 70% of your projects in the bottom right corner over here. And don't forget the, governance and responsibility. Right? So, at Jenny, I becomes core to operations. Governance matters more than ever.

Adopt, recognize standards, map to regulatory requirements, maintain a robust inventory, and keep humans in the loop for high risk decisions. Budgets for regular AI audits because oversight and transparency will only become more critical as adoption scales. So it's really important that you kinda think about this. So I wanna end with two things. First, I wanna give you an executive playbook. Right? What is your roadmap? Define a clear number one is like define a clear AI north star tied to the business outcome. You always wanna look for business outcome, not vanity metrics. Number two, set up an AI center of excellence in partnership with the risk and compliance. You wanna form a small team that actually kinda is a tiger team that kinda goes out in this AI center of excellence.

You wanna put some folks who are just thinking about this. Run a ninety day pilot project and measure their real impact. And number four, unify your data and leverage RAG for deep context. And what does RAG mean? It's like it's using personalized context. For example, Google just today announced that when you do a search result or search, it can actually look through your Gmail and some of the other, Google assets to customize the results for you. So if you have always liked to kinda travel, in a nature friendly, resorts, versus going out and living in a buzzing city, by looking at historical hotel receipts, They will start recommending that when you search for vacations. Right? So that is retrieval augmented generation, but you wanna unify your data so that it is accessible. Make it accessible by, make make it accessible by AI. And finally, number five, instrument your business with ROI dashboards for productivity, revenue, and risk. This is how you personalize Gen AI for lasting value. Right?

If you don't know, if you are, you know, if you are having building value versus not, then you will end up spending time and energy on initiatives that is not returning value to you. So my question to you is, which decision this quarter can you augment with AI? You start somewhere. You don't have to be perfect, but you start somewhere. The core cost of winning is real. Right? The competitive gap widens every day. So start small, measure fast, and scale what works. Lead the change and risk having the change lead you. With that, I wanna thank you. Please, check out my, Substack where I write a lot about the various things in generative AI. And don't hesitate to connect with me or reach out to me if you have, any questions or thoughts around generative AI.

I would love to kinda hear, about the problems that you're facing and brainstorm around that. Thank you.