 First, the apologies for being virtual here, but a huge kudos to the UX India team for pivoting overnight and making this streaming solution possible. I'm Sundar and I'm from Intuit. Intuit is hopefully a household brand, but if not, it's getting there very soon. At Intuit, our mission is barring prosperity around the world and we serve 100 million customers worldwide with our products such as TurboTax, QuickBooks, Mint, Credit Karma and MailChimp. Hopefully these are brands that you've heard, if not worked with. And through these products, we serve three customer benefits. First is putting more money into the pockets of our consumers, of our small business customers. Secondly, eliminating work so that they can do what they do best. And third, enabling them to make financial decisions with a lot of confidence. So that's our mission, that's how we deliver prosperity for our customers. In my role, I serve small businesses with our product Intuit QuickBooks, which is an accounting and business management software. And I also play the role of a product management and design leader at Intuit India. We are a 100 plus strong passionate product force, some of whom you hopefully have met yesterday and probably tomorrow. In this talk, what I want to do is share some of my thoughts on how AI has evolved from being this occasional or inspiring tool to becoming a utility, right? And help us deliver the kind of benefits that we couldn't in the past for our customers. And I also want to spend some time talking about our role as designers in making this a reality. But let me start with a personal story, I'm married to an accountant. I just want to confirm or reassure you that that is not a prerequisite to get hired at Intuit, especially not necessary to work at Intuit QuickBooks. It is purely a coincidence. So I'm not trying to stereotype, but I get excited about technology. I get excited about track technology, all of that. But she is more interested in numbers, not that I've not tried to. But I have to say that to some extent, you could argue that it's helped my marriage. That I don't talk necessarily the same language. So we just finished 19 years of our marriage earlier this month. So one of the reasons is we have complimentary interests. Having said that, it's not that I've not tried to get her interested. But that's the new Apple Vision Pro or drones or EVs, all of that. Like I've tried my best, but usually she knows how to tune me out. If I'm trying to get her interested on AI, you can pretty much imagine her reaction is going to be. So having said that, it's not that she has not experienced AI, one way or the other. Whether it is watching Netflix or Amazon Prime and discovering what's next, what's best for us based on our viewing habits or using Amazon or a Mintra recommendation for picking our next outfit, all of that. She's consuming AI. In fact, recently she was talking about how she's really impressed with the likes of Instamart, Blanket, if any of you folks are in the audience, you have a huge fan in her. It's almost as if they know what we want to order even before we know what we want to order and because they are ready in less than 10 minutes. All of these, and she's consuming AI, but it's also the occasional awe. And in many cases, she really doesn't even know that she's consuming, right? So I just want to give you a bit of this context because we had a pivotal moment in our relationship a few days ago. It's vivid in my memory because this is over a coffee break. Few of the benefits of working from home is that you can enjoy coffee with your spouse in the evening. So one of these days, a few days before that, she'd asked me about chat GPT, or maybe I told her about it like I always do. And what is this? Why are so many people interested in it? How can I access it? I just briefly explained to her. I told her that I'll WhatsApp you the URL, you go figure it out. And that's it. It was out of memory. But on this conversation, she was gushing over chat GPT to me. She had tried it. She had put that to use. She actually used it for a real work situation and was so impressed with the answers. Chat GPT pretty much gave her a work plan for how to address the problem, which tax government agency to go to what requirements, et cetera, et cetera. The entire plan that she could lay out, delegate some of that to her team and off she went. Something like this would have apparently taken like two to three weeks with her external consulting firm. In fact, she went even further to think, how are they going to survive in this revolution, right? So that was like I said, it's a huge moment where both of us are talking technology in all earnest. So clearly we are in the hype cycle of AI, specifically generative AI. So that got me thinking, how did we get here? What's happening? And where are we headed? And I believe the answer is in two pieces. One is you have to look at the evolution of computing and AI on one hand, and then the evolution of interfaces or user interfaces or user interactions at the other end. While AI is on the rise, we have evolved through many, many iterations of human interfaces, including chatbots, which you can see is actually very past its peak of inflated expectations, right? So that's where we are. Having said that, AI is not new, right? Let me, I'm going to just go quickly into AI and then I'll go to the user interface side. But AI is not new. Elisa, that you see in front of you, was built in 1966, probably the first instance of a human-like interaction with the machine. On the right hand side, you see, if for folks who have watched the movie, Imitation Game, that's the code breaking machine that Alan Turing and team built. A lot of these were actually built during the World War. So it's not new, right? And it's only that it's gotten better over time, so much so that I believe it's actually better than human in many use cases. In fact, the good part is, I know I saw the topic about how do we survive this revolution. The good part is humans are still better in many use cases. So we hold the trump card in many such instances, but AI is actually already better than humans in many of these areas, such as prediction fraud. In fact, as an example, a few years ago, we had set up our AI investments a few more than five years ago. One of the earliest ones were in fraud and risk management, and the team had set up a game for a few of us to look at data and artifacts. For context, when we bring in a merchant into a payment skateway, we put them through validation just to make sure that they're all valid merchants, not rogue merchants, so we don't harm any of our other customers. And say they submit the driving license, bank details, yeah, they are. And usually agents look at all of them, and then they make a call on whether this is valid or not. That used to be the past, and we now then put AI to do that. When we were doing it, of course, we were not trained agents, we absolutely could not. We could have just thrown a dart on the board, and we would have probably been equally right. But AI was cracking it. In fact, even better than agents. In fact, agents get involved only if it's on the border, sort of corner cases, when they feel even the answer may not be accurate. So AI has come a long way in getting better. Through the years, we actually made a lot of investment. In fact, as of today, we have more than, add into it, we have more than 810 million customer interactions every year. We have 2 million AI models in production, refreshing daily, 65 billion predictions every day. So it's an AI at scale. But the point I wanted to make is that it's not new. AI is not new for us, add into it. It's not new for the industry. So now if I look at sort of change, focus, and look at the user interface side, Jacob Nielsen, for those of you who follow him, he had this article recently, which is really beautifully articulated. I love the frames, I'm going to use the same frame to think about how user interface and interactions have evolved. We started with, like the code-breaking machine, we started with issuing instructions to a machine. Probably one instruction at a time, the evolution was, or revolution was batch instruction. You could give instructions in a batch. And over a period of time, you of course slapped a UI in front of it so you can visualize it, et cetera, et cetera. But all of this was centered around this basic premise of providing information and providing instructions to a machine. And in that process, we of course made it easier to do all of that with whether it's a pointing device like a mouse or a touch or a gesture, more recently voice or conversational interfaces. We have managed to reduce the values of communication. But at the heart of it, it's still you telling the machine what to do. And we are not able to bring the user into the creative process. The problem with this is that the user needs to know what to do, right? You're telling the machine what to do, you're issuing instructions. We know from our small business customers, for instance, we have dashboards, we have reports. But many of the times, our customers don't necessarily know what question to even ask, let alone figure out which report and what will give you the answer and how to piece all this together. And people in the audience who have built dashboards or insights pages, if you honestly rate yourself, you probably thought that this is going to solve all the problems for your customers, deliver all the answers. But we know, for instance, our small business customers have a job to do which is run the business, not come and pour over insights and try to figure out, okay, great insight, but what do we do about it, right? So that's the fundamental issue with the paradigm that we have. Now with JNAI, that changes. For the first time, with all the computing and the map behind it, for the first time, we are able to create something. We're able to bring the user into the creation process. As an example, there's just one example, but many such. As an example, these are two architecture designs or buildings based on parametric design. To be completely honest, I didn't know anything about parametric design a couple of days ago until I searched for it and now I'm the world's expert on it. And these are two example buildings from this renowned architect that the pioneer in the space called Zaha Hadid. I couldn't design it even if I wanted to, but now I probably could. I could do this in tracking the system, feeding some of these designs from the past, telling the system what my requirements are and boom, I could probably have a parametric design building for my house. Not that I'm building one, but if I wanted to. And you can apply the same examples in any... This is going to revolutionize architecture for sure, but the same applies in many other spaces. Now I'll talk about that in our MailChimp example in a second. The beauty, like I said, suddenly now you are bringing the user into the creation process. You're not just telling the system what to do, but you're basically saying what you want. So that's the... So JNAI is therefore sort of flipping this whole interaction paradigm on its head, because now you can tell the system just not do what I say, but do what I need. So there's an example with MailChimp, right? And even before that, if I go back to the dashboard example, I could probably piece together three different reports, three different performance charts, and come up with an answer. If I knew the question, I probably would come up with an answer. But in many other cases, I may not even be able to do that. So now JNAI is able to help you create something or get to a place where you articulate what the question is and come up with a better answer. With MailChimp, we launched these features recently where as a small business with no marketing department or marketing knowledge with the owner, you can have MailChimp create a complete marketing plan just based on your website. It can, and of course, some additional information that you may want to give, it can create brand assets for you. It can create the copy for the email campaign if you want to run or the same campaign in different channels, including Instagram, all of that in one click and in a few seconds. So therefore, you're now moving from issuing instructions to the machine to telling the machine what you need and let it go figure out. It's not your job. It's the machine's job. It's this combination of AI's ability now to create and this new interaction paradigm where the heavy lifting is now moved, onus is moved from the user to the machine that I believe is getting people like my wife engaged and bringing AI into the mainstream. And this is engagement at scale. It's not engagement with people like you and I. We are also the biased parties, but it's engagement at scale. So that's where we are. Now the question is where is AI headed? Where are we headed? I believe and we believe this into it as well that AI will become a utility-like electricity-like internet. Now, I don't think any of you in the audience, me included, dates all the way to pre-electricity, although I know we get a glimpse of it on a daily basis in Bangalore. But many of us, including me, are pre-internet and I could not imagine what that world looks like. In fact, I couldn't do this presentation if it were not for internet. We believe AI will become exactly that. It become a utility where you couldn't imagine how you operated with our AI. And as an example, with our input QuickBooks product and the ecosystem of products that we have, we are already solving different parts of the facets of a small business. And some of the top three problems, such as how do I get more customers? How do I retain them? How do I get paid? How do I get capital? How do I pay my employees? And so on and so forth. So it's becoming mainstream already. It's becoming part of their day-to-day business. And we are therefore putting AI to work with our platform, AI-driven expert platform. We are putting AI to work with tip the odds in the favor of small businesses and increase their survival rate. In fact, one of our true North goals for 2030 is to up the survival rate in the US at least by 10 points. And we believe AI will be a part of that solution. Now, how do we get that, right? How do we get from engagement to utility? It's not only linear path. Large language models are evolving rapidly. They are becoming adept at managing breadth as well as depth. And depth specifically in specific domains. And in the input example, I'll see the video that I will play in a second. You will see that as well. It's a case where we have not just the breadth but also the depth of in a certain domain and we can make it work. The one challenge right now that we do have is that although the large language model is supposedly super intelligent, it still requires the user to be very smart in providing the right problems for it to give you a good answer. Otherwise, it's going to give you a dumb answer. Which I think is not frankly, not acceptable. That's the best we have right now. And I believe we have to get and we will get better where the system can guide the user to get to the better prompt, better question and eventually to the better answer. But that's on the AI side. I believe we'll get there. On the interaction side, I think we have a lot to do and that's where in all of us and in this audience has a role to play. Just as a recap and we started with issuing instruction to a computer, we even learned that computer language not the other way around. We put a UI on top to visualize. We had conversational inputs. We are trying to reduce those degrees of separation, but we're still bound by the boundaries of language and there are other challenges that I'm not going to, but it's still working its way. And the chat bit experience is one of the best we have right now for J&ER. I will not pretend I know exactly where we will land, but I can say with a lot more conviction that it has to be vastly better than where we are today. So that's my sort of call for all of you, for all of us to dig deeper, think harder in understanding the behavior and also influencing behavior because this is an area where even our customers or users don't necessarily know how to interact and make the best use of. So we also have influenced behavior so that we can leverage AI to build the next generation of experiences that takes AI from engagement all the way to utility. That's a lot on the future, but I just want to stay in the present as someone advised me. So before I play the video, so there's a lot to figure out, but I want to just take a moment in celebrating where we are. We've come a long way and one of the examples of that success is Intuit Assist. It's a new financial assistant. Like I said, it's just a beginning. We have a long way before we can actually deliver an experience where it's done for you as a customer. Where we do the heavy lifting for you. So we're just literally scratching the surface. This is a vision that we articulated more than five years ago. At that time, I don't think we necessarily knew exactly how we will get there, but we had conviction that this is where things are headed. So we started making the investments in building the infrastructure starting from data, clean data. And like I said earlier, we have 2 million AI models running every day. We have more than 800 million interactions every day, every year. And it's only growing. Now, all of that with that foundation already in place with Gen AI coupled with it, we now have what we call Intuit Assist. Like I said, it's a financial assistant that uses the power of Gen AI for delivering intelligent and personalized recommendation. It's the interaction of the third kind that Kaladhar and team were talking about. It will do the hard work to fuel the success of our small business customers and consumers. So that's our vision. I'll leave you with that video. Thanks for listening and hit us up if you want to chat, at least until we find a better interaction paradigm. Thank you. Let me play the video in a second.