 Hello everyone, welcome to the special CUBE presentation. We're here at Google Next in Las Vegas, 2024. I'm John Furrier, host of the CUBE. This is an accelerating innovation with persistent special report. I'm here in the suite in the 60 second floor in the Mandalay Bay with two great guests breaking down the topics, unleashing the potential, creating AI-driven enterprise driving business. Very well, Caroline Yapp, managing director, global AI business and applied engineer at Google. Great to see you. Good to see you again, John. How are you on Banerjee, chief strategy and growth officer persistence? Great to have you on here too. And thanks for inviting us up to your penthouse suite here in Las Vegas. Great to meet both of you. We've had many conversations, both of us together in the past. We've known each other for a long time when cloud was just getting going. And now we're on the front end of a massive way with Jenner of AI. And it's just been really kind of intoxicating to see the excitement and level. And the confidence is getting there. And we're seeing a lot of people certainly enthused. This next level of JNAI is going to have massive technology change, but also business value is the biggest conversation. Show me the proof and show me the results of just a little bit of a win. Maybe just some short win or a big win. And at the same time, it's accelerating and changing digital transformation, which we've talked a lot about with persistence. So Caroline, you've got a unique view at Google. How do you see this? Because you're in applied engineering, you're looking at the landscape. The business value seems to be the topic. Oh yeah, over the last year, I've met with about a thousand customers. And this is on a, from a global perspective. And it's been really nice because most of the conversations have been at the C suite and at the board level. And they're realizing that they actually need to use more of AI to get their insights. And I go, I usually ask them, I say, okay, do you really need AI? What's your business strategy for the next, you know, two, three, five years? And then, do you have to say something to your shareholders because everybody wants you to use AI? Or can you actually just say, here are all the analytics, automation, machine learning things we've already done, which now we can accelerate results for because Generative AI is allowing us easier for us to actually query all of our data sets to get to that business value drivers that you might need. And so it's been so nice to see business and IT having conversations again. Because, you know, as you mentioned, we've known each other for a long time. IT never really got to have conversations with business other than deliver the system for me. And so now it's so great that we're changing that dynamic again with AI and getting to start to see what those values are. Even inside at Google, you know, earlier today, Brad Calder gave us some of the stats. Our own developers are seeing, you know, 40, 50% production increase, less, 55% less unit tests, for example, I think was one of the stats. And it really is very meaningful, even for a company like us. And the business success has been great. It's watching Google bring out a workspace which is a user experience phenomenon that's growing really fast with Generative AI. Fits the bill with people seeing with Generative Generating Value. And then the back end, you need the horsepower, you need the chips. And so all these things are driving right down to the digital transformation changeover. Now, question on the persistence. We've had many conversations about digital transformation. There's a pre-Gen AI digital transformation journey. Now there's the post-Gen AI marker. So if you have to put a marker down and say, Gen AI happened, okay, everyone's excited, the hype's up, but there's still legit things happening. There's now an accelerated changeover to digital transformation. That's the business value. What are you guys seeing now from that standpoint? How has that changed or accelerated or more? What's your view on that? You know, first things first. I think so, you know, we sat on the keynote earlier in the morning today and I was just thinking that this was almost like if we were to recreate the movie, 2001, A Space Odyssey, and let's say we did a movie in 2030, what would that movie look like? And to me, it felt like the movie is actually unfolding in front of eyes right now, here, right? I mean, it's no longer an act of fiction. It is happening here now and you can touch it and you can feel it. I think it's from our vantage point and even if I speak on behalf of our entire industry, digital transformation was really about looking at a whole bunch of projects to digitalize internal operations or spawn of digital native businesses, right? And it was almost like a long laundry list of things. Whereas if we talk about, you know, the advent of AI, AI has been going on now for close to a decade right now, but I would say in the last 15 months, you know, it's really becoming ubiquitous. It's becoming, you know, it's becoming practical and it's also getting much core to the business. It's becoming more affordable. It's no longer left to data scientists. It's left to people like Carolyn and, you know, John, you and everybody out here, right? So in our minds, you know, if you talk about the customer conversations we are having today, it's all about saying that, look, I'm a retailer and I want to drive an uptick in my revenue or I want to drive a completely different paradigm from a customer experience standpoint. What's the solution? It's not saying that, hey, can I go and actually get this stack in and go work on modernizing that particular bit of the cloud. So I think so the narrative is now starting more top down rather than bottom up. I think that's one stark difference that we are saying. Go ahead, continue. And then I guess the second thing is, you know, I mean, we've also been working with all our clients in the last 15 months since Generative AI has come on and we are now, I was saying the last quarter or two, you're starting to see many clients saying that, look, it is no longer an experiment. It's about really making it real and starting to push enterprise level KPIs. How do I take it to production? So I think the digital transformation to me has become more about business transformation than digital transformation with digital just being a tool in the toolkit. That's how we are seeing it. It's interesting, you mentioned that. We're hearing that same thing here at Google Next. You're seeing on stage the announcements from Google vids, which is like really mind blowing. Also Thomas Kurian said, grounded with Google search, another big announcement that they announced which means that they're bringing all their data to the table. Then he said right after they said, grounded in your enterprise data, which is interesting because he's comparing the enterprise data as a grounding opportunity, meaning there's value in the data. So this seems to be the theme this year is, and you're seeing retrieval, augmentation generation, or RAG being very popular, why? Because you can use some data with it and use AI, not building some algorithm, you could get value. So there's a theme of, to use a baseball analogy, get on base, get a win, get a single. And so getting wins is a theme because we're experimenting, but businesses are starting to rethink their transformation, which is different because it's not like the old days with BI team will run something, the data scientists will get something, they're data people, they're not systems people, they're not writing code, they're not thinking about how things interact. So the theme this year that we're seeing is, the business landscape is very much about the system, not just the data, but the data is critical and they might look different. So this seems to be the new thing we're unpacking on theCUBE is, okay, one, what is the business view system? And then two, how do I use my data? I think for us, our last Google Cloud Next was in August. So Google Cloud Next 2023 was in August and now we're in April. You can see the stark difference in even just the enterprise experience since August to now. But from the time we moved into not just changing vertex and expanding vertex I should say into having like the model garden and vertex conversations and vertex search, we always told clients that if you want to reduce hallucinations, it has to be grounded or fine tuned or creating adapters to then have it where it's using your data to augment the answers because otherwise you're just using a large language model which knows lots about everything but really next to nothing about your business nor should you be putting that data, which is your IP, out there for a public model to learn from you. And so I really like that the clients who were experimenting to your point are now moving into production and one of the panels I'm on is actually about that. It's moving to action, right? How do you take all the experiments and now go into production and what are the right metrics to even measure because AI still is very much a probabilistic kind of experience. It is absolutely an experience. So then how do you give it some of this more deterministic or persistent, pun intended, for the results to be there because you need to be able to measure these KPIs. You need to be able to show the metrics and I completely agree with you. It's very much a business transformation conversation now and how AI is the enabler versus just being a very techie tool, deep in data scientists, just machine learning. It's finally in a way people can really grok and really feel and actually and it gives them, it actually makes them feel something. It's really good. It's interesting, the Google relationship persistent, you guys have a great working relationship. They got all that Google brain power and all that scale. You guys have expertise with customers. It's interesting you mentioned about the eight months from the last Google Next. The big thing last Google Next was the model garden which I love, was to love the name model garden. It implies something's growing in there. I feel like something's happening but they've added some things to it. Model building and agent builder. So again, back to the next level. Okay, here we go. This is now the business conversation. The agents are hot, not chatbots, agents that are doing things with data. You can take small data sets and get some wins internally, right? So I'm seeing a lot of customers cannot risk the external app side of it but go internal and say, hey, you know what? I'll knock a win out by taking some code. I'll ingest it or I'll do some mundane tasks. Oh yeah, talk about the importance of this phenomenon of experimentation. That's almost next level dot connecting but you still got to get the momentum and the wins that we were talking about. This agent building, how hot will agents be this year? And what does that mean for the business conversation? Yeah, I mean, I think that goes back to the previous question, right? Which is agents are all about saying that it's a business problem, right? So Caroline wants to get the dress of the rock star who that she watched and here all of a sudden she's saying, well, I want that checkered shirt. It's not about the agent going and saying that, hey, find me the cheapest checkered shirt. It's about Caroline saying that, hey, you know what, this particular singer who performed at that concert at the Royal Philharmonic, you know, I want that shirt in a place which is close to me at the right price point and is available because I have a concert coming up the day after tomorrow and I want to go solve that problem, right? And it's all conversational, right? So the agent is not about solving a small piece of that problem in that entire workflow but it's about taking the entire problem statement and finding you that checkered shirt. So in my mind, whether it is develop a productivity, whether it's security agents, whether it is agents on the CX and on the business process transaction side, I think so you're going to see more and more proliferation of agents solving end-to-end business problems to assist what usually was done by humans rather than having chat boards or RPA or any of those smaller, you know, what I would call niche applications. So in my mind it's pervasive, it's going to happen. How about the value of both of you guys? We don't mind common because that's a great example. It's a simple example, not use case of ordering but now the data has to be available. Now it's not like yesterday's data modeling. Again, you have unstructured data, you might have different databases. The intent is a query essentially so there's some reasoning involved. So you got kind of all the elements happening on AI. You got the reasoning, you got a query, a prompt, the answer, but you got a reason. It's got to do some reasoning. And you got to figure out, okay, is the data available because it's not a hallucination if it doesn't see anything. There's no data, there's a big void there. So data is critical having access to the right data. I think that it's also critical to identify as much of that as possible in the design. And I really like that people are really thinking about design now more so than before because I like what you said about the agents and it's not just the chat board. In the past, you know, conversational design and conversational architectures with chat boards were awful in the sense of it was so difficult to maintain. You could have five, 600 different head intents for just something like billing inquiry. But now on the generative aspect, it doesn't matter how someone's asking the question. You know that it's just about the billing inquiry, which means these are all of the different answers that it could be. But like you said though, if you don't design it correctly and identify these are the data sources that are needed to answer that billing inquiry question correctly, then you end up with the hallucination. And so it doesn't necessarily mean you need a large data set to answer billing inquiry. It only needs access to where the bills are. It doesn't need access to the rest of it, you know? And like you said, the checkered shirt, right? And it's local. You can really have identifiers of where the person is. That should just give you, there's also narrow down the data set that you need to actually deliver that experience. And so I think I'm really happy to see that people are digging in more in the design versus just assuming that everything is, you know, a nail because you're also a hammer. This is the changing role of the person at the table, sitting at the big table, call it the CISO, CIO, CEO. The IT people have to come up to the table with them, not just be back holding the store down and grinding hard on IT. They got to come up and level up. And the next question is, what's the prerequisites? So as you guys sit down with customers and craft the strategy and figure out the data, you have to think, okay, what's some of the prerequisites needed to do this? What would you say to that if that was the question? Because everyone says, okay, I'm in. I get it. The bridge to the future is pretty clear. What's working now? What's the reality? Is it even available? So what's those steps? What are some of those prerequisites that can you share that customers need to be aware of to start moving down the road of AI for the future? And look, I mean, you know, 10 out of 10 customers will come and tell us that, okay, you know, what we have these prioritized use cases and now I want to take it to production. And in every single one of them, the prerec is that what is the data source, right? So, and in that, if you think about it, it's not just about available data that's available to the enterprise or data that's available, you know, in a public, you know, large-angle model. But it's about saying that in certain cases, you know, pick a drug maker as an example. You're talking about accelerating molecular research, right? And there is only so much data that's available. But then how do you start building synthetic data? How do you start combining structured and unstructured data with synthetic data to be able to really accelerate the life cycle of defining that molecule? And that molecule, the beauty is, and you know, I'd actually written about this in a LinkedIn post a few months back in the peak of daily pollution, which is to say that, you know, people have bronchial issues. But the thing is that this is all accelerating, you know, things like, you know, cancer and so on and so forth. Now cancer is, you know, completely built, the treatment of cancer is completely built on Western, you know, strains, but the strains are different. How do you come and find out the right genomic set in Asia to be able to build drugs which are tailored to the strains in Asia, right? That's going to happen all on the back of the structured unstructured data and synthetic data. And I guess that's where we as a company come in because what we do well and where we thrive in is to really help our customers get equipped to be able to solve that problem of finding that right drug, to treat that right strain of lung cancer in Asia. So yes, to answer your question, whether it's retail or healthcare or fraud and financial services, I mean, data is everywhere and that's really where the future is. I think your point about the data and that workflows and unique situation for the enterprise is their IP. And you bring the power of scale to it and they can win. Final question before we get the closing statements on this session and thank you for your time. Appreciate it. Give an example of generative AI in action that you would tell a friend or a non-techie out there that this is real proof point. If you guys can conjure up an example that you've seen maybe from your best customer call or customer use case where the doubting, the doubters out there, the haters or whatever you were in the anti-AI, I don't understand it's generating stuff. They don't, some people aren't getting it. They'll come on board. But give an example. This is a situation where it's so obvious that that's the future. Can you guys think of a good story? I will actually talk about construction. I know a lot of people are always surprising. I wrote a blog post a few months ago about what we saw over the last year from meeting all of my clients. And I said, you know, there's these two industries and luxury and construction with the two surprising industries who are really benefiting and leveraging generative AI. And the company is actually part of Lendless Digital and it's called Podium. So it's P4D and it launched in December in Singapore. And one of their customers is actually the Housing Development Board of Singapore who have to work on about 16,000 either new or refurbishing properties for the housing market in Singapore. Where it got really interesting is you think about construction, takes forever to do bids. You have to know all of the setbacks, all of the requirements, what can be retail, what can be residential, how many units can you have? Is it going to be within transit? How do you do all the analysis? And so they use generative AI to put a system together where they could just go, here's the bid, tell me what needs to be, you know, responses to. And then is this going to be a good model for us? What's the maximum housing I can have? What's the maximum retail I can have? What citizen services or resident services do I need to provide? Now they can go from months to two weeks or less to even decide if they want to bid on a project. And based on the shape of the land, they can just outline, I think it needs to look like this. And it just goes, yep, if you do it this way, maximum is 15 stories, you will have, say, 50 units of housing with four rooms. And you know, here's how many retail capacity that you've got. And this is what the Urban Development Plan is going to look like. Construction, it's happening. It's real. It's not an obvious example, but you go, wow, that's really a lot of value. Yes, it's a lot of value, especially for, you know, think of Singapore, that's not a lot of land. So they have to be really, really smart with what they do and all the services that they provide. And so I think that it's been great to see how the technology and the tools are being leveraged in such a way that it's going to deliver real value. And that's just one of many, you know. It's a real, it's a real world example, too. Yes, it's a real world example. Oh yeah, what do you have in your top of mind that you could point to and say, this is a deep needle moving example proves the point. Yeah, and look, I mean, in our own worlds, you know, in the role that I play, we spend an enormous amount of time just preparing for, you know, announcing to the street how we've done, right? And if you talk about the whole life cycle of, you know, preparing for the board meetings and stuff like that, I can honestly see that there's a ton of stuff that can actually be automated and done on the back of, you know, just generate API. And, you know, whether it is reporting, whether it is investor, road shows, whether it is, you know, going and actually creating Q and A's, you know, just the amount of research that's possible. I mean, now we're talking about a matter of a day rather than a matter of a month, just order of magnitude, but maybe on a lighter note, I can almost envision that the next time there's a football world cup that happens and betting on the MVP player, you know, who's scored the maximum goals and who had the best, you know, hit rates, predicting that I can almost envision that Betters would be out of business because AI would have done that job and there will be no betting industry left. So there is- It'll be simulation. It'll be simulation, exactly. They're going to know every shot, every foot mark on the field. I mean, it's going to be completely a day to driven, almost getting to know with sensors. I mean, it's an IoT world, awesome example. All right, let's close it out. The future of AI built organizations is coming. We see it being embedded in immediately into features of apps, that's instance value. It'll maybe take the shape into more operationally or reorganize businesses, system wide, not just data. What do you guys see as the future and how can businesses and partners be prepared to work together? What's the future look like? I think upskilling is probably the biggest thing everybody needs to get moving on today. And a lot of people assume that AI is just going to take huge swaths of jobs. And I've had to tell people it might remove tasks. It's, and it will change jobs because the tasks won't no longer be required. So I think it's really important, just for us also as an industry, if I may say, as leaders and also employers that we need to make sure we have really good human capital upskilling, but also re-skilling of folks so that they can move into these sort of like future roles where AI is absolutely second nature and not just another tool, but it's really part of the core system. The more the AI pieces fade into the background and the more people can actually deliver the value of what humans can bring to the table is key. I mean, even today, right? What is the best thing in the design? There's always reinforcement learnings for human feedback and human in the loop. So we have to be part of it. I loved your healthcare use case. We're so nomadic, people live everywhere, work everywhere, but if all of the medicines here is only to treat folks who always were here, then it won't know how to treat those of us who are migrants who have moved here and now live here. So I like that. So I think the re-skilling, upskilling, we do really important. And being generous, having the data. So final point, role of partnerships, your relationship with Google, pretty strong. What's the future look like from your perspective? Two comments. One is, I think that this is just a start. And even though we've been working with Google now for the last decade right now across a multitude of common clients, I feel we've just about started. So the potential opportunities are enormous. That's point 1.2. I think so to your comment on talent, I couldn't agree more with that. On the other end of the spectrum is that companies who will take very deliberate moves in terms of adjusting their business model to the business model of the future will be the ones who'll be able to really exploit the value from this. The rest of them will keep experimenting. And I think that's where, again, the power of someone like Google and Persistent comes together because we're talking about really making that applicable, making that real, and actually moving the needle for our common clients. So that's how we think about it. Awesome. Thanks so much, Caroline. Great. Data-driven, AI-driven. All this has now changed in the landscape and accelerating with AI. Thanks for taking the time. Thanks for having us, Sean. Accelerating innovation with Persistent. I'm John Furrier of theCUBE. You're watching a special presentation here at Google Next 24 on the 62nd floor. Here at the Mandalay Bay. Thanks for watching.