 Good evening and welcome back to theCUBE. It's theCUBE after dark. I think it almost- It's an evening? I think it's still light outside, kid. But it's 5.40, specific time. Lisa Martin with the Vellante. It's officially summer. It is officially summer, that is true. We've been having amazing conversations the last two days at Snowflake Summit. This is our second full day of coverage. Nothing compared to the conversation we're about to have. No, AI will predict it will be the best one of the entire conference, don't you think? No pressure. No pressure. They kind of set us up for that before we started. Please welcome our two guests, Lisa Aguilar, VP, Industry Field CTOs and Product Solutions at Data Robot. And Sarah Gibson joins us as well the manager US Business Analyst at Aflac. Yes. Welcome. Thank you. You must get people all the time going, Aflac, Aflac. Yeah, so we won't do that. Outside of the, I mean just doing it twice. Lisa, let's go ahead and start with you. Data Robot, you guys have been on theCUBE before, trusted by 40% of the Fortune 50. Talk about Data Robot for anybody in the audience that's not familiar, what's new? Well, Data Robot is the full end to end AI and machine learning lifecycle platform. We're the only one that actually does that. The things that make Data Robot absolutely unique is how open and flexible we are. So it doesn't matter where you have your data, where you need to get your AI and insights out. Doesn't matter if you have models that have been built since the time of Genghis Khan, we just want to make sure that you're getting value out of them. So everything about us is about making sure that you're getting from what you need to do with AI and predictions, getting it right into the hands of your business so you can actually start to get the value from them. I love that, a true Data Robot. Sarah, talk to us a little bit about your role. We know Aflac works with Data Robot and stuff like that. Give us a little bit of your backstory. So I've been an Aflac for seven years working across operations and sales. And now I just recently become a manager of a team that works mainly focusing on ML projects across both operations and sales, which we've been able to leverage Data Robot to help improve projections and fraudulent data. So it's been really exciting. So connect the dots for us if you would. So where did you start with Data Robot and where does Snowflake fit? So back in 2020, we started with Data Robot and our first big project was Suggestive Selling Tool, which basically takes information from customers that an agent can go to and say, hey, based on our current customers, we think that X, Y, and Z lines of business would be best for you. And that was our first really push into Data Robot. And then within the last couple of years, we've been using Snowflake to now pull the data from Data Robot from Snowflake using an API and then pushing our results and our predictions back into Snowflake from Data Robot. And then who are the primary consumers of that data? Mainly our stakeholders, such as, so for our agents, which would be Suggestive Selling or our operations leaders when we're trying to tell them how many FTEs they need based on projected volumes for say claims or how many calls are going to get into the call center. Mainly like that, mostly internal though. Okay. How have your organizations, both of you responded to the AI heard around the world? That was kind of, obviously you participate directly, you were a leader in that space, so that must have been an interesting conversations going on and then you guys in insurances must be an infinite number of applications. But let's start with Data Robot. We're excited. Finally, I can tell people I do what I've been doing. AI machine learning and people get it. They got it. No idea, yes. For years it was like, what do you do and you how to describe it? So I think what's been great about generative AI is it's actually made it mainstream as a concept. So people have a real understanding of what it can do if you point it at the right problems in terms of solving. So for us, we're extremely excited about it. I think the way that we look at it as generative AI is an incredible technology, but what makes it extremely powerful is when you can add in the predictive AI as part of it. So it's not an or, it's an and for us. So we are looking at ways that getting a chatbot that answers a simple question like, is it going to, you know, who is the most famous runner in the world? That's an easy thing to actually solve. But if you're like in a real business like with Affleck and you're trying to understand who are my best customers that we want to make sure that we understand and then what is maybe the probability that they're going to be much better at wanting to participate in maybe this event with us or do this thing with us. That's where the power of prediction actually comes in. So having them together is what we're really excited about, being able to bring to the market, help our customers and take generative AI as a concept that's kind of cool and fun and actually drive real value from it that impacts the business and that everyone can actually use and benefit from. Is that makes sense to you, Sarah? How will you consume generative AI? Yeah, it's been a slow process for us. A lot of our leadership or inside the business is based on feelings more and so trying to have us push, this is the direction we want to go, this is the direction that we feel is best going to give you the best results because of X, Y and Z which DataRobot gives us prediction explanations which really helps drive our users in the direction we think is best fit for them. And I would imagine people just individually in the organization, oh, everybody's just using chat GPT and for whatever purposes. I mean, we do obviously do a lot of writing here and so editing and things like that is quite useful, writing, marketing, copy and things of that nature. But when you were talking about real enterprise, real business, how are customers thinking about applying this? Because when I talk to IT people, it's mostly the same things we're all doing with generative AI right now but it seems like the potential's so much greater. Yes, there's actually a lot of very unique things that can't be done and we've been working with a lot of our customers on their early stages of their generative AI journey. For example, some of our healthcare customers and clients are actually looking at it for what is the best recipes that I need to do because this patient needs a certain calorie intake or they're recovering from a certain procedure and they need a certain mix of nutrients or something that they may be missing. It's very difficult to understand that maybe based on their food allergies or anything like that. With generative AI, we can start to take a look at that and create custom recipe plans for them. We can also take a look at helping maybe a doctor who's doing specialized work from a scan that they had and they want to interpret what's all the data that's happened in here. I just really need to understand my patient as quickly as possible. It's another use case that we're working on with some of our healthcare customers and clients but there's a variety of different applications with generative AI. I think the thing that we are getting most excited about is in two ways. One, helping make the work of the data science and the data teams much easier. So being able to do things like create net new features from their data using generative AI that didn't exist. So they're augmenting their data sets with things that they may not be able to understand or helping to streamline and then curate all the best coding practices that they've had for creating something and then creating a very specific, personalized chat GPT for cogeneration that's specific to their solution and then also looking at the use cases and applications that we can have. So, Sarah, it's your industry, obviously been around a long time, tends to be pretty staid. I've seen, we talked about this, Lisa, I've seen kind of three reactions to this whole generative AI. The first is amazing excitement and hype and we saw that Monday night, it was great. And then there's a lot of skepticism. Like, ah, we've heard this, we've seen this movie before, lots of promises, technology industry again, and there's a lot of fear. How do you see it? How would you categorize the sentiment in Aflac? I think you have all three. I think you can definitely have, from the analytics side, we're excited for what the future can hold and then you have middle management which is more like, I don't know if I trust it and then there's a lot of us that are like, the fear of what could it do to really damage the image of Aflac for what it can really, so I think that if you combine all three really, you really are going to get the best results because you're going to have those people that are gung-ho and really excited but then you're also going to have someone that's holding them back and pulling on the reins and saying, maybe not too quick, let's start slow and work our way up. Balance of power, yes. Yeah, and how was that balance happening within the organization? I imagine it's quite high up but really from a data strategy perspective, to have an AI strategy as Jensen Wong was saying, Monday night you got to have a data strategy. We do and I think we're going very slowly which is what I think is best for our company because we are such an old company and starting slow and really showing value is what our focus is. Yeah. And you talked, you asked about the fear and that is a real thing, like let's be honest about it. Generative AI is very powerful, it can be very fun. We've got Balenciaga Pope that went viral but I mean, that's not very valuable for an organization and you can do everything with Generative AI if you're not setting it up properly, it's going to give you an answer. That's its whole goal. So I'll give you an example. I asked it, what is the world record for crossing the English Channel on foot? And not only does it answer, it gave a person, it gave the time, it gave exactly the distance and then it gave a caveat of don't, we don't recommend you try this, this is very dangerous. And that's what we mean exactly. Right? So there is a right to be hesitant about it and I think the important thing is if you're looking at embracing it as an actual technology, it's much better to be very methodical and slow because there is a brand reputation risk. You don't want it to come up with an answer just because it can. And there are certain steps that you can take, like making it smaller in terms of your vector database, narrowing down the world, so if that answer doesn't exist in the world of possibilities, it's going to know that and how you prompt it and all the security that you need to do, managing those hallucinations and managing people who are trying to break it. So it's a real journey. So DataRobot would design its Generative AI with that purpose. So if it doesn't know an answer, how does it work? Would it say, I don't know the answer, I need more information? Correct. Okay, so it's specifically designed to do that, whereas obviously chatGPD today anyway isn't. I mean, maybe eventually it will be. And it's how you set up the problem, like how do you use a large language model to create that environment that is safe? So that's one of the things that there's technology and then there's expertise to apply that technology. And because it's so new as a technology, that a lot of that expertise doesn't really exist. And this is where our DataRobot are applied AI expert teams. Because we have so much work, we've worked with clients like Affleck, we work with those teams, we get to see all these problems thousands of times. We've curated that expertise over hundreds of thousands of use cases. So we can help provide those guardrails and recommendations. Is your objective to actually build the large language models or to support the building of large languages? That is a fabulous question. So here's the thing with our perspective and Sarah said it. It's all about values, exactly what she said about her business and wanting to apply AI. And what ends up happening is this is just, what's happening with generative AI is what's happening with AI back in the day. You went from a model section co-poach which is looking at all the LLMs. Look at how many parameters it's been built on. Great, but that doesn't solve the business problem. It's how you apply that algorithm. Then we went to data centric. Well, let's add more data. We're going to get better answers from it. That's data centric. Now we're entering the next evolution of value centric AI. And that is the outcome that you want to get. So you want to really start with what's the objective you're trying to get to, work backwards. Don't focus on the model. The model is literally just an ingredient to getting what you need. But it's like the flour, not the cake. And what you want to start with is what kind of, what do you want? And you have a lot of experience with that. Sure, if you can't solve your problem, even if you have all of these cool technologies and tools behind you, there's no point to having an end issue because you're never going to answer it. So I think that's what we're trying to focus on by starting small is, you know, be methodical and really focus on issues that we think that Aflac can use them for. To start it small, what are some of the steps you might be taking in the near future? I think is getting buy-in from our leadership, from our hierarchy is making sure that he can see value in what we have produced so far to make sure that we can showcase everything that we've done. Well, I mean, despite being conservative, it's whatever you're doing is working. I mean, Aflac's an incredible company. It's extremely profitable and successful. Thank you. I mean, you just look at the stock chart. And you have, but you have a lot of data, right? And you have to do things like pricing. You've got, you know, experts trying to figure out, okay, what's the probability of X, Y, or Z outcome? And so one would think that generative AI could really help, you know, drive value. Oh yeah, for sure. And we have teams that that's all they work on and we think that generative AI can, you know, maybe be a tool on top of all of the technology that we currently have. Well, I got to tell you, I just asked chat GPT, how do you cross the English channel by foot? And it actually said it's challenging and potentially dangerous. But then I kept reading, research the title walk. Title walks across the English channel are only possible during certain low tide periods. So maybe it's possible. Yeah, no. Ha ha ha ha ha ha ha ha. Ladies, thank you so much for joining us, talking about data robot app, like how you're using generative AI and some of the great use cases. We really appreciate you taking the time to talk with us. Thank you so much. We want to thank you so much for watching the program. You have just watched the end of day two of theCUBE's coverage of Snowflake Summit 2023. Tomorrow we've got a great lineup, including the CMO of Snowflake Denise Person. She's going to be joining us. You can find all of our content on thecube.net, all of our editorial and analysis on siliconangle.com. Dave Vellante, Lisa Martin, we'll see you tomorrow.