 Good afternoon, Google fans, and welcome back to beautiful Las Vegas, Nevada. We're here at Google Cloud Next, bringing on the back end of day one of three days of coverage here on theCUBE. My name's Savannah Peterson, joined by analyst Rob Stretche. Rob, what a power pack to afternoon we're about to open. I think this is going to be a lot of fun. I am really excited for these, a lot of partnership with Google and a lot of how you actually make this stuff work around AI in particular. Which is why we have such a fantastic guest. We've got Kathleen from Elastic with us. Welcome to the show, Kathleen. Thank you, I'm glad to be here. Glad you're having me. Yeah, we're having you back. You're a cube pro now at this point. It only takes twice, okay. Yeah, well you're alumni after, so you're an alumnus after, so yeah. Yeah, that's all it takes in most cases, especially in AI, only two prototypes, right? Awesome. And then you're an expert. How's the show going for you so far? Yeah, it's going great. Just the usual fast packed, fast paced, action packed, a lot of really interesting things going on here this year. Yeah, a lot of really interesting talks I noticed too. Elastic is all about search powered AI. Break that down for us a little bit. What are the advantages you're offering to your customers? Well, as you can imagine, all search experiences are a little bit different. And so if you think about almost a year ago from now, we were all talking about this new thing called generative AI. And there are a lot of questions out there like, is search, is traditional search dead? What's going on? How is this going to change the face of the space? And what we found is it absolutely has, but it's actually drawn more attention to the need for some of that traditional search functionality. So if you think about, I have a customer, I want to build a search application that provides this awesome conversational AI experience. It needs the same underpinnings as traditional search applications that people were building years ago. In fact, part of this was we had a good discussion with Sachin Gupta earlier on today about infrastructure in distributed cloud. And he actually referenced Elastic, unplugging AlloyDB on-premise and using in that infrastructure to distribute it cloud, Google distributed cloud, using Elastic for the vector database. And it would seem like, I mean to me again, I've been saying it for a while that Vector is more a feature than it is an actual, you know, entire database. How does, how does that really come out? Like how has that manifested itself within Elastic and with your customers? Yeah, so what we really say over and over again is you think about a vector database and some people are thinking about it in the terms of like storing and searching embeddings. But that's really just the smallest use case. You really also need to understand how are those embeddings getting in there? And then once you have all this information there, like how are you going to retrieve the best information? And that requires a lot more than just that kind of technical embeddings piece. You need to be thinking about, am I going to be able to aggregate my data? What about facets and filtering? Like how am I actually going to run this? Is it, can I do it on-prem? Can I do it on the cloud? Am I going to have the technology around it for my specific use cases that might require things like text search because surprisingly, text search is going nowhere. It's still really applicable to a lot of those search scenarios where I'm an e-commerce site. One of the most common searches I get is a part number. There's no real intent I need to infer from that. I have a part number, tell me more about it. So it still has a place in the search space and you need more than just kind of these more later technologies like semantic search et cetera to really fill a lot of these use cases. Let's build on that a little bit. So that's a perfect example. We're going to an e-commerce site with three million SKUs. Don't need to know much more detail beyond the intent of someone to learn or to purchase or whatever it is they're doing there. What are some of the use cases that really sing for search powered AI? There's just so many. So a customer that we just recently had at one of our user groups, Stack Overflow was talking about the fact that they have over 100,000 people contributing highly technical information to their environment. And within that environment, there are absolutely other people there that would benefit from that information. But that's this perfect scenario for semantic search where the way that I say something and the way I share information may be different from somebody else. So you really need that technology to help you understand what's the intent behind this question so that it can match you with the best results. So that's a different scenario using some different kind of technology that to answer that question. But what we're finding is most people actually need a combination of both. And so you need a situation where text, traditional text search is going to be the best answer. There are others where semantic search is really required to get the most relevant answer. So that's what we're seeing hybrid search really taking off and exploding. You need your technology to be able to recognize, okay, which is the most relevant answer for this particular query? And then surface that to the person asking the question. Is that where your partnership with Google comes in? Yes, that and more. So one of the things that we've really kept at the forefront for Elastic is kind of our background and just being open. And so we recognize that there are different decisions that have different right answers across our customer base. So if you look at something that a lot of people are talking about right now, like what is the right large language model for me to be using? There's a lot of factors that go into making that decision and a lot of different answers that could be absolutely correct. And so what we focused on is making sure that we're partnering with these technology, with these technologies, working closely with them to make sure that Elastic Search works effectively but also better than other things so that people can pick the things that are right for their use case. And Google's a perfect example of that. Where we got together and we built some reference architectures that we could share with our mutual customers about how to build conversational experiences on Elastic Search on Google. And we had a great webinar on that. It was one of our super high attended ones because people are really interested in this topic as you can imagine. Absolutely and I can imagine they also like a little guidance from a company like Elastic on how to prioritize that and how to train those models to know which type of search to default to in that scenario, which is really cool. You've got some power packed logos on your website illustrating that there are some very large well-known brands that have adopted this. You mentioned a use case with Cisco and how quickly they were able to make change over the course of a year. Can you tell us a little bit about that? In fact you had them on with us. Yeah, last time you saw me, we had Cisco here with us and one of the things that I was really impressed by was just how quickly they moved, embracing generative AI and building their own application. And the application that they built was something that helped their customer support engineers find answers for their customers who were calling in with problems. Always a sensitive one, always when you want to handle quickly and efficiently. And so talking to them a year on out where they got time under their belts, they've made improvements along the way, we've partnered with them the whole time. What they have found is that they can attribute 90% of the answers that they're surfacing to their support engineers are coming from this platform and they've been able to save about 5,000 hours a month for these customer support engineers. You have to imagine that has to have a huge impact on customer satisfaction, right? Yeah, absolutely. I'm getting your answers fast and I'm also able to serve more customers now. And your agents feel better too because their time is better utilized and you have a greater sense of purpose for other types of work. Yeah, so just a really great example about being an early adopter, building something really powerful and getting some great results. Do you see that people are coming to Elastica new because of these features and because, I mean again, Elastica's been around for a while and been known in really how you aggregate data. A lot of pipelines into Elastica like you talked about, getting the data in is part of the problem and then being able to do the vectorizing and all of the tagging and what have you. Is this a new introduction to people who may have not thought about Elastica that way? I think, I would say that Elastica has some great recognition across the market. I think that when you think about search, a lot of people think about Elasticsearch right away. But I think that what we've been able to do is kind of take this moment, this market moment that's been presented and really re-educate people on how much we have done over the last several years to really dive in to the promise of machine learning, dive in to the promise of AI. And it's already surfacing in our products. So when the big buzz started happening around vector database, for instance, it's like we're the most downloaded vector database in the market and that may have been a surprise to some people, but it's the truth. We've been around for a while, we've been investing in these technologies and we're helping our customers see the promise. Were you predicting this market moment? I wish I could say that I predicted all market moments, but I think- I mean don't we all for the record, yeah. I would say I was definitely pleasantly surprised by just the massive amount of buzz that happened because I think all of us saw chat GBT and those initial experiences and we're like, this is very cool. But I think it really opened everybody's eyes and minds to okay, this has a place everywhere. I mean, we've got to figure out what that is because if our competitor just do first, then that's a, we're doing a disservice to our customers. So I think that was really exciting and interestingly enough, we had, we just did a developer survey recently where these were elastic and non-elastic developers, but they all had search on their minds. And one of the things that we found were 87% of these developers had a generative IAU's case in mind or in progress, but only 11% had gotten those out into production. And so that's one of the things that we're seeing is when you start kind of going down this path, all these considerations come up like, okay, how am I going to keep my private data safe? How are we going to make sure that we understand the costs around this and working with somebody who's had a long, long time leadership spot in the search space, I think can help there. Absolutely. How are you helping? We had a really interesting insight from Mackenzie earlier. There's a running joke of death by 1,000 POCs at this point. And so how are you helping customers go from experimentation phase to really figuring out their solution at scale? You know, we have a lot of fun doing generative II workshops not for multiple customers at a time or for individual customers if they want to use their own data. And a lot of times we're kind of starting to talk about it in the framework of like a playground where it's like, let's just, you know, let's see what's possible or if you have something in mind or if you're earlier on in the process, let's come in and by the time you leave, you're going to have written your first generative II application. That's got to be really inspiring and empowering for people. Yeah, it is because it's, you know, I think while there's a lot of considerations, our goal is obviously to make a platform that helps you get up and running quickly, helps you ingest data quickly and efficiently and enhance that data as you bring it in so that ultimately, you know, you have a great experience at the end. Makes total sense. Yeah, it really does make total sense. Are there any applications? So we've talked about the logistics and how it works a little bit. Are there any applications of this in the wild or maybe in the Instant Pot of you will, cooking for the future, that really have you personally excited, not just as a company, but as an individual? Well, I think, yeah, we've been having a lot of exciting conversations about just the art of the possible. And I think some of the things that our customers bring to us are like, wow, this is extremely inspiring. And so I think you have your traditional experiences like Cisco has jumped on and like others where I have so much internal data that I want to surface to my employees to help them work more efficiently that that efficiency is only increased. So I think if you'd asked me, which spaces are you seeing the fastest adoption of this technology? I'd say customer success type of, or internal work place search. These are two areas where people are jumping on quickly because it's just so obvious that there are benefits to jumping in this space. But then we've also seen customers go out and do really exciting things about anomaly detection and threat monitoring and things like that that are really helping keep the world safer. And so that's very exciting too. Yeah, I was going to say what was the top use case that you're seeing from your customers? And where are they focusing when they get started a lot of times? Yeah, like I said, I think it's, let's potentially create a experience where our customers can come and ask questions to and have kind of a conversational experience at our support site. So I'm having this issue, how can I solve it? But outside of kind of that traditional, because that in and of itself is not new, but what can be new is if you think about, I know who this is, I know what kind of certifications they have, I know what products they have of mine, I know what may be a recent issue that has surfaced is, all this information can come together with your standard documentation, troubleshooting guides, and provide like a step-by-step, here's, yes, we understand this is what you have, this is what's happening, this is how you can fix it. And lovely to be able to just get a step-by-step guide to solve my problem without having to call anybody, right? Absolutely, and also better for them to, I know how many times I've called up and they don't know how, yes, I know that, I've rebooted it, I've done all this stuff, here's where I am in your call screen, that makes a lot of sense. Yeah, so I think between that and just again, if you think about your corporate internet and evolving that into an interaction scenario where you can ask, what's my 401K policy? Well, I know who you are, I know what region you're from, and in some cases I know what role you are, so if I'm asking a more sensitive question, like what's the compensation ratio for this title at this level, you know, I as an individual contributor probably shouldn't be seeing that information, but as a manager of that type of role, I should, and that's where things like role-based access control become really important, making sure that we only serve the data to the people that have the right to see it, so. That's a really good point, I'm glad you said that. Closing question for you, because time has flown. What do you hope, you gave us that great Cisco example here today, a year later, what do you hope the next time we have you on theCUBE, that you can say a year from now, that you couldn't say today? There's just so much exciting stuff that our customers are working on, I feel like we just have endless possibilities to talk about next year, which we hadn't, you know, necessarily even thought about two months ago, and it's kind of the combination of the creativity of the people that we're working on the customer side, with, you know, a technology that we're willing to hear that input and take the product in the direction that's going to serve them best. I love it, well, we look forward to telling that story with you on stage at the next Google Cloud Next. Kathleen, thank you so much for being here. Rob, fantastic insights and questions, as always, and thank all of you for tuning in around the world here for our live coverage from Google Cloud Next. It's the end-ish of day one of three. My name's Savannah Peterson, you're watching theCUBE, the leading source in enterprise tech news.