 Hey everyone, we're excited that you're with us. It's theCUBE, the leader in live tech coverage, Lisa Martin and Dave Vellante here, covering day one of Snook Lake Summit 2023 from Las Vegas. Dave, one of my favorite things to do is with our vendors talk to their customers and really hear how the technology is driving significant business outcomes. We've got Instacart here next. I've been a user for a long time. Do you use it? I do. It's one of the favorite parts of this show is that you're able to talk to customers. A lot of events, you know, it's a lot of tech. That's cool. I love tech. I think the real customer input is always the best. It is. That's that validation that you just can't buy. We've got two guests here with us. Arton Avanes, director of product management at Snook Lake joins us. Mahadev Konar is here as well, senior director of engineering at Instacart. Guys, it's great to have you on theCUBE. Thank you for joining us. Thank you for having us. Thank you for having us. So, Arton, you were saying before we went live, you are eight plus years with Snook Lake. You've seen a tremendous amount of change and news and announcements. Lots unveiled yesterday evening and today regarding the platform, making it easier for organizations to really get so much more value out of their data. Give the audience just a little bit of an overview, if you will, of all the news that has come out in the last 12 hours. Yeah, it's crazy to see. I think when we started, I think I never dreamed of being where we are today. I think it comes back to really the architecture and the engine of Snook Lake that allows you to do so much more. And I think if you look at the announcements, whether it's now the announcement around the large language models that we are collaborating with a number of our partners, the architecture allows you to bring these new capabilities to the data. And we are in a very unique position to leverage the vast amount of data that our customers like Instacart are bringing to the platform. There are other announcements like our ability now to process fully unstructured data and making sense out of documents and PDF files, which is again a testament to our platform and our approach to a single engine and a single product that allows you to do so. So I think I feel sometimes as an observer that I mean, I know there's always that these shows a lot of excitement, a lot of selling, a lot of hype, but I feel like in some ways you undersell what you have. So I want to test this. I know it's a very friendly question, but you have this unified data management platform. And I write down here, a lot of different ways to query. I can store and manage a lot of different data types. I can query those different data types. I can pull together and get an answer that is from those as opposed to some, I won't even mention them. I'll say it, Amazon. I get 12 different databases, okay? And I can't integrate them in a way. I don't think it, unless I'm missing it, that's what you do. This is a very valid observation. I think it's a key differentiation for the years to come. As it relates to building an integrative experience, it is very, very hard to build. If you think about the various ways, as you called out, how you can access data. We started with SQL as the language of databases and data access, and as you heard, our investment around Snowpark and opening up the platform for other languages that you can now use and build on top of Snowflake. Building that in a fairly integrated fashion is very, very hard. And there are a number of examples, whether it's around data governance, security. If you think about our collaboration and data sharing use cases, where now you can share not just data, but applications of any kind that you have built with your favorite language, it is a hard problem. So yes, it is our key differentiation, but it's hard to implement. So the reason I love talking to customers is because I got data frame, I got search, you know, you have a Neva, you got SQL, I got document database, I got graph database, vector database. So my question is, does that matter to you? Or are you trying to solve a more narrow problem? So for us, for that at Instagard, it does matter because all of it is important in the journey to get the answers we want. Like we are interested in vector databases, we are interested in the snowflake data cloud and whatever these guys can help us with, because all these new technologies like vector databases and all, they help us do a lot of more machine learning kind of workload that we want to do. So as and when they come up with new solutions, we are very interested in trying it out. So are you, hey, tongue in cheek, were you working on machine learning before OpenAI invented it last November? Yes, yes, if you look at Instagard, there's a lot of machine learning that goes into the Instagard app, like all the ads that you see, all the recommendations that you see, buy it again that you see, you might like this product or that product and those kind of things. And we are also now getting into very personalized experiences for our customers. So all of that is machine learning based and all of that requires quite a bit of machine learning that we want to do and now with new technologies coming up, we want to do a lot more. That personalized experience is no longer a nice to have. It's essential. It is, exactly. We all expect to be able to go, I want to know if I'm doing grocery shopping, I'm on Instacard or whatever I need, they're going to be able to tell me what I bought last time I can buy it again with a click, with one click, or make suggestions of things that are relevant. So having that personalized experience is table stakes these days. Exactly, it is. We are seeing it from our customers. They really like the personalized. When we're taking that journey, so when we see that things launching, we see that people are excited, the customers are excited about it. And the retailers must be as well to really help them get the picture of what's going on in the store shelves, what's inventory like, how do we deliver that personalized experience as a retailer? Exactly, the retailers are very excited about it. Like they really want to work with us because of this, because of the data that we collect, because of the experiences we can provide, and they want to really see what's working, what's not working, what are the customers interested in, what are they looking at the stores, what are they buying, what are they not buying, those kind of things, they are very interested in knowing. So seeing document integrated, the app, when you acquired Applika, it was like, hmm, okay, how is this going to all come together? And yet again, that theme of a unified platform. So how does Instacard, do you use that capability, that document understanding, or do you plan on using it? So we haven't used that capability yet, document understanding, we don't deal with that much of documents as you would expect, like most of the insurance companies might be using more documents and all. So the document usage on our side is low. We do a lot of image processing and all, but document usage is not that much. So maybe the back office might be. Maybe the back office. So how does this affect, how do you think this affects like the RPA movement? I mean, you can, I would think capture a lot of those use cases that are actually a part of that unified platform. Is that a fair way to think about it, or not necessarily, is it more complimentary? No, I think not every customer will use every feature that we will offer, right? So I think to your point, some customers might not deal with PDFs or don't have PDFs, but we have a number of customers who do. And I mean, I personally have a lot of PDF and I have the search challenge myself, right? So I think if you look at that, not every customer will use every feature, but it will make it very easy. And you have also to think about the future. Maybe today it's not top of mind for certain customers, but it might become a topic. So having a unified approach will at least give you the peace of mind to not think about, okay, I need to purchase yet another tool or another way to process data. How would you describe your data strategy? So our data strategy in the sense that we, first of all, Instagram is extremely data focused. So every decision we make, everything we do is like an A-B test experiment. So everything we do is A-B testing. So our data strategy, all of our data is stored in the Snowflake data cloud and including all the catalog that we get from the retailers, like all the things that are in the stores that are available. Then we have all the data that we, all the data that we collect from the app, all the events we collect from the app. And then we have all the data that we want to provide as a solution to our retailers to see, look at the visibility of what's happening in the stores. So all of this data is provided, we collect in Snowflake and that's our data cloud where we put all the data and we run on top of that. We run all the experiments, everything on top of Snowflake data cloud. How much history do you keep? Is it like a power law where the most recent data is the most valuable? So there's compliance that's required. So we have some compliance restrictions in place. Obviously everybody has that and you guys are very well aware. So we do have data, but as much as we can keep, we do keep that. And then we have to remove the data that certainly is old. Oh, you're required to remove it as opposed to you want to remove it. Yeah, so we don't want to remove it because we can do a lot more with the data that we have, right? So. Talk a little bit about the use case. Instacart is part of the Powered by Snowflake program. Talk maybe both of your perspectives on that program. How has it helped Instacart really build on that data cloud? Oh yeah, so we have a lot of collaboration with Snowflake. So we have been able to figure out how to optimize our data cloud. How to use, we use Snowpipe, sorry. Snowpipe, yes, sorry. Ingestion, yeah. Yeah, the ingestion layers, those kind of things. So just having those experts working with us to be able to know how to use it and where to optimize. As an example, we had a fairly extensive collaboration with Snowflake to reduce our costs. We were spending a lot on Snowflake and these guys knew that we were actually, during COVID we were just looking to expand, just taking the workload in. And then later we realized we didn't take care of the costs. And then they've partnered with us and they've been great partners and we've been able to reduce our costs and through that program with these guys. And it's been great. The Snowflake performance index, we heard today there's been eight months, you've reduced query times by, I think, 15%. 15%. A lot of people might say, what's the big deal, 15%, but explain how you've done that and why that's a big deal. I think it's a big deal for us because of many reasons. One is obviously we wanted for very own personal benefit, understand how much are we improving the workloads. Because if you think about it, we always say we have to earn every single workload. We are consumption based, which is a great model because we are only successful if our customers are successful. And as just mentioned, I think we want to help our customers to build even more efficient workloads on top of Snowflake. So you have to have the visibility in the first place. And what we did is we look at across the entire fleet in Snowflake and that is not an easy challenge because there's a lot of dynamics involved. Customers might change the data volumes, customer might change their application and workload. And so we started with the so-called simpler problem where we said let's look at the stable workloads, which we call the stable warehouses, and look at how much improvement across the fleet we are bringing back to our customers. So 15% might not sound like a big number at a first glance, but if you look at it across the board, it is actually a pretty impressive number, but this is not where we will stop to be very clear. So I want to follow up on this because when I think about it, I think about when I get a new iPhone and it's the same performance because I have so much more data. So you're bringing in what you call a snow pipe, you're bringing in a lot more data through a snow pipe, and yet you're still able to improve query performance by 15%. To me, that's why it's a big deal. It's actually, you're talking about wall clock time, right? Yeah, you're absolutely right. I think if you look at the data volumes that come in, if you think about the workloads that are coming in, that's definitely a consideration, right? But we started by looking at the stable workloads because that's the one that we can really quantify and is the so-called simpler challenge. And I think it's a testament that we want to give also to our customers and prove out that we will and will continue investing and making the platform more efficient and faster. Okay, so that's like your control group and then you'll extend from there. Mahadev, is Instacart using the performance index? Well, we're not really using it, but obviously we get very excited about, yeah, we are very, very excited about the improvements these guys are making and that benefits us because it reduces our costs and makes things faster for us. So we are very, very excited about that. And as part of the engagement, we work together to understand where are the opportunities for optimization, right? We work together to make sure that we give also the right guidance in terms of, maybe you don't need this particular table or this particular data set because you're not being, or you're not accessing the data as frequent as you might think of. So it's really a close collaboration with Instacart and without any other customer of ours. Transparency visibility is one of the things that's popped in my mind when you said that. That is key. When we talk about earning the trust of our customers, it starts with visibility and transparency. Coming into 2023, did the priorities, were they different than they are now? Given all the uncertainty in the economy and the Fed and recession and interest rates, et cetera, et cetera, et cetera, or was it pretty much sort of constant? There is, as I said, there is a certain degree of change in priorities. As an example, we just, before it was very evident that we'd seen in Wall Street and everywhere else that there wasn't, obviously, everybody's worried about profits, but there's not really that much of focus on, hey, growth versus profit, the pendulum was on the growth side. But now we see that the pendulum has swung on the profit side. So everybody's careful about that. Snowflake themselves would be careful about that. So we need to worry about how we are spending the money, how we are using that data so that we can do all the things we want to, but in a very optimized way. So there is certain change we've seen. And so that manifests itself in terms of your ability to dial down consumption, as an example? No, our consumption is always- Other way around, okay. Yeah, our consumption is always increasing. It's always about optimization. So what we've been working with them is that just how to optimize things. Consumption, like in this world, there is no way consumption's going to decrease. It's always going to increase. It's not about slowing down. It's about running faster, but more efficient. Exactly. And that's the theme that we have heard over all the years I have been with Snowflake. It is certainly this year a very dominant theme. But I think it doesn't change at least how we operate. We will always focus on it whether 2024 comes along and it will be hopefully a great year for everyone. We will not change focusing on making the platform more efficient. So yes and no to your question from my side. Yeah, but so doing more with what you have is really the perpetual trend. Yes, correct. Mahat, I have a last question for you. From a cultural perspective, it sounds like Instacart is really, you talked about it, it sounds like it's in the DNA to be data driven. What's next as Dave was saying as we go from 2023 to 2024? What are some of the things that you guys want to do with data to delight those customers that are on the horizon? There's obviously a lot of excitement around this year with all the LLM news that you've seen. So there's a pretty big effort at Instacart. I think you must have seen all the news with Instacart and OpenAI, Instacart plugin in OpenAI, Instacart announcement with Google. So we have a lot of effort going in to figure out what the next stage of grocery shopping or grocery industry looks like with all the LLM data that all the LLM announcements that have come up. So we're working through and figuring out what the new experience for a customer should look like in this new world. So meal planning as an example, just talking to Instacart and saying, hey, what do I do for dinner tonight? And it knows about you. It knows about your personal choices and it tells you, hey, you have a family, it doesn't tell you, but maybe you tell it about your family and then you can figure out the meal plan for you and suggest what to shop and what to get. Oh, I'm going to take that. I need that. That is fantastic. Mahadev, Arjun, thank you so much for joining us on the program, talking about really kind of breaking down the snowflake announcements and how they're really coming to fruition with a company like Instacart. We really appreciate your time. Thank you for being here. Thank you guys. Our pleasure. Thank you so much. Welcome for Dave Vellante. I am Lisa Martin. Stay here. Stay here. Up next, our next conversation. Sonny Betty is here. You're pinned to the table. Chief Information and Data Officer. He's going to be talking about Snowflake's inaugural 2023 data trends report, the transforming role of the CIO-CDO and how Snowflake's reference to itself as customer zero impacts profitability. We'll see you in a minute.