Snowflake Data Cloud Summit 2024 | George Fraser | Fivetran & Prasanna Krishnan | Snowflake

Clips
News
More from Snowflake Data Cloud Summit 2024

George Fraser

CEO

Fivetran

Prasanna Krishnan

Head of Horizon Catalog and Collaboration

Snowflake

Breaking data silos: Fivetran and Snowflake partner for seamless AI integration

Data management is the bedrock of the artificial intelligence era, so Fivetran Inc. and Snowflake Inc. are partnering to help customers bring their data together by making Fivetran a native application on the Snowflake Marketplace.Fivetran and Snowflake have a longstanding collaboration, with Fivetran helping deliver data into Snowflake’s cloud platform. The upcoming release of Polaris will allow customers to access Iceberg data from Fivetran.Fivetran’s George Fraser and Snowflake’s Prasanna Krishnan talk about data management and creating a faster, easier data pipeline

play_circle_outline Introduction of guests

play_circle_outline Fivetran solves data pipeline problem

play_circle_outline Snowflake and Fivetran's Simplified Partnership: Streamlining Data Movement and Procurement Processes

play_circle_outline Leveraging AI for data insights and actions

play_circle_outline Snowflake's Successful IPO Amidst Pandemic: Echoing Arctic Wolf's Aspirations, Real Data Reveals Momentum and Simplicity

Info

George Fraser | Fivetran & Prasanna Krishnan | Snowflake

George Fraser

CEO Fivetran

Prasanna Krishnan

Head of Horizon Catalog and Collaboration Snowflake

search

Rebecca Knight

>> Good morning everyone, and welcome back to theCUBE's Live coverage of the Snowflake Data Cloud Summit, here at the Moscone Center in San Francisco. I'm your host, Rebecca Knight, alongside my co-host and analyst, Dave Vellante. It is Developer Day here at the Data Cloud Summit, but on theCUBE, it might as well be Founder Day. We're having a lot of really cool tech founders.

>> Founders and partners. You and I have done many shows and we've observed ecosystems evolving, it's such an important part of a platform's strategy.

Rebecca Knight

>> Yeah. No, it's really exciting. So with that, I'd like to introduce our next two guests. We have Prasanna Krishnan, head of Collaboration and Horizon at Snowflake. Welcome back to theCUBE.

>> She's back.

Rebecca Knight

>> Exactly. Welcome.

Prasanna Krishnan

>> Thank you for having me back.

Rebecca Knight

>> And George Fraser, CEO and co-founder of Fivetran. Welcome, George.

George Fraser

>> Very nice to be with you, good to be back.

Rebecca Knight

>> Yeah, exactly. We just had your co-founder on the show earlier, but I'd love to hear from your perspective, tell us a little bit about Fivetran and the massive problem the companies and businesses face when they're trying to harness the power of their data.

George Fraser

>> Yeah. So Fivetran is a data pipeline, we solve a very straightforward problem, which is replicating all of the data from all of your systems of records, systems like SAP, Workday, Oracle, Salesforce, you name it, into your data warehouse. And we've been working with Snowflake for many years, since we were both much smaller companies. At this point, almost a third of Snowflake customers use Fivetran to ingest some or all of their data into Snowflake. People love Fivetran because it just works, same reason they love Snowflake.

Prasanna Krishnan

>> Absolutely, yes.

Rebecca Knight

>> Tell us a little bit about your role.

Prasanna Krishnan

>> Tell us about what?

Rebecca Knight

>> Your role.

Prasanna Krishnan

>> Oh, my role. So I'm on the product management team at Snowflake, and my role spans data sharing, the marketplace, native applications, working with partners like Fivetran, as well as Horizon.

>> So we just had one of your customers on, Lyra, and he told us, "We're all on Iceberg." Okay, how are you going to govern that? What's your strategy? He said, "We're figuring that out." So help us understand Snowflake's approach. We've been talking about it all week and where Fivetran fits, so maybe I could set it up. You open source Polaris, which is the technical metadata catalog-

Prasanna Krishnan

>> Yes....

>> and Horizon is the Snowflake, the role-based access controls and all the heavy governance is your product.

Prasanna Krishnan

>> Yep.

>> That's not open source. And so, you're leaning into Iceberg, let the customers put the data where they want, that's what they're going to do. Pick it up from there, what can you tell us about your strategy? And then, George, you're going to tell us where Fivetran fits.

Prasanna Krishnan

>> Absolutely. So we want to give our customers a choice. We want them to be able to use Iceberg as an open table format, and we've always supported Iceberg, and I think Polaris takes that support a step further, being the Iceberg technical catalog that is fully open source. Now, the way Polaris fits in with Horizon is that Horizon is the built-in discovery and governance layer. It provides you a lot of the business and semantic cataloging side of things. And so, Horizon works with Polaris, and so you can use Horizon's discovery and governance capabilities, on top of both data that is natively in Snowflake's format, as well as data that is in Iceberg. And so, for data that is natively in Snowflake's format, Snowflake is also the technical catalog, and for data that is in Iceberg, Polaris is our open source Iceberg catalog, which gives our customers the ability to have their data in one place, and have multiple engines be able to read and write to it.

>> George, where you fit is the customer says, "I want my data in Iceberg, help me get it there," right?

George Fraser

>> Absolutely. The announcements about Iceberg have been great this year. I have nothing but praise for Snowflake's embrace of Iceberg. It really represents doing right by the customer and giving them the freedom to store their data where they want, how they want, even if that means less lock-in for Snowflake, Snowflake has embraced it. I was a little surprised by this. I asked Snowflake's former CEO, Frank, last year, I'm like, "Hey, aren't you worried about less lock-in?" He was like, "You've got to follow what the customer wants. You've got to do it their way." And so, it's great to see. We have several hundred customers that use Fivetran to deliver data to Iceberg to today, and if you want to read that data in Snowflake, you have to go through the Glue metastore right now, which it's not important exactly what that is, but we're looking forward to the release of Polaris, and embracing Polaris as the mechanism to give Snowflake customers access to Iceberg data delivered by Fivetran.

>> So I would say the word I would use for myself is I was perplexed, maybe not surprised, but perplexed, and my perplexion, is that even a word, was I said, "Okay, but then how is Snowflake going to preserve its value-add?" And the answer, I'm realizing, is a combination of Horizon, but also, we're going to compete on the basis of our compute engine.

Prasanna Krishnan

>> Yes.

>> We've got the best one in the world, at least you believe that. I think it's one of the best, if not the best. So we're going to compete with our strength.

Prasanna Krishnan

>> Yes, absolutely.

>> And let the customers decide.

George Fraser

>> And the compute engine has always been Snowflake's biggest strength.

>> Right.

George Fraser

>> Terry, Benoit, and Marcin spent two years in that apartment in San Mateo building it. It was great when it came out, it's only gotten better since then. And so, I think Snowflake will compete very effectively on that dimension.

>> Well, and we said earlier, when Benoit was on stage, he described what I called the hard plumbing, it's all in place now, and what you're doing is building on top of that, bringing in AI, bringing in an application development ecosystem, that's the key to a platform.

Prasanna Krishnan

>> Yes.

>> Okay. So tie that in to your partnership with Fivetran.

Prasanna Krishnan

>> Yeah.

>> How are you guys working together to add incremental customer value over time?

Prasanna Krishnan

>> Yeah, absolutely. So as Benoit also mentioned in his keynote, data is at the center, and customers' enterprise data sits in a variety of places. Not all of it originates in Snowflake, they have their enterprise applications. And so, we and Fivetran are aligned on our vision of enabling you to break down silos between data, bringing together all your data. And this is where we partnered with Fivetran to help bring that data into Snowflake, so that it's now available for you to be able to unlock value from through AI and apps. And so, now, where we are going with this partnership is furthering that to enable customers to leverage Fivetran as a native application on Snowflake, which gives them further governance benefits.

George Fraser

>> Yeah. We're very excited to make this announcement this week, that Fivetran will be available as a native application in the Snowflake marketplace. The reason this is so important is many Snowflake customers spend a lot of time validating Snowflake as an environment where it's safe to put their data, sometimes years doing that validation. Snowflake, for them, it functions like their fourth cloud. And after they've done all that work, it is helpful if we can deploy the parts of Fivetran that move the data into that security perimeter so that they don't have to repeat the entire process again for Fivetran. Over the last year, we've built this capability to execute in a hybrid mode, where you can have the same easy to use experience at Fivetran.com, but the parts that move the data all run on your own infrastructure, or now on Snowflake infrastructure.

>> So I've called it a super cloud, that was my-

George Fraser

>> You have indeed.

>> But you can also monetize as well in the marketplace, right?

George Fraser

>> That is correct.

>> So Snowflake's the icebreaker, and then... Please, describe that.

George Fraser

>> Yeah, that's correct. Customers will be able to use Snowflake credits to pay for data movement into Snowflake through the marketplace running in Snowflake Container Services.

Rebecca Knight

>> Yeah. So talk a little bit more about the value that the customers will actually derive from this partnership, in terms of not having to validate all over again, as you said, but also just what they'll get out of it.

Prasanna Krishnan

>> Yeah, absolutely. I think this goes back to the three hurdles that you run into typically when you're trying to use an application in your enterprise. The first is security approvals and being confident in the security and governance, second is procurement, and then the third is, of course, legal. And we are really eliminating all three of those hurdles with native applications. And what native applications do is really enable you to bring the application to the data, to the customer's security and governance perimeter. And so, with this partnership, as George was saying, we're enabling customers to run Fivetran's logic as a native application, within the security and governance perimeter of their Snowflake account, which enables them to, A, speed up procurement, and then from a monetization or from a paying for this perspective, we are also simplifying that by giving them the option to use the capacity dollars that they've already committed with Snowflake to be able to pay for the application.

>> Can you add some color to that, Prasanna, and George as well? Because native, we hear that term a lot, cloud native, and we're hearing AI native, native apps. I remember when people lifted and shifted their stack into the cloud and they wrap it in a container, and really, they weren't taking advantage of some of the native capabilities of cloud. So that's when cloud native came out, everybody started to talk about that and the additional value that you bring. We're now hearing that with AI. What does native apps mean for the customer in terms of the experience? Are there capabilities native inside of Snowflake that Fivetran can take advantage of and other partners, and what are some of those examples?

Prasanna Krishnan

>> Yeah. So when we say native application, we're really referring to applications that are built using our native app framework to leverage Snowflake's capabilities, including our governance and role-based access controls, and the ability to operate and run this app within the customer's Snowflake account. And so, what this means is that when I install a native application, it's much like installing an app on my phone, I can permission it and I can say, "What can this do in my account?" Just like on my phone, I can say, "Should this app be able to use my camera or not?" And so, I can say, "What tables can this write into, can it run tasks?" And then, I can monitor it. So the capabilities that are available to the application are all of Snowflake's capabilities, from creating tables, to running tasks, to deploying an entire container now using Snowpark Container Services.

>> So it's the experience for the customer, if I understand it correctly, how is that different, pre versus post?

George Fraser

>> Yeah. At the end of the day, from a customer perspective, it's all about ease of use, it's all about breaking down barriers to getting all of your data together, whether those barriers are, like Prasanna said, security or procurement or networking, this is going to be a tool that customers will be able to use to make it easy to say yes to I'm going to get my data into Snowflake in this way with Fivetran through the native application.

Rebecca Knight

>> I want to ask about the partnership between Fivetran and Snowflake, because when you were first talking about it, George, you said, "We're a good product and we work, we're kind of like Snowflake." Talk a little bit about the culture of both of your organizations, and the customer-centric approach, and this joint goal of, as you say, making things easier, breaking down silos. How do you two work together to collaborate to make sure that you are in fact producing the best possible product for customers?

George Fraser

>> Yeah. I think the companies have very similar cultures and very similar product values. We try to minimize knobs, we try to make it simple on the outside, even if that means we have to do more work behind the scenes. At Snowflake, a lot of that is about performance and optimization. At Fivetran, that's about reliability. Fivetran, behind the scenes, we are maniacal about troubleshooting data pipelines. Data pipelines is like... And this is an area where we're maybe more different, data pipelines is like the incidental complexity business. There are so many weird things that all these different data sources do, you cannot believe it. And the real value Fivetran provides is all the people working madly behind the scenes to map out and work around all these funny things that all these data sources do, so that you just get a mirror image of your data in Snowflake as though it was born there. The good news is that because we have 7,000 customers, there is a certain economy of scale. Everyone ultimately has the same bugs, and so the longer we're around and the more customers we have, the easier it is to make it work. And we've been around for 10 years now, so we're getting pretty good at it.

>> I love this notion of minimizing knobs, because there are a lot of knob-turners, a lot of times, developers want to turn knobs, but I feel like we're entering the era where customers don't want to turn knobs anymore. I've had this conversation with Andy Jassy before, years ago, when he said, "Look, we choose to give primitives because it gives us flexibility in the market when things change." Does AI change that? Simplicity always works, but then at some point, markets change and you've got to be able to get in and get to those primitives, do that hard work. That's what you guys have done, that's clearly what you guys are doing as well. Does AI change that knob-turning desire at all in your view?

Prasanna Krishnan

>> I think it doesn't. With AI, ultimately, I still want to be able to leverage AI and get the business value that I want out of it, and be able to do that in a way that is simple and just works. If I don't need to go fine-tune, if I can use AI in minutes, that's a lot better, and so I think the simplicity applies to AI as well.

>> And I meant it the other way, like AI abstracts that complexity, and can be an agent to do some of that knob-turning that people maybe like to do, but it doesn't really add a lot of value, if that makes sense?

George Fraser

>> It's very hard to predict what the impact of AI is going to be on all of our businesses. We're at the first stage of a revolution here. It's 2009, and the iPhone's been out for one year, and everyone knows mobile is going to be really important, but nobody's sure exactly how. I think we have studied, for the last couple of years, how to use the large language models to write connectors. We have made progress on using them internally. When we write a new connector at Fivetran, the scaffold is actually written by a language model, but there always has to be a lot of developer work to go and map out all of the very arbitrary rules that you need to get right in order to correctly replicate data. It's easy to replicate data, it's hard to correctly replicate data. So we haven't seen it turn things upside down for us yet, but we certainly are focused on it and trying to figure out how it'll change the data replication business.

>> So I've got to ask you... Oh, go ahead, please.

Rebecca Knight

>> No, Prasanna, we're at the real moment in time here, and as George said, we're at the first stage of the revolution. I'm curious to hear, from your standpoint, as this landscape is changing so fast, what are the trends that you're seeing right now starting to bubble up that you think we're going to be talking about at next year's summit?

Prasanna Krishnan

>> Yeah, great question. Clearly, AI is not just a fad, it's here to stay, and the way I think of it is, ultimately, in every line of business, I want to go from data to insights, and from insights, I want to take actions, and then actions generate more data, so you have a nice loop. And going from data to insights to 12 steps, and AI is helping us reduce that down to a few steps, and I think that'll continue. In terms of the trends that we're seeing, I see this on two sides of the coin. One is, with AI-powered applications, we're seeing interesting needs for unstructured data, and maybe you're seeing some of that too, that's growing for all the generative AI. And then, the other side of it is we're also seeing new kinds of applications that this enables, back to my point about applications that enable different lines of business to go from data to insights to actions, we see those kinds of applications being built and distributed.

George Fraser

>> I think that what AI is going to do in this context is it's finally going to give us the ability to do something with unstructured text. We have been delivering text data to Snowflake data warehouses for years. We deliver Zendesk tickets, and Slack messages, and sales transcripts, and CMSs with docs, and you name it, and we deliver it as part of a larger schema that contains a lot of metadata about it, but the text is there, it's been there this whole time, but it's very hard to do anything with texts-

Prasanna Krishnan

>> Right....

George Fraser

>> with traditional computer science. And I think language models, their most profound impact from an enterprise data perspective is they are going to give us the ability to actually wrangle and make use of that unstructured text data.

>> Sentiment analysis is a good example, we've seen many demos here at the show. Sentiment analysis has not been that great, in my opinion, and it now has the potential to actually be much greater signals and higher value.

George Fraser

>> Yeah. And other things like that that we can't even predict right now.

Prasanna Krishnan

>> Right.

George Fraser

>> But these language models, in some sense, they actually comprehend that unstructured text, and what are people going to do with that? Time will tell, but they're going to do something, because there's a lot of value in there.

>> And amazing speed, not always, but oftentimes better than humans can, certainly faster.

Prasanna Krishnan

>> Yeah. And it's multimodal, not only text-

>> Yeah....

Prasanna Krishnan

>> but also images and videos and so on.

>> Well, and too, I think as you start to be able to blend the best of multiple models, you guys have done a lot of work with a mixture of experts, that's the first step-

Prasanna Krishnan

>> Right....

>> but if I can build apps that actually can, in near real time, tap multiple models, that's going to open up a whole new layer of value. Like you said, it's very hard to predict, but you can predict one thing, it's going to blow our minds.

Rebecca Knight

>> Indeed.

>> I've got to ask, so Taylor was here, you guys have a pretty big entourage following you, I'm imagining... So you've seen Snowflake, incredibly successful company, great IPO during the pandemic. You guys have raised, I think, nearly a billion dollars. I'm imagining, George, up on the balcony at the New York Stock Exchange, someday, IPO, bunch of people screaming, I don't know if we could... Let's hear it guys. Come on, make some noise. I don't know if that's going to happen. But where do you want to take the company?

George Fraser

>> Our goal is to go public. We've always thought that it makes sense for there to be an independent company that solves the problem of data replication from any source to any destination, and we're on that journey. I can't say exactly when it will happen, but it's closer now than it's ever been.

>> Well, we've had many of these discussions. I was just talking to Nick Schneider, who's the CEO of Arctic Wolf, similar aspirations. He's like, "We're getting ready. We're just waiting to see what the window looks like, and when the time is right, we'll go." What we've seen from the data, the survey data I shared with you earlier, from our partner, ETR, you obviously have a lot of momentum. It's tough sometimes to tell how a private company is doing, other than reading effusive press releases, but we have real data that shows momentum, so congratulations on that.

George Fraser

>> We're growing well. Our message of simplicity is getting through.

>> Great. Thank you.

Rebecca Knight

>> I think Thoreau said it first, but yes, absolutely, simplicity, simplicity, simplicity. Prasanna and George, thank you both so much for coming on theCUBE, a really fun conversation.

Prasanna Krishnan

>> Thank you so much.

George Fraser

>> Nice to speak to you again.

>> Thanks you guys.

Rebecca Knight

>> I'm Rebecca Knight. Stay tuned for more of theCUBE's live coverage of the Data Cloud Summit. You're watching theCUBE, the leader in enterprise tech news and analysis.

George Fraser | Fivetran & Prasanna Krishnan | Snowflake

search

Rebecca Knight

>> Good morning everyone, and welcome back to theCUBE's Live coverage of the Snowflake Data Cloud Summit, here at the Moscone Center in San Francisco. I'm your host, Rebecca Knight, alongside my co-host and analyst, Dave Vellante. It is Developer Day here at the Data Cloud Summit, but on theCUBE, it might as well be Founder Day. We're having a lot of really cool tech founders.

>> Founders and partners. You and I have done many shows and we've observed ecosystems evolving, it's such an important part of a platform's strategy.

Rebecca Knight

>> Yeah. No, it's really exciting. So with that, I'd like to introduce our next two guests. We have Prasanna Krishnan, head of Collaboration and Horizon at Snowflake. Welcome back to theCUBE.

>> She's back.

Rebecca Knight

>> Exactly. Welcome.

Prasanna Krishnan

>> Thank you for having me back.

Rebecca Knight

>> And George Fraser, CEO and co-founder of Fivetran. Welcome, George.

George Fraser

>> Very nice to be with you, good to be back.

Rebecca Knight

>> Yeah, exactly. We just had your co-founder on the show earlier, but I'd love to hear from your perspective, tell us a little bit about Fivetran and the massive problem the companies and businesses face when they're trying to harness the power of their data.

George Fraser

>> Yeah. So Fivetran is a data pipeline, we solve a very straightforward problem, which is replicating all of the data from all of your systems of records, systems like SAP, Workday, Oracle, Salesforce, you name it, into your data warehouse. And we've been working with Snowflake for many years, since we were both much smaller companies. At this point, almost a third of Snowflake customers use Fivetran to ingest some or all of their data into Snowflake. People love Fivetran because it just works, same reason they love Snowflake.

Prasanna Krishnan

>> Absolutely, yes.

Rebecca Knight

>> Tell us a little bit about your role.

Prasanna Krishnan

>> Tell us about what?

Rebecca Knight

>> Your role.

Prasanna Krishnan

>> Oh, my role. So I'm on the product management team at Snowflake, and my role spans data sharing, the marketplace, native applications, working with partners like Fivetran, as well as Horizon.

>> So we just had one of your customers on, Lyra, and he told us, "We're all on Iceberg." Okay, how are you going to govern that? What's your strategy? He said, "We're figuring that out." So help us understand Snowflake's approach. We've been talking about it all week and where Fivetran fits, so maybe I could set it up. You open source Polaris, which is the technical metadata catalog-

Prasanna Krishnan

>> Yes....

>> and Horizon is the Snowflake, the role-based access controls and all the heavy governance is your product.

Prasanna Krishnan

>> Yep.

>> That's not open source. And so, you're leaning into Iceberg, let the customers put the data where they want, that's what they're going to do. Pick it up from there, what can you tell us about your strategy? And then, George, you're going to tell us where Fivetran fits.

Prasanna Krishnan

>> Absolutely. So we want to give our customers a choice. We want them to be able to use Iceberg as an open table format, and we've always supported Iceberg, and I think Polaris takes that support a step further, being the Iceberg technical catalog that is fully open source. Now, the way Polaris fits in with Horizon is that Horizon is the built-in discovery and governance layer. It provides you a lot of the business and semantic cataloging side of things. And so, Horizon works with Polaris, and so you can use Horizon's discovery and governance capabilities, on top of both data that is natively in Snowflake's format, as well as data that is in Iceberg. And so, for data that is natively in Snowflake's format, Snowflake is also the technical catalog, and for data that is in Iceberg, Polaris is our open source Iceberg catalog, which gives our customers the ability to have their data in one place, and have multiple engines be able to read and write to it.

>> George, where you fit is the customer says, "I want my data in Iceberg, help me get it there," right?

George Fraser

>> Absolutely. The announcements about Iceberg have been great this year. I have nothing but praise for Snowflake's embrace of Iceberg. It really represents doing right by the customer and giving them the freedom to store their data where they want, how they want, even if that means less lock-in for Snowflake, Snowflake has embraced it. I was a little surprised by this. I asked Snowflake's former CEO, Frank, last year, I'm like, "Hey, aren't you worried about less lock-in?" He was like, "You've got to follow what the customer wants. You've got to do it their way." And so, it's great to see. We have several hundred customers that use Fivetran to deliver data to Iceberg to today, and if you want to read that data in Snowflake, you have to go through the Glue metastore right now, which it's not important exactly what that is, but we're looking forward to the release of Polaris, and embracing Polaris as the mechanism to give Snowflake customers access to Iceberg data delivered by Fivetran.

>> So I would say the word I would use for myself is I was perplexed, maybe not surprised, but perplexed, and my perplexion, is that even a word, was I said, "Okay, but then how is Snowflake going to preserve its value-add?" And the answer, I'm realizing, is a combination of Horizon, but also, we're going to compete on the basis of our compute engine.

Prasanna Krishnan

>> Yes.

>> We've got the best one in the world, at least you believe that. I think it's one of the best, if not the best. So we're going to compete with our strength.

Prasanna Krishnan

>> Yes, absolutely.

>> And let the customers decide.

George Fraser

>> And the compute engine has always been Snowflake's biggest strength.

>> Right.

George Fraser

>> Terry, Benoit, and Marcin spent two years in that apartment in San Mateo building it. It was great when it came out, it's only gotten better since then. And so, I think Snowflake will compete very effectively on that dimension.

>> Well, and we said earlier, when Benoit was on stage, he described what I called the hard plumbing, it's all in place now, and what you're doing is building on top of that, bringing in AI, bringing in an application development ecosystem, that's the key to a platform.

Prasanna Krishnan

>> Yes.

>> Okay. So tie that in to your partnership with Fivetran.

Prasanna Krishnan

>> Yeah.

>> How are you guys working together to add incremental customer value over time?

Prasanna Krishnan

>> Yeah, absolutely. So as Benoit also mentioned in his keynote, data is at the center, and customers' enterprise data sits in a variety of places. Not all of it originates in Snowflake, they have their enterprise applications. And so, we and Fivetran are aligned on our vision of enabling you to break down silos between data, bringing together all your data. And this is where we partnered with Fivetran to help bring that data into Snowflake, so that it's now available for you to be able to unlock value from through AI and apps. And so, now, where we are going with this partnership is furthering that to enable customers to leverage Fivetran as a native application on Snowflake, which gives them further governance benefits.

George Fraser

>> Yeah. We're very excited to make this announcement this week, that Fivetran will be available as a native application in the Snowflake marketplace. The reason this is so important is many Snowflake customers spend a lot of time validating Snowflake as an environment where it's safe to put their data, sometimes years doing that validation. Snowflake, for them, it functions like their fourth cloud. And after they've done all that work, it is helpful if we can deploy the parts of Fivetran that move the data into that security perimeter so that they don't have to repeat the entire process again for Fivetran. Over the last year, we've built this capability to execute in a hybrid mode, where you can have the same easy to use experience at Fivetran.com, but the parts that move the data all run on your own infrastructure, or now on Snowflake infrastructure.

>> So I've called it a super cloud, that was my-

George Fraser

>> You have indeed.

>> But you can also monetize as well in the marketplace, right?

George Fraser

>> That is correct.

>> So Snowflake's the icebreaker, and then... Please, describe that.

George Fraser

>> Yeah, that's correct. Customers will be able to use Snowflake credits to pay for data movement into Snowflake through the marketplace running in Snowflake Container Services.

Rebecca Knight

>> Yeah. So talk a little bit more about the value that the customers will actually derive from this partnership, in terms of not having to validate all over again, as you said, but also just what they'll get out of it.

Prasanna Krishnan

>> Yeah, absolutely. I think this goes back to the three hurdles that you run into typically when you're trying to use an application in your enterprise. The first is security approvals and being confident in the security and governance, second is procurement, and then the third is, of course, legal. And we are really eliminating all three of those hurdles with native applications. And what native applications do is really enable you to bring the application to the data, to the customer's security and governance perimeter. And so, with this partnership, as George was saying, we're enabling customers to run Fivetran's logic as a native application, within the security and governance perimeter of their Snowflake account, which enables them to, A, speed up procurement, and then from a monetization or from a paying for this perspective, we are also simplifying that by giving them the option to use the capacity dollars that they've already committed with Snowflake to be able to pay for the application.

>> Can you add some color to that, Prasanna, and George as well? Because native, we hear that term a lot, cloud native, and we're hearing AI native, native apps. I remember when people lifted and shifted their stack into the cloud and they wrap it in a container, and really, they weren't taking advantage of some of the native capabilities of cloud. So that's when cloud native came out, everybody started to talk about that and the additional value that you bring. We're now hearing that with AI. What does native apps mean for the customer in terms of the experience? Are there capabilities native inside of Snowflake that Fivetran can take advantage of and other partners, and what are some of those examples?

Prasanna Krishnan

>> Yeah. So when we say native application, we're really referring to applications that are built using our native app framework to leverage Snowflake's capabilities, including our governance and role-based access controls, and the ability to operate and run this app within the customer's Snowflake account. And so, what this means is that when I install a native application, it's much like installing an app on my phone, I can permission it and I can say, "What can this do in my account?" Just like on my phone, I can say, "Should this app be able to use my camera or not?" And so, I can say, "What tables can this write into, can it run tasks?" And then, I can monitor it. So the capabilities that are available to the application are all of Snowflake's capabilities, from creating tables, to running tasks, to deploying an entire container now using Snowpark Container Services.

>> So it's the experience for the customer, if I understand it correctly, how is that different, pre versus post?

George Fraser

>> Yeah. At the end of the day, from a customer perspective, it's all about ease of use, it's all about breaking down barriers to getting all of your data together, whether those barriers are, like Prasanna said, security or procurement or networking, this is going to be a tool that customers will be able to use to make it easy to say yes to I'm going to get my data into Snowflake in this way with Fivetran through the native application.

Rebecca Knight

>> I want to ask about the partnership between Fivetran and Snowflake, because when you were first talking about it, George, you said, "We're a good product and we work, we're kind of like Snowflake." Talk a little bit about the culture of both of your organizations, and the customer-centric approach, and this joint goal of, as you say, making things easier, breaking down silos. How do you two work together to collaborate to make sure that you are in fact producing the best possible product for customers?

George Fraser

>> Yeah. I think the companies have very similar cultures and very similar product values. We try to minimize knobs, we try to make it simple on the outside, even if that means we have to do more work behind the scenes. At Snowflake, a lot of that is about performance and optimization. At Fivetran, that's about reliability. Fivetran, behind the scenes, we are maniacal about troubleshooting data pipelines. Data pipelines is like... And this is an area where we're maybe more different, data pipelines is like the incidental complexity business. There are so many weird things that all these different data sources do, you cannot believe it. And the real value Fivetran provides is all the people working madly behind the scenes to map out and work around all these funny things that all these data sources do, so that you just get a mirror image of your data in Snowflake as though it was born there. The good news is that because we have 7,000 customers, there is a certain economy of scale. Everyone ultimately has the same bugs, and so the longer we're around and the more customers we have, the easier it is to make it work. And we've been around for 10 years now, so we're getting pretty good at it.

>> I love this notion of minimizing knobs, because there are a lot of knob-turners, a lot of times, developers want to turn knobs, but I feel like we're entering the era where customers don't want to turn knobs anymore. I've had this conversation with Andy Jassy before, years ago, when he said, "Look, we choose to give primitives because it gives us flexibility in the market when things change." Does AI change that? Simplicity always works, but then at some point, markets change and you've got to be able to get in and get to those primitives, do that hard work. That's what you guys have done, that's clearly what you guys are doing as well. Does AI change that knob-turning desire at all in your view?

Prasanna Krishnan

>> I think it doesn't. With AI, ultimately, I still want to be able to leverage AI and get the business value that I want out of it, and be able to do that in a way that is simple and just works. If I don't need to go fine-tune, if I can use AI in minutes, that's a lot better, and so I think the simplicity applies to AI as well.

>> And I meant it the other way, like AI abstracts that complexity, and can be an agent to do some of that knob-turning that people maybe like to do, but it doesn't really add a lot of value, if that makes sense?

George Fraser

>> It's very hard to predict what the impact of AI is going to be on all of our businesses. We're at the first stage of a revolution here. It's 2009, and the iPhone's been out for one year, and everyone knows mobile is going to be really important, but nobody's sure exactly how. I think we have studied, for the last couple of years, how to use the large language models to write connectors. We have made progress on using them internally. When we write a new connector at Fivetran, the scaffold is actually written by a language model, but there always has to be a lot of developer work to go and map out all of the very arbitrary rules that you need to get right in order to correctly replicate data. It's easy to replicate data, it's hard to correctly replicate data. So we haven't seen it turn things upside down for us yet, but we certainly are focused on it and trying to figure out how it'll change the data replication business.

>> So I've got to ask you... Oh, go ahead, please.

Rebecca Knight

>> No, Prasanna, we're at the real moment in time here, and as George said, we're at the first stage of the revolution. I'm curious to hear, from your standpoint, as this landscape is changing so fast, what are the trends that you're seeing right now starting to bubble up that you think we're going to be talking about at next year's summit?

Prasanna Krishnan

>> Yeah, great question. Clearly, AI is not just a fad, it's here to stay, and the way I think of it is, ultimately, in every line of business, I want to go from data to insights, and from insights, I want to take actions, and then actions generate more data, so you have a nice loop. And going from data to insights to 12 steps, and AI is helping us reduce that down to a few steps, and I think that'll continue. In terms of the trends that we're seeing, I see this on two sides of the coin. One is, with AI-powered applications, we're seeing interesting needs for unstructured data, and maybe you're seeing some of that too, that's growing for all the generative AI. And then, the other side of it is we're also seeing new kinds of applications that this enables, back to my point about applications that enable different lines of business to go from data to insights to actions, we see those kinds of applications being built and distributed.

George Fraser

>> I think that what AI is going to do in this context is it's finally going to give us the ability to do something with unstructured text. We have been delivering text data to Snowflake data warehouses for years. We deliver Zendesk tickets, and Slack messages, and sales transcripts, and CMSs with docs, and you name it, and we deliver it as part of a larger schema that contains a lot of metadata about it, but the text is there, it's been there this whole time, but it's very hard to do anything with texts-

Prasanna Krishnan

>> Right....

George Fraser

>> with traditional computer science. And I think language models, their most profound impact from an enterprise data perspective is they are going to give us the ability to actually wrangle and make use of that unstructured text data.

>> Sentiment analysis is a good example, we've seen many demos here at the show. Sentiment analysis has not been that great, in my opinion, and it now has the potential to actually be much greater signals and higher value.

George Fraser

>> Yeah. And other things like that that we can't even predict right now.

Prasanna Krishnan

>> Right.

George Fraser

>> But these language models, in some sense, they actually comprehend that unstructured text, and what are people going to do with that? Time will tell, but they're going to do something, because there's a lot of value in there.

>> And amazing speed, not always, but oftentimes better than humans can, certainly faster.

Prasanna Krishnan

>> Yeah. And it's multimodal, not only text-

>> Yeah....

Prasanna Krishnan

>> but also images and videos and so on.

>> Well, and too, I think as you start to be able to blend the best of multiple models, you guys have done a lot of work with a mixture of experts, that's the first step-

Prasanna Krishnan

>> Right....

>> but if I can build apps that actually can, in near real time, tap multiple models, that's going to open up a whole new layer of value. Like you said, it's very hard to predict, but you can predict one thing, it's going to blow our minds.

Rebecca Knight

>> Indeed.

>> I've got to ask, so Taylor was here, you guys have a pretty big entourage following you, I'm imagining... So you've seen Snowflake, incredibly successful company, great IPO during the pandemic. You guys have raised, I think, nearly a billion dollars. I'm imagining, George, up on the balcony at the New York Stock Exchange, someday, IPO, bunch of people screaming, I don't know if we could... Let's hear it guys. Come on, make some noise. I don't know if that's going to happen. But where do you want to take the company?

George Fraser

>> Our goal is to go public. We've always thought that it makes sense for there to be an independent company that solves the problem of data replication from any source to any destination, and we're on that journey. I can't say exactly when it will happen, but it's closer now than it's ever been.

>> Well, we've had many of these discussions. I was just talking to Nick Schneider, who's the CEO of Arctic Wolf, similar aspirations. He's like, "We're getting ready. We're just waiting to see what the window looks like, and when the time is right, we'll go." What we've seen from the data, the survey data I shared with you earlier, from our partner, ETR, you obviously have a lot of momentum. It's tough sometimes to tell how a private company is doing, other than reading effusive press releases, but we have real data that shows momentum, so congratulations on that.

George Fraser

>> We're growing well. Our message of simplicity is getting through.

>> Great. Thank you.

Rebecca Knight

>> I think Thoreau said it first, but yes, absolutely, simplicity, simplicity, simplicity. Prasanna and George, thank you both so much for coming on theCUBE, a really fun conversation.

Prasanna Krishnan

>> Thank you so much.

George Fraser

>> Nice to speak to you again.

>> Thanks you guys.

Rebecca Knight

>> I'm Rebecca Knight. Stay tuned for more of theCUBE's live coverage of the Data Cloud Summit. You're watching theCUBE, the leader in enterprise tech news and analysis.