Snowflake Data Cloud Summit 2024 | Vanja Josifovski | Kumo AI & Brian Rikuda | BET

Clips
News
More from Snowflake Data Cloud Summit 2024

Vanja Josifovski

CEO & Co-founder

Kumo.AI

Brian Rikuda

Executive Vice President, Strategy, Operations & Programming Strategy

BET

Four insights you might have missed from theCUBE’s coverage of Snowflake’s Data Cloud Summit

The Snowflake Data Cloud Summit in San Francisco last week offered an in-depth look at how artificial intelligence, data and open source are continuing to reshape the enterprise world.As Snowflake Inc. moves from its Data Cloud into the AI Data Cloud, the company is seeking to write its next chapter under the leadership of new CEO Sridhar Ramaswamy, supporting an AI-driven environment where customers are demanding data simplicity and value for the business

BET turns to Kumo’s predictive AI solution for attracting and retaining viewers

For major media content providers, such as Black Entertainment Television LLC, or BET, meeting the expectations of its viewers is of paramount importance. This is where a predictive AI solution takes on greater significance.BET’s Brian Rikuda and Kumo.ai’s Vanja Josifovski talk with theCUBE about predictive AI.“We are looking at data in different ways than the industry has looked at data before to start predicting what customer behavior is going to look like,” said Brian Rikuda (pictured, left), executive vice president of strategy, operations and programming scheduling at BET. “We have a range of subscribers that have certain consumption habits

play_circle_outline Exploring the Power of Predictive AI and the Transition to Snowflake: Unveiling Kumo.AI's Simple SQL-like Queries

play_circle_outline The importance of data in understanding customer behavior and improving subscriber retention

play_circle_outline The impact of Kumo on improving subscriber retention for direct-to-consumer platforms

play_circle_outline Kumo's growth as a company and partnership with Snowflake

Info

Vanja Josifovski | Kumo AI & Brian Rikuda | BET

Vanja Josifovski

CEO & Co-founder Kumo.AI

Brian Rikuda

Executive Vice President, Strategy, Operations & Programming Strategy BET

search

Rebecca Knight

>> Good afternoon 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, sitting alongside my co-host, co-analyst, co-founder of theCube, Dave Vellante. We're talking about predictive AI, especially as it relates to media entertainment, which I think is so cool and a little creepy how well they can determine what I want to watch, what I want to listen to, things I might be interested in.

>> Well, as an analyst, I love to make predictions, although my predictions aren't necessarily that accurate.

Rebecca Knight

>> Right. Well, you need a little help from AI. So I'd like to introduce our next guest, Vanja Josifovski. He is the CEO and Co-founder of Kumo.AI. Welcome to the show.

Vanja Josifovski

>> Thank you.

Rebecca Knight

>> And Brian Rikuda, Executive Vice President, strategy, operations and programming at Black Entertainment Network. Thank you so much for coming on the show.

Brian Rikuda

>> Thank you. Thank you.

Rebecca Knight

>> So I'm going to start with you, Vanja. Tell us a little bit about Kumo.AI. I know the company was founded in 2021.

Vanja Josifovski

>> Yeah.

Rebecca Knight

>> Why'd you found it?

Vanja Josifovski

>> You know what? In the last decade or so, we've seen this transition of the data from homegrown system into systems like Snowflake where it can do so much more. 10, 15 years ago in my prior jobs, we built data warehousing ourselves. So you'll get three or four engineers and you will build something data warehouse. Now, nobody would do this today with the same mind. You would use something like Snowflake. So while the data is in there, in this data warehousing, and you can look at the past, you can understand what has happened, because analytics is about the past. You still need to put a lot of effort to see what's going to happen in the future. And the stage a which predictive AI is today with all the tooling available is the same as the stage of data warehousing 10 years ago. You have these small teams of three to five people building stuff in-house, and we need to get it to the point where it's easy, it's a press button and you get the predictions out. And that's why we founded this company. Kumo allows you to do that. You can get predictions as easy as analytics. You can write a simple SQL-like query, but not about the past, but about the future.

>> Brian, we've certainly heard the story before from the technology industry, and every time, whether it's EDW, Hadoop, big data was going to change the way in which we interacted with data. But to your point, Vanja, it's always been looking in the rearview mirror and then projecting forward. The past is not prologue, we all know. Are we at the point finally where the industry's promises are going to be realized by customers? What's your experience been?

Brian Rikuda

>> Yeah, I think absolutely. We are looking at data in different ways than I think the industry has looked at data before, to start predicting what customer behavior is going to look like, to start predicting what are the types of content that customers want to engage with? If we take a step back and think about BET's overall mission, it's to entertain, engage, and empower the Black community. Data now is fundamental to us being able to do that better. If we think about the entertainment segment, for example. With our direct-to-consumer platform, BET+, we have a range of subscribers that have certain consumption habits. They're looking at certain content, but what we don't know is what are the other types of content that we have on our service that they might want to get exposed to, that they might want to engage with? That with our own abilities we might not be able to see and determine, but now with services like Kumo and products, it allows us to look at customer behavior, it allows us to look at engagement patterns and it allows us to look at the inventory of content that we have and start making more dynamic recommendations.

Rebecca Knight

>> Can you back up a little bit and just talk about the challenges of the attention economy, with so many streaming services available now as well as old-fashioned newspapers and books and TikTok, and these are the things that are competing for our attention. And just talk a little bit about the challenges that presents.

Brian Rikuda

>> Absolutely. So more than ever before, we have more options to keep our minds and our eyes occupied at any given point in time. 20, 30 years ago, there was the TV and there was cable, and the introduction of satellite was the big thing. That was like, "Oh wow, this is a new way to consume," what was effectively though cable content, it was just getting delivered through a different mechanism.

>> Right.

Brian Rikuda

>> Now with smartphones, with social media platforms, with all the different direct-to-consumer platforms, FAST, AVOD, insert another acronym, there are so many different services that compete for our attention any given point in time. And so now more than ever, it's more important for us to understand what do customers actually want? And not only what do they want, just generally, but when do they want it? At different points in the day, at different times of the week, at different seasons of the year, what are the types of content that our customers want to be able to consume? And so yes, to your point, it's more competitive than ever, and so it's more important than ever for us to intimately understand our desires and our wants of our customers.

>> Yvonne, what is the secret sauce that you guys have that has allowed you to do this? Algos are plentiful, everybody can use them it's how you apply them, but you can describe how you're different and why are you effective?

Vanja Josifovski

>> Yeah, secret sauces are usually secret, right? So joke aside, I think at nutshell what's happening here is that when we do predictions from tabular data as we have in Snowflake, we have usually multiple different tables. And the traditional way that we've been doing this in the last 15 years is to join that data in a single table, produce signals which are called features, and then learn on top of those features. We've been doing the same thing for the last 15 years. This process, and it sounds a little complicated, and it is. It's a very technical process. You need data engineering, you need data science to produce this training set, and that's why it takes a long time. So why have been doing this for the last 15 years is because relational data is complicated. It's like how do we think about this multiple different tables? So what Kumo has done, it has brought the deep learning and transformer revolution to tabular relational data. At the core of what we do is we represent the data as a graph and then we do transformers over graphs to be able to find out how to combine the signal in all these different tables automatically without human to be able to produce the predictions. The similar revolution happened in computer vision. 15 years ago I wrote code that was finding edges in images, and then we would build classifier in the top. And you can imagine then, we use only a small subset of the image because the edge is a very small part of the image. Today, we use the entirety of the image automatically without features. That's what we have done and that's why we get strong improvements over models that are built by hand and making it a lot quicker, of course.

>> So is it math? You mentioned a graph.

Vanja Josifovski

>> I would say it's a formalism that enables you to learn over the entirety of the data and not on a small summary in forms or features.

>> Okay. So it's that visibility and then that observability space, if I can use that phrase, and then what you do with it?

Vanja Josifovski

>> Yeah, in some ways we have found a way to learn over everything and pump everything in the network architectures to be able to use the signal entirely.

>> And it works?

Vanja Josifovski

>> It does work.

Brian Rikuda

>> Absolutely.

>> How has it affected your business?

Brian Rikuda

>> So we're still in the early stages, just full disclosure, but one of the reasons why we're excited in engaging with Kumo is if we look at subscription based direct-to-consumer platforms, there are two really important drivers. One is subscriber acquisition, the number of new subscribers you're getting to sign up in a given day or a week, for obvious reasons.

>> Sure, yeah.

Brian Rikuda

>> But the other one is subscriber retention. And almost just as important to the subscribers you're adding is you don't want to lose those subscribers that you spent all this hard marketing work to build awareness and to get them to come on board. But once they're on board, they're really, really, really valuable. And so we are looking at all the range of levers. How we found Kumo was we were looking at all the range of that we could pull to improve subscriber retention. If you look across all of the direct-to-consumer platforms, the Netflix and Max, and you name them, we're all trying to solve that retention problem. Because if you could solve the challenge of retention in the weekly or monthly leakage of your subscribers, you actually solve a huge problem for the growth of the overall business. And so where Kumo came in for us was, I alluded to it earlier, but with our vast array of content that we have for our subscribers, our subscribers aren't always going to organically find it on their own. They'd have to search, they'd have to go through all of our catalog and our library and identify content that they think might be appealing. In some cases they won't even know just based upon a title or a short description or even watching a short trailer. And so where Kumo's value comes in is it allows us to look at the viewing patterns of customers at a very specific level and determine what are the other types of content that that customer might enjoy and engage with. We know that as our customers engage with our service more and watch more content, they're less likely to churn out. And so this recommendation engine that can empower a recommendation engine is extremely valuable to us as a result.

Rebecca Knight

>> So does that go as granular as storylines, certain celebrities that are in these kinds of programs? What is the data that you're actually looking at to determine what will keep those people coming back and wanting more?

Brian Rikuda

>> So again, can't reveal everything as far as all that.

Rebecca Knight

>> Okay, fair enough. Fair enough.

Brian Rikuda

>> But we're looking at the expanse of our customer behavior and the different elements of the content that they consume.

Rebecca Knight

>> Okay, okay.

>> Vanja, how do you see this changing the way organizations operate their data estates?

Vanja Josifovski

>> Yeah, as I said before, I think every technology when it starts initially is very complex. And that happened to data warehousing, it has happened to other technologies. And then as it matures, it becomes easier and easier to use. And I think it will be the same here with predictive analytics, predictive AI is that today it requires a lot of skills, a lot of times. And I think as we progress forward, it's going to become more available within the enterprise to a wider set of users, and to users that are not necessarily the highest technical skilled users. I believe that even you can use Kumo.

Rebecca Knight

>> Even you, Dave.

>> Less hyper specialized roles.

Vanja Josifovski

>> I think so too. Much wider use.

>> More access for business people.

Vanja Josifovski

>> And exactly what happened with Snowflake, for example. 15 years ago you had to be able to write Hadoop jobs, you need to be able to manage the cluster and everything else to figure out, for example, the sum of spend per customer in the cloud. Today, everybody can write a simple SQL query and you get that in minutes and it's done. So that's the same type of transformation eventually you'll see, I believe that that's going to change the organizational structure of the enterprise as well accordingly.

>> So if I may, in the Hadoop example, you had to call a services company to help you do it for you.

Vanja Josifovski

>> Either that or hire people.

>> Or hire people, but the people if they're not available, which-

Vanja Josifovski

>> Exactly....

>> there's a lack of people today. So are we at that point yet or we're in that middle ground?

Vanja Josifovski

>> That's exactly what's happening today. You either hire a data science, data engineering team to solve these problems or you hire somebody outsourcing to do this for you. In spite of all these tools around, it's still a highly technical thing to do.

Rebecca Knight

>> But the future is a more democratized, frankly, employment market too, if we are changing these things. So where, Vanja, do you see the future of predictive AI going?

Vanja Josifovski

>> Well, let me pull Kumo as a crystal ball.

Rebecca Knight

>> Exactly. You actually have the data to back it up.

Vanja Josifovski

>> Exactly. I think a couple years back, I was at Snowflake conference as I was really impressed by this image that was a blue globe showing all data centers, all instances of Snowflake around the world, and there were lines exchanging information and there were hundreds of those around the globe. So you realize that this is almost like an organism that's providing analytics at scale in the world. And I feel that that's what's going to happen with predictive analytics. It'll be a global system as Kumo will be around that will provide millions and millions of models and trillions and trillions of predictions, make it available to everybody as easy as a simple question. And data will be all available, the models will be there and you'll be able to ask questions and know everything.

>> I remember that graph. It's not the right name, it's something like Shared Edges or something like that.

Vanja Josifovski

>> Yes.

>> They show it every now and then, and they track that and it's growing quite dramatically.

Vanja Josifovski

>> Yeah. I took that and I made it pink because Kumo's color is pink, and then I used it.

>> Love it. You started the company just as we were exiting COVID, right?

Vanja Josifovski

>> Yeah.

>> And you did two raises in '22 before, I guess-

Vanja Josifovski

>> '21 and '22.

>> '21 and '22, before the market got tight.

Vanja Josifovski

>> Yeah.

>> So where are you at as a company now?

Vanja Josifovski

>> Yeah, as a company, Kumo has been around for about two and a half years. We have an all-star team. Kumo is a complex technical system. As Snowflake, it took a while to build something that works and get the clients. We started getting revenue somewhere last year. We've ramped up really rapidly and we are working on expanding through our partnership with Snowflake. We are one of the SPCS partners. And things are going really well, and we are very excited. We have a few clients that we're working with and moving from there on, scaling.

>> You feel like you have product-market fit?

Vanja Josifovski

>> Oh, I think that's really clear. Just here at the conference, when we show people the product, universally they have, "Wow," response to it. So I think we are on a great path to change the world.

>> You agree?

Brian Rikuda

>> Absolutely. I wouldn't be sitting here if I didn't.

>> Well, the key is retention.

Brian Rikuda

>> Right.

Rebecca Knight

>> We know from you. Exactly.

>> As you said, Brian.

Rebecca Knight

>> Great. Brian, Vanja, thank you both so much for coming on theCUBE. A really fascinating conversation.

Brian Rikuda

>> Absolutely.

Vanja Josifovski

>> Thanks for having us here.

Brian Rikuda

>> Thank you.

Rebecca Knight

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

Vanja Josifovski | Kumo AI & Brian Rikuda | BET

search

Rebecca Knight

>> Good afternoon 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, sitting alongside my co-host, co-analyst, co-founder of theCube, Dave Vellante. We're talking about predictive AI, especially as it relates to media entertainment, which I think is so cool and a little creepy how well they can determine what I want to watch, what I want to listen to, things I might be interested in.

>> Well, as an analyst, I love to make predictions, although my predictions aren't necessarily that accurate.

Rebecca Knight

>> Right. Well, you need a little help from AI. So I'd like to introduce our next guest, Vanja Josifovski. He is the CEO and Co-founder of Kumo.AI. Welcome to the show.

Vanja Josifovski

>> Thank you.

Rebecca Knight

>> And Brian Rikuda, Executive Vice President, strategy, operations and programming at Black Entertainment Network. Thank you so much for coming on the show.

Brian Rikuda

>> Thank you. Thank you.

Rebecca Knight

>> So I'm going to start with you, Vanja. Tell us a little bit about Kumo.AI. I know the company was founded in 2021.

Vanja Josifovski

>> Yeah.

Rebecca Knight

>> Why'd you found it?

Vanja Josifovski

>> You know what? In the last decade or so, we've seen this transition of the data from homegrown system into systems like Snowflake where it can do so much more. 10, 15 years ago in my prior jobs, we built data warehousing ourselves. So you'll get three or four engineers and you will build something data warehouse. Now, nobody would do this today with the same mind. You would use something like Snowflake. So while the data is in there, in this data warehousing, and you can look at the past, you can understand what has happened, because analytics is about the past. You still need to put a lot of effort to see what's going to happen in the future. And the stage a which predictive AI is today with all the tooling available is the same as the stage of data warehousing 10 years ago. You have these small teams of three to five people building stuff in-house, and we need to get it to the point where it's easy, it's a press button and you get the predictions out. And that's why we founded this company. Kumo allows you to do that. You can get predictions as easy as analytics. You can write a simple SQL-like query, but not about the past, but about the future.

>> Brian, we've certainly heard the story before from the technology industry, and every time, whether it's EDW, Hadoop, big data was going to change the way in which we interacted with data. But to your point, Vanja, it's always been looking in the rearview mirror and then projecting forward. The past is not prologue, we all know. Are we at the point finally where the industry's promises are going to be realized by customers? What's your experience been?

Brian Rikuda

>> Yeah, I think absolutely. We are looking at data in different ways than I think the industry has looked at data before, to start predicting what customer behavior is going to look like, to start predicting what are the types of content that customers want to engage with? If we take a step back and think about BET's overall mission, it's to entertain, engage, and empower the Black community. Data now is fundamental to us being able to do that better. If we think about the entertainment segment, for example. With our direct-to-consumer platform, BET+, we have a range of subscribers that have certain consumption habits. They're looking at certain content, but what we don't know is what are the other types of content that we have on our service that they might want to get exposed to, that they might want to engage with? That with our own abilities we might not be able to see and determine, but now with services like Kumo and products, it allows us to look at customer behavior, it allows us to look at engagement patterns and it allows us to look at the inventory of content that we have and start making more dynamic recommendations.

Rebecca Knight

>> Can you back up a little bit and just talk about the challenges of the attention economy, with so many streaming services available now as well as old-fashioned newspapers and books and TikTok, and these are the things that are competing for our attention. And just talk a little bit about the challenges that presents.

Brian Rikuda

>> Absolutely. So more than ever before, we have more options to keep our minds and our eyes occupied at any given point in time. 20, 30 years ago, there was the TV and there was cable, and the introduction of satellite was the big thing. That was like, "Oh wow, this is a new way to consume," what was effectively though cable content, it was just getting delivered through a different mechanism.

>> Right.

Brian Rikuda

>> Now with smartphones, with social media platforms, with all the different direct-to-consumer platforms, FAST, AVOD, insert another acronym, there are so many different services that compete for our attention any given point in time. And so now more than ever, it's more important for us to understand what do customers actually want? And not only what do they want, just generally, but when do they want it? At different points in the day, at different times of the week, at different seasons of the year, what are the types of content that our customers want to be able to consume? And so yes, to your point, it's more competitive than ever, and so it's more important than ever for us to intimately understand our desires and our wants of our customers.

>> Yvonne, what is the secret sauce that you guys have that has allowed you to do this? Algos are plentiful, everybody can use them it's how you apply them, but you can describe how you're different and why are you effective?

Vanja Josifovski

>> Yeah, secret sauces are usually secret, right? So joke aside, I think at nutshell what's happening here is that when we do predictions from tabular data as we have in Snowflake, we have usually multiple different tables. And the traditional way that we've been doing this in the last 15 years is to join that data in a single table, produce signals which are called features, and then learn on top of those features. We've been doing the same thing for the last 15 years. This process, and it sounds a little complicated, and it is. It's a very technical process. You need data engineering, you need data science to produce this training set, and that's why it takes a long time. So why have been doing this for the last 15 years is because relational data is complicated. It's like how do we think about this multiple different tables? So what Kumo has done, it has brought the deep learning and transformer revolution to tabular relational data. At the core of what we do is we represent the data as a graph and then we do transformers over graphs to be able to find out how to combine the signal in all these different tables automatically without human to be able to produce the predictions. The similar revolution happened in computer vision. 15 years ago I wrote code that was finding edges in images, and then we would build classifier in the top. And you can imagine then, we use only a small subset of the image because the edge is a very small part of the image. Today, we use the entirety of the image automatically without features. That's what we have done and that's why we get strong improvements over models that are built by hand and making it a lot quicker, of course.

>> So is it math? You mentioned a graph.

Vanja Josifovski

>> I would say it's a formalism that enables you to learn over the entirety of the data and not on a small summary in forms or features.

>> Okay. So it's that visibility and then that observability space, if I can use that phrase, and then what you do with it?

Vanja Josifovski

>> Yeah, in some ways we have found a way to learn over everything and pump everything in the network architectures to be able to use the signal entirely.

>> And it works?

Vanja Josifovski

>> It does work.

Brian Rikuda

>> Absolutely.

>> How has it affected your business?

Brian Rikuda

>> So we're still in the early stages, just full disclosure, but one of the reasons why we're excited in engaging with Kumo is if we look at subscription based direct-to-consumer platforms, there are two really important drivers. One is subscriber acquisition, the number of new subscribers you're getting to sign up in a given day or a week, for obvious reasons.

>> Sure, yeah.

Brian Rikuda

>> But the other one is subscriber retention. And almost just as important to the subscribers you're adding is you don't want to lose those subscribers that you spent all this hard marketing work to build awareness and to get them to come on board. But once they're on board, they're really, really, really valuable. And so we are looking at all the range of levers. How we found Kumo was we were looking at all the range of that we could pull to improve subscriber retention. If you look across all of the direct-to-consumer platforms, the Netflix and Max, and you name them, we're all trying to solve that retention problem. Because if you could solve the challenge of retention in the weekly or monthly leakage of your subscribers, you actually solve a huge problem for the growth of the overall business. And so where Kumo came in for us was, I alluded to it earlier, but with our vast array of content that we have for our subscribers, our subscribers aren't always going to organically find it on their own. They'd have to search, they'd have to go through all of our catalog and our library and identify content that they think might be appealing. In some cases they won't even know just based upon a title or a short description or even watching a short trailer. And so where Kumo's value comes in is it allows us to look at the viewing patterns of customers at a very specific level and determine what are the other types of content that that customer might enjoy and engage with. We know that as our customers engage with our service more and watch more content, they're less likely to churn out. And so this recommendation engine that can empower a recommendation engine is extremely valuable to us as a result.

Rebecca Knight

>> So does that go as granular as storylines, certain celebrities that are in these kinds of programs? What is the data that you're actually looking at to determine what will keep those people coming back and wanting more?

Brian Rikuda

>> So again, can't reveal everything as far as all that.

Rebecca Knight

>> Okay, fair enough. Fair enough.

Brian Rikuda

>> But we're looking at the expanse of our customer behavior and the different elements of the content that they consume.

Rebecca Knight

>> Okay, okay.

>> Vanja, how do you see this changing the way organizations operate their data estates?

Vanja Josifovski

>> Yeah, as I said before, I think every technology when it starts initially is very complex. And that happened to data warehousing, it has happened to other technologies. And then as it matures, it becomes easier and easier to use. And I think it will be the same here with predictive analytics, predictive AI is that today it requires a lot of skills, a lot of times. And I think as we progress forward, it's going to become more available within the enterprise to a wider set of users, and to users that are not necessarily the highest technical skilled users. I believe that even you can use Kumo.

Rebecca Knight

>> Even you, Dave.

>> Less hyper specialized roles.

Vanja Josifovski

>> I think so too. Much wider use.

>> More access for business people.

Vanja Josifovski

>> And exactly what happened with Snowflake, for example. 15 years ago you had to be able to write Hadoop jobs, you need to be able to manage the cluster and everything else to figure out, for example, the sum of spend per customer in the cloud. Today, everybody can write a simple SQL query and you get that in minutes and it's done. So that's the same type of transformation eventually you'll see, I believe that that's going to change the organizational structure of the enterprise as well accordingly.

>> So if I may, in the Hadoop example, you had to call a services company to help you do it for you.

Vanja Josifovski

>> Either that or hire people.

>> Or hire people, but the people if they're not available, which-

Vanja Josifovski

>> Exactly....

>> there's a lack of people today. So are we at that point yet or we're in that middle ground?

Vanja Josifovski

>> That's exactly what's happening today. You either hire a data science, data engineering team to solve these problems or you hire somebody outsourcing to do this for you. In spite of all these tools around, it's still a highly technical thing to do.

Rebecca Knight

>> But the future is a more democratized, frankly, employment market too, if we are changing these things. So where, Vanja, do you see the future of predictive AI going?

Vanja Josifovski

>> Well, let me pull Kumo as a crystal ball.

Rebecca Knight

>> Exactly. You actually have the data to back it up.

Vanja Josifovski

>> Exactly. I think a couple years back, I was at Snowflake conference as I was really impressed by this image that was a blue globe showing all data centers, all instances of Snowflake around the world, and there were lines exchanging information and there were hundreds of those around the globe. So you realize that this is almost like an organism that's providing analytics at scale in the world. And I feel that that's what's going to happen with predictive analytics. It'll be a global system as Kumo will be around that will provide millions and millions of models and trillions and trillions of predictions, make it available to everybody as easy as a simple question. And data will be all available, the models will be there and you'll be able to ask questions and know everything.

>> I remember that graph. It's not the right name, it's something like Shared Edges or something like that.

Vanja Josifovski

>> Yes.

>> They show it every now and then, and they track that and it's growing quite dramatically.

Vanja Josifovski

>> Yeah. I took that and I made it pink because Kumo's color is pink, and then I used it.

>> Love it. You started the company just as we were exiting COVID, right?

Vanja Josifovski

>> Yeah.

>> And you did two raises in '22 before, I guess-

Vanja Josifovski

>> '21 and '22.

>> '21 and '22, before the market got tight.

Vanja Josifovski

>> Yeah.

>> So where are you at as a company now?

Vanja Josifovski

>> Yeah, as a company, Kumo has been around for about two and a half years. We have an all-star team. Kumo is a complex technical system. As Snowflake, it took a while to build something that works and get the clients. We started getting revenue somewhere last year. We've ramped up really rapidly and we are working on expanding through our partnership with Snowflake. We are one of the SPCS partners. And things are going really well, and we are very excited. We have a few clients that we're working with and moving from there on, scaling.

>> You feel like you have product-market fit?

Vanja Josifovski

>> Oh, I think that's really clear. Just here at the conference, when we show people the product, universally they have, "Wow," response to it. So I think we are on a great path to change the world.

>> You agree?

Brian Rikuda

>> Absolutely. I wouldn't be sitting here if I didn't.

>> Well, the key is retention.

Brian Rikuda

>> Right.

Rebecca Knight

>> We know from you. Exactly.

>> As you said, Brian.

Rebecca Knight

>> Great. Brian, Vanja, thank you both so much for coming on theCUBE. A really fascinating conversation.

Brian Rikuda

>> Absolutely.

Vanja Josifovski

>> Thanks for having us here.

Brian Rikuda

>> Thank you.

Rebecca Knight

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