Snowflake Data Cloud Summit 2024 | Matthew Scullion | Matillion & Jeremy Post | Coca-Cola Consolidated

Clips
News
More from Snowflake Data Cloud Summit 2024

Matthew Scullion

Founder & CEO

Matillion

Jeremy Post

Manager of Business Intelligence

Coca-Cola Consolidated

Coca Cola Consolidated transforms workforce with AI for all

Many companies wanting to implement artificial intelligence often lack experts in data and AI learning, but Matillion Ltd. aims to help organizations such as Coca-Cola Consolidated empower all their employees to run data pipelines.Coca-Cola Consolidated’s Jeremy Post and Matillion’s Matthew Scullion and talk about the importance of democratizing data.“The reason we don’t all have the data that we need is because there aren’t enough data engineers to do all the work,” said Matthew Scullion (pictured, right), founder and chief executive officer of Matillion. “The problem is even more urgent and even more acute [with generative AI]

play_circle_outline The Crucial Role of Snowflake as Matillion's Indispensable Partner in Accelerating Data Engineering Teams' Productivity

play_circle_outline Automation and reduction of toil as a result of using Matillion for data processes.

play_circle_outline Transforming Business Processes and Improving Customer Service with Generative AI in the Data Productivity Cloud

play_circle_outline Importance of building generative AI pipelines due to the disruption it can cause

Info

Matthew Scullion | Matillion & Jeremy Post | Coca-Cola Consolidated

Matthew Scullion

Founder & CEO Matillion

Jeremy Post

Manager of Business Intelligence Coca-Cola Consolidated

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, California. I'm your host, Rebecca Knight, sitting alongside my co-host, co-analyst, co-founder of theCUBE, Dave Vellante. Dave, we've just had lunch. I could go for Coca-Cola.

>> I love to have repeat guests on, especially ones that bring customers.

Rebecca Knight

>> Well, indeed.

>> Which you do every time.

Rebecca Knight

>> And speaking of Coca-Cola, I'd like to introduce CUBE alum, Matthew Scullion, and he's the founder and CEO of Matillion. Welcome back, Matthew.

Matthew Scullion

>> Thank you so much for having me, Rebecca. Yeah, CUBE alum. I'm wondering if I overstay my welcome. This is going to be my fifth, sixth time on this.

Rebecca Knight

>> You're a CUBE veteran, I should say.

Matthew Scullion

>> Yeah, yeah, exactly. Thank you for having us today.

Rebecca Knight

>> And Jeremy, post manager of business Intelligence at Coca-Cola Consolidated. Welcome.

>> Thank you. I'm glad you had me today. I'm happy to be here.

Rebecca Knight

>> Well, we're excited to hear about your experiences, but Matthew, let's start with you. Talk a little bit about Matillion and what you're doing here at the Snowflake Summit.

Matthew Scullion

>> Well, Snowflake Summit, one of the most important tent pole events of the year for Matillion. Snowflake, incredibly important partner to our company. But jointly, together, we help our customers, like Jeremy at Coca-Cola Consolidated, accelerate their time to value with data. Matillion's platform, the data productivity cloud is designed to wildly increase the productivity of data engineering teams, and it's an important problem. No one has all the data that they need. 70% of the work of making a data use case come to life is data engineering. The only real way you can square the circle is to make an individual data engineer or data engineering team wildly more productive or to get more players on the pitch by changing the skills profile required to do data engineering. And those are the two things that our platform's designed to do. This year though, I think a particularly important year in our industry and at this particular Snowflake Summit, we've been really excited to announce our next big swath of generative AI pipeline building functionality on the data productivity cloud support for Snowflake cortex, for the Arctic large language model, and for Snowflake Container Services as well, meaning that customers build out data and GenAI pipelines on data productivity cloud, they can run them using Cortex functionality and on Snowflake containers. So super exciting. Really important event for us. Thrilled to be here, particularly joined by many of our customers, including Jeremy.

>> You always bring the customers. We love that on theCUBE. And yeah, the Container Service, interesting, announced, I guess a couple years ago, big one last year, but now we're seeing a lot of the Nvidia innovations get in. We've seen NeMO, we've seen NIMs. Jeremy, tell us about your role at Coca-Cola. What's your data journey been like? When did you find out about Snowflake and start adopting it and bring in Matillion?

>> Well, I got started in BI, as most people do, as a truck driver. As we know, everybody gets started in BI as a truck driver, right?

Rebecca Knight

>> Standard career path. Yeah.

>> Be quiet.

>> Yeah. Yeah. It's a required skillset. And Coca-Cola Consolidated presented kind of a very unique situation where they had a private fleet of trucks like a lot of companies do, and they recognized that they were taking stuff from their plants and their manufacturing facilities and delivering it to distribution centers. And then they were coming back empty, which is very expensive for any trucking company. So the thing that they identified as a way to turn a cost center into a profit center. And so they spun that private fleet off into a four hire trucking company so that now that Coke is the largest customer of Red Classic, and they're also the owner of Red Classic, but now Red Classic is free to go out and broker other freight and go pick up other freight and bring it back. And so it was a whole separate company. And I was brought in to do business intelligence and data analytics with them, and they kind of got their own fate for a long time. And so they needed an ETL solution that would help them with a very small number of people. We didn't have the same level of people there. We only had a few people in it and a few people in BI. And we needed a solution that would allow us to work with it effectively and quickly to get things standing up that don't need 20 people and you don't need highly technical people to do it because there's a component for that really for everything. And so it allows somebody like me with not a huge technical background to come in and do work. And if a crusty old truck driver can do this stuff, just about anybody can given just a little bit of effort.

Matthew Scullion

>> Jeremy, I don't think we should accept for a moment the crusty old truck, but it's such a brilliant story because it's a micro example of the macro problem. The reason we don't all have the data that we need is because there aren't enough data engineers to do all the work. And I think, as I was mentioning a couple of minutes ago, here at the summit, a lot of the talk is about the next vanguard of data innovation around generative AI and building the pipelines that our businesses all need there. The problem is even more urgent and even more acute. There is even fewer people that can build out these sophisticated GenAI pipelines. So it's wonderful to hear Jeremy's story and hear an example of how we can save people with the domain expertise and turn them into highly productive data engineering teams. And I think that's what we've managed to do together.

Matthew Scullion

>> Yeah, and I think democratization of data is where the future is, and it's only going to become more pronounced when there are fewer and fewer technical people trying to do those things. And I think that the people that accept that and try to influence that change are the ones that are going to be successful, and the people that can work and the companies that can work with those folks that are less technical are going to be successful because democratization is going to steamroll you. It's happening. I was at around 20 years ago when we were trying to keep that from happening and keep that data there, and it's Pandora's box. It's out now, and everybody sees the value in data and the value in AI, even though it's still in its infancy, all the power of the things you can do with it.

>> It's quite an amazing career story. So many of us, we see the world, you have... The data is plentiful, but insights are lacking. So can you take us through how you've used some of that data, turned it into insights and what it's done for your business?

>> Yeah, so I've worked for three different trucking companies over my career in about 15 years, and one thing they all had in common is they didn't have a really good handle on where all their equipment was. Tractors are a little bit easier to track because they've traditionally had GPS on them or satellite communications, and they have a driver that's actually communicating their location. But with trailers, they didn't have that, right? We have trailer tracking providers, but you might be interested to know that at least the first two trucking companies I worked for put bounties out because they lost trailers all the time, because these things go and they sit and nobody knows where they're at. And one of the big challenges that I was presented with is we have three companies. We have our own company trailers and we have two leasing companies that we lease from, and we didn't know where all of those are. And they all had different tracking providers. In fact, one of the companies had two different tracking providers. So now I'm presented with... I've got to do four different API calls just to get my locations, and then another three APR, four API calls to get what's the active equipment, and then another one to find out where our landmarks at so we can tell where our trailers were. And I use Matillion to do all of that by myself. It was one person working on this. My team started with it, and then the plan kind of morphed and I ended up having to take it across the finish line myself. My team did a wonderful job, so don't let me take all that credit. But without Matillion, if I'd had to do a lot of the heavy lifting myself, it would have been very difficult for me. And so that's what it provided me, just an ability to do that. And now we know where our trailers are, how long they've been there. Are they at a landmark? How many do we have and how many do we need there? And where do we need to move them from?

>> And then you can utilize them, which you couldn't before.

>> Yes. Yes. We can turn them in if we're leasing them and they're not being used. We can talk to our customers and say, "Hey, we've had a trailer that sat there for two years. Are you using it for storage? Do we need to unload it and get freight loaded on and get it moving?" Or "Hey, we need to service these things. Where are they at so we can get them in for service?" There's just a lot of different reasons why we needed that data and why it's so important to us.

>> Can you describe the old way versus new way? In other words, you said, "If I didn't have Matillion, I would've had to do this... I couldn't have done it myself," or you would've had to bring in a team. Let's assume you found the people. What would you have had to do that you didn't have to do with Matillion? Can you compare and contrast?

>> Yeah. So 10, 12 years ago, it would've been you talked to a developer and they develop a custom solution, maybe in JavaScript or C-Sharp or some other application's language to move that data around, and it would kind of be an ad hoc case by case basis, right? I can just build an orchestration out and then I say that's the way I want to do it every time, and just copy and paste, and copy and paste, because that's where all good developers get everything from, is Stack Overflow and ChatGPT. I'm just copying and pasting. I've never had an original idea, right? I just find something that... I've got a problem, and I find something that someone else does, and I try to implement that. But Matillion gave me a way where I could just go in and I could see these orchestrations visually, and I could work in what I was comfortable with. So I've been working in SQL for 10 or 12 years now, and I'm really, really good with that. And so there's a lot of applications out there that maybe wouldn't fit me because they aren't SQL heavy. But Matillion has... You look at stuff I build in Matillion and it's all these SQL components. It's just one after the other because that's what I'm comfortable in. But if I go into Matillion and I'm a Python developer, I can do that. I know how to laterally flatten JSON and unstructured data, but it's very tedious. Well, there's a component I can just pull over and say, "I don't have to do that. I'm just going to pull it over here and let Matillion do the work for me." So it takes out hours of work that way when you can just drag a component over, change a few settings, and you have your data.

Rebecca Knight

>> And not particularly interesting work too. I mean. It's really taking away the toil, the scut work.

>> Yeah. Yeah.

Matthew Scullion

>> Rebecca, that's the thing. Jeremy, such an amazing story. Thank you. But also, the only thing I'd very respectfully slightly disagree with you about, nervously because Jeremy's my customer, is I don't think it's 10 to 12 years ago, right? So we heard Sridhar from Snowflake talking about Snowflake's vision for enterprise data management at the keynote, whenever that was, day before yesterday, I think. Time has a funny quality at this event, but the key messages that Snowflake we're putting forward, extremely compatible with how Matillion thinks about the world, is around the simplicity around the single platform, the security and compliance and stability, the basic idea that we want to turn up at Snowflake, or in fact on Matillion, and just get on with making our businesses better. Now that contrasts Snowflake from other cloud data platform providers and other technologies where it's a bit more assemble your own open source, and I think of it as the Mercedes sedan that just works and is lovely and comfortable and gets you where you need to go. Like Snowflake compared to the F1 car, which is really cool, but needs 40 people just to start it up, right? And that's the data integration, data productivity layer, and now AI productivity layer, that's exactly what Matillion tries to do as well. I think Jeremy's just so eloquently explained that perhaps laborious work, which can be fun if you're an engineer or an enthusiastic, but if what you're focused on is optimizing your fleet of logistics, tractors and trailers and making your business more money as productively and quickly as possible, we kind of want to get that stuff out of the way. I think that's what Snowflake wants to do. That's certainly what Matillion has always been focused on. I think. Jeremy, you just brought it to life so beautifully my friend, so thank you.

>> Thank you.

>> So how is AI going to change this, generative AI specifically? What kind of new use cases are you finding and envisioning with customers?

Matthew Scullion

>> Yeah, so we've both, Dave, been around for a few of these mega trends.

>> Right.

Matthew Scullion

>> Part of the trick is figuring out if it is one. This one definitely is. The world is changing our pace. Every aspect of how we were live and play is going to be changed, we hope for the better, by generative AI. But coming back to our world, there's maybe two levels to answer the question out. What are customers doing? And then what are we doing as a technology vendor? From a customer point of view, I like to think of it, we may have even discussed this last year, that there's an adoption curve that we go through in our mind as CEO's of companies like Jeremy's, of CIO's, chief data officers. This time of year ago, it was maybe at the interest or denial stage. It then goes through the curiosity and existential fear stage. Then there's the excitement and action stage, and it feels to me like we're somewhere between excitement and action right now with customers just starting to lay down their first generative AI use cases. And we've seen those at early Pioneer customers, even internally at Matillion, where we've been building out gen AI use cases to improve our own business. And the effects can be profound. The amount of time that you can take out of business processes, the improved customer service that you can offer, most of what we've seen has really been augmenting the power of humans, by the way. It hasn't been replacing them. It's been making them go faster and be better informed, and once again, Rebecca, cutting out some of that manual work that we all do in life, right? On the platform side, there's been two big areas that we've been focusing on. So first of all, we've been focusing on using generative AI to make Matillion's data productivity cloud even better at what it's always done. So it's a data productivity platform. It helps Jeremy and his team go faster at bringing together API data to drive business insight. We want to use generative AI to get better at that. And so we've built a copilot into our software, so you can just ask for the pipeline to be built. It will now auto document what your pipeline does, and so other people can figure it out later. And that's part of why Matillion pipelines require 70% less time to maintain over time than other ways of building pipelines.

>> Sorry, what percent?

Matthew Scullion

>> So Matillion customers typically spend 70% less.

>> 70.

Matthew Scullion

>> Yeah, it's huge.

>> I thought I heard 17. 70.

Matthew Scullion

>> So six-zero percent faster to build a pipeline in the first place. Seven-zero, 70% less time maintaining it over time.

>> Thank you.

Matthew Scullion

>> That adds up to a 270% ROI using Matillion build data pipelines historically. We'll get that better over time as well, but 3x is pretty good. The real exciting stuff perhaps is if you think about this thesis that the world's always needed more data but can't get it because we don't have enough data engineers, well, that's true, but more urgent and more pronounced for generative AI. It's more urgent because generative AI is so disruptive. We can completely revolutionize our business with it. So if we don't and our competitors do, that is a real problem. And we have less abilities to do anything about it because there are even fewer engineers in the world that know how to build out GenAI pipelines. So over the last 12 months since we last met, what we've been doing is quickly imbuing the data productivity cloud with the capability of building generative AI pipelines, rag pipelines in the visual composable way for which we've always been famous. So Jeremy's taught through how he can pick from the palette and bring in APIs. Well, now you can pick from the palette and bring in data from the hundreds of sources we support already. You can then chunk that data, vectorize that data into the common vector data sources like Pinecone and Postgres . You can then prompt into LLMs like Arctic, like OpenAI, or the LLM of your choice. You can take that data, and then embellish it into your Snowflake scaler data in the Snowflake warehouse. Or in fact, you can send it onto the destination systems, like Salesforce or Zendesk or some internally built system that you're using. Very quickly, using the skills that we all have in our businesses today, kind of common old garden data engineering skills. We can compose these generative AI pipelines. And we think this is a huge need. We're really proud of the progress we've made on it. Really thrilled this week at Snowflake Summit to be announcing the cortex support and the container services support, which also allow you to bring in those custom large language models if you need to use those as well. That's the direction we're taking it, Dave.

>> Jeremy, for you, this just drives more automation into your business, takes away some of that toil that Rebecca was talking about.

Matthew Scullion

>> The things that really excite me about GenAI, and I'm seeing more of it here, I saw it at Tableau a few years back, is that everybody wants to be an analyst, and we talked about that democratization of data. And now, one of the biggest barriers to that is convincing people that, "Hey, you have to understand this data very well to really truly analyze it." Now, they're able to ask large language models, plain English questions to get stuff out of their data. And then the other thing, I think most engineers have been in the existential dread phase of all this, right? Because data infrastructure can be really bad. And I think one thing that this is going to drive that's going to make everybody's lives better is AI and GenAI is not going to work, and it's only going to be as good as the data you feed it. And it's going to force these CDOs and CIO's and CTOs to get with their people and say, we have to get this right because there's too much upside to be left on the table if we don't. And it's going to help solve that problem for us and get us better data. And then our users are going to be able to get at their data without having to be data experts, right? And it's that democratization making things easy for people that aren't really technically skilled because eventually, hey, data's the most single most valuable commodity on this planet, and it's only going to get more valuable.

Rebecca Knight

>> And you're a living, breathing example of it. Well, Jeremy and Matthew, thank you both so much for coming on theCUBE. Really great conversation.

Matthew Scullion

>> Thank you for having us, Rebecca.

>> 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're watching theCUBE, the leader in enterprise tech news and analysis.

Matthew Scullion | Matillion & Jeremy Post | Coca-Cola Consolidated

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, California. I'm your host, Rebecca Knight, sitting alongside my co-host, co-analyst, co-founder of theCUBE, Dave Vellante. Dave, we've just had lunch. I could go for Coca-Cola.

>> I love to have repeat guests on, especially ones that bring customers.

Rebecca Knight

>> Well, indeed.

>> Which you do every time.

Rebecca Knight

>> And speaking of Coca-Cola, I'd like to introduce CUBE alum, Matthew Scullion, and he's the founder and CEO of Matillion. Welcome back, Matthew.

Matthew Scullion

>> Thank you so much for having me, Rebecca. Yeah, CUBE alum. I'm wondering if I overstay my welcome. This is going to be my fifth, sixth time on this.

Rebecca Knight

>> You're a CUBE veteran, I should say.

Matthew Scullion

>> Yeah, yeah, exactly. Thank you for having us today.

Rebecca Knight

>> And Jeremy, post manager of business Intelligence at Coca-Cola Consolidated. Welcome.

>> Thank you. I'm glad you had me today. I'm happy to be here.

Rebecca Knight

>> Well, we're excited to hear about your experiences, but Matthew, let's start with you. Talk a little bit about Matillion and what you're doing here at the Snowflake Summit.

Matthew Scullion

>> Well, Snowflake Summit, one of the most important tent pole events of the year for Matillion. Snowflake, incredibly important partner to our company. But jointly, together, we help our customers, like Jeremy at Coca-Cola Consolidated, accelerate their time to value with data. Matillion's platform, the data productivity cloud is designed to wildly increase the productivity of data engineering teams, and it's an important problem. No one has all the data that they need. 70% of the work of making a data use case come to life is data engineering. The only real way you can square the circle is to make an individual data engineer or data engineering team wildly more productive or to get more players on the pitch by changing the skills profile required to do data engineering. And those are the two things that our platform's designed to do. This year though, I think a particularly important year in our industry and at this particular Snowflake Summit, we've been really excited to announce our next big swath of generative AI pipeline building functionality on the data productivity cloud support for Snowflake cortex, for the Arctic large language model, and for Snowflake Container Services as well, meaning that customers build out data and GenAI pipelines on data productivity cloud, they can run them using Cortex functionality and on Snowflake containers. So super exciting. Really important event for us. Thrilled to be here, particularly joined by many of our customers, including Jeremy.

>> You always bring the customers. We love that on theCUBE. And yeah, the Container Service, interesting, announced, I guess a couple years ago, big one last year, but now we're seeing a lot of the Nvidia innovations get in. We've seen NeMO, we've seen NIMs. Jeremy, tell us about your role at Coca-Cola. What's your data journey been like? When did you find out about Snowflake and start adopting it and bring in Matillion?

>> Well, I got started in BI, as most people do, as a truck driver. As we know, everybody gets started in BI as a truck driver, right?

Rebecca Knight

>> Standard career path. Yeah.

>> Be quiet.

>> Yeah. Yeah. It's a required skillset. And Coca-Cola Consolidated presented kind of a very unique situation where they had a private fleet of trucks like a lot of companies do, and they recognized that they were taking stuff from their plants and their manufacturing facilities and delivering it to distribution centers. And then they were coming back empty, which is very expensive for any trucking company. So the thing that they identified as a way to turn a cost center into a profit center. And so they spun that private fleet off into a four hire trucking company so that now that Coke is the largest customer of Red Classic, and they're also the owner of Red Classic, but now Red Classic is free to go out and broker other freight and go pick up other freight and bring it back. And so it was a whole separate company. And I was brought in to do business intelligence and data analytics with them, and they kind of got their own fate for a long time. And so they needed an ETL solution that would help them with a very small number of people. We didn't have the same level of people there. We only had a few people in it and a few people in BI. And we needed a solution that would allow us to work with it effectively and quickly to get things standing up that don't need 20 people and you don't need highly technical people to do it because there's a component for that really for everything. And so it allows somebody like me with not a huge technical background to come in and do work. And if a crusty old truck driver can do this stuff, just about anybody can given just a little bit of effort.

Matthew Scullion

>> Jeremy, I don't think we should accept for a moment the crusty old truck, but it's such a brilliant story because it's a micro example of the macro problem. The reason we don't all have the data that we need is because there aren't enough data engineers to do all the work. And I think, as I was mentioning a couple of minutes ago, here at the summit, a lot of the talk is about the next vanguard of data innovation around generative AI and building the pipelines that our businesses all need there. The problem is even more urgent and even more acute. There is even fewer people that can build out these sophisticated GenAI pipelines. So it's wonderful to hear Jeremy's story and hear an example of how we can save people with the domain expertise and turn them into highly productive data engineering teams. And I think that's what we've managed to do together.

Matthew Scullion

>> Yeah, and I think democratization of data is where the future is, and it's only going to become more pronounced when there are fewer and fewer technical people trying to do those things. And I think that the people that accept that and try to influence that change are the ones that are going to be successful, and the people that can work and the companies that can work with those folks that are less technical are going to be successful because democratization is going to steamroll you. It's happening. I was at around 20 years ago when we were trying to keep that from happening and keep that data there, and it's Pandora's box. It's out now, and everybody sees the value in data and the value in AI, even though it's still in its infancy, all the power of the things you can do with it.

>> It's quite an amazing career story. So many of us, we see the world, you have... The data is plentiful, but insights are lacking. So can you take us through how you've used some of that data, turned it into insights and what it's done for your business?

>> Yeah, so I've worked for three different trucking companies over my career in about 15 years, and one thing they all had in common is they didn't have a really good handle on where all their equipment was. Tractors are a little bit easier to track because they've traditionally had GPS on them or satellite communications, and they have a driver that's actually communicating their location. But with trailers, they didn't have that, right? We have trailer tracking providers, but you might be interested to know that at least the first two trucking companies I worked for put bounties out because they lost trailers all the time, because these things go and they sit and nobody knows where they're at. And one of the big challenges that I was presented with is we have three companies. We have our own company trailers and we have two leasing companies that we lease from, and we didn't know where all of those are. And they all had different tracking providers. In fact, one of the companies had two different tracking providers. So now I'm presented with... I've got to do four different API calls just to get my locations, and then another three APR, four API calls to get what's the active equipment, and then another one to find out where our landmarks at so we can tell where our trailers were. And I use Matillion to do all of that by myself. It was one person working on this. My team started with it, and then the plan kind of morphed and I ended up having to take it across the finish line myself. My team did a wonderful job, so don't let me take all that credit. But without Matillion, if I'd had to do a lot of the heavy lifting myself, it would have been very difficult for me. And so that's what it provided me, just an ability to do that. And now we know where our trailers are, how long they've been there. Are they at a landmark? How many do we have and how many do we need there? And where do we need to move them from?

>> And then you can utilize them, which you couldn't before.

>> Yes. Yes. We can turn them in if we're leasing them and they're not being used. We can talk to our customers and say, "Hey, we've had a trailer that sat there for two years. Are you using it for storage? Do we need to unload it and get freight loaded on and get it moving?" Or "Hey, we need to service these things. Where are they at so we can get them in for service?" There's just a lot of different reasons why we needed that data and why it's so important to us.

>> Can you describe the old way versus new way? In other words, you said, "If I didn't have Matillion, I would've had to do this... I couldn't have done it myself," or you would've had to bring in a team. Let's assume you found the people. What would you have had to do that you didn't have to do with Matillion? Can you compare and contrast?

>> Yeah. So 10, 12 years ago, it would've been you talked to a developer and they develop a custom solution, maybe in JavaScript or C-Sharp or some other application's language to move that data around, and it would kind of be an ad hoc case by case basis, right? I can just build an orchestration out and then I say that's the way I want to do it every time, and just copy and paste, and copy and paste, because that's where all good developers get everything from, is Stack Overflow and ChatGPT. I'm just copying and pasting. I've never had an original idea, right? I just find something that... I've got a problem, and I find something that someone else does, and I try to implement that. But Matillion gave me a way where I could just go in and I could see these orchestrations visually, and I could work in what I was comfortable with. So I've been working in SQL for 10 or 12 years now, and I'm really, really good with that. And so there's a lot of applications out there that maybe wouldn't fit me because they aren't SQL heavy. But Matillion has... You look at stuff I build in Matillion and it's all these SQL components. It's just one after the other because that's what I'm comfortable in. But if I go into Matillion and I'm a Python developer, I can do that. I know how to laterally flatten JSON and unstructured data, but it's very tedious. Well, there's a component I can just pull over and say, "I don't have to do that. I'm just going to pull it over here and let Matillion do the work for me." So it takes out hours of work that way when you can just drag a component over, change a few settings, and you have your data.

Rebecca Knight

>> And not particularly interesting work too. I mean. It's really taking away the toil, the scut work.

>> Yeah. Yeah.

Matthew Scullion

>> Rebecca, that's the thing. Jeremy, such an amazing story. Thank you. But also, the only thing I'd very respectfully slightly disagree with you about, nervously because Jeremy's my customer, is I don't think it's 10 to 12 years ago, right? So we heard Sridhar from Snowflake talking about Snowflake's vision for enterprise data management at the keynote, whenever that was, day before yesterday, I think. Time has a funny quality at this event, but the key messages that Snowflake we're putting forward, extremely compatible with how Matillion thinks about the world, is around the simplicity around the single platform, the security and compliance and stability, the basic idea that we want to turn up at Snowflake, or in fact on Matillion, and just get on with making our businesses better. Now that contrasts Snowflake from other cloud data platform providers and other technologies where it's a bit more assemble your own open source, and I think of it as the Mercedes sedan that just works and is lovely and comfortable and gets you where you need to go. Like Snowflake compared to the F1 car, which is really cool, but needs 40 people just to start it up, right? And that's the data integration, data productivity layer, and now AI productivity layer, that's exactly what Matillion tries to do as well. I think Jeremy's just so eloquently explained that perhaps laborious work, which can be fun if you're an engineer or an enthusiastic, but if what you're focused on is optimizing your fleet of logistics, tractors and trailers and making your business more money as productively and quickly as possible, we kind of want to get that stuff out of the way. I think that's what Snowflake wants to do. That's certainly what Matillion has always been focused on. I think. Jeremy, you just brought it to life so beautifully my friend, so thank you.

>> Thank you.

>> So how is AI going to change this, generative AI specifically? What kind of new use cases are you finding and envisioning with customers?

Matthew Scullion

>> Yeah, so we've both, Dave, been around for a few of these mega trends.

>> Right.

Matthew Scullion

>> Part of the trick is figuring out if it is one. This one definitely is. The world is changing our pace. Every aspect of how we were live and play is going to be changed, we hope for the better, by generative AI. But coming back to our world, there's maybe two levels to answer the question out. What are customers doing? And then what are we doing as a technology vendor? From a customer point of view, I like to think of it, we may have even discussed this last year, that there's an adoption curve that we go through in our mind as CEO's of companies like Jeremy's, of CIO's, chief data officers. This time of year ago, it was maybe at the interest or denial stage. It then goes through the curiosity and existential fear stage. Then there's the excitement and action stage, and it feels to me like we're somewhere between excitement and action right now with customers just starting to lay down their first generative AI use cases. And we've seen those at early Pioneer customers, even internally at Matillion, where we've been building out gen AI use cases to improve our own business. And the effects can be profound. The amount of time that you can take out of business processes, the improved customer service that you can offer, most of what we've seen has really been augmenting the power of humans, by the way. It hasn't been replacing them. It's been making them go faster and be better informed, and once again, Rebecca, cutting out some of that manual work that we all do in life, right? On the platform side, there's been two big areas that we've been focusing on. So first of all, we've been focusing on using generative AI to make Matillion's data productivity cloud even better at what it's always done. So it's a data productivity platform. It helps Jeremy and his team go faster at bringing together API data to drive business insight. We want to use generative AI to get better at that. And so we've built a copilot into our software, so you can just ask for the pipeline to be built. It will now auto document what your pipeline does, and so other people can figure it out later. And that's part of why Matillion pipelines require 70% less time to maintain over time than other ways of building pipelines.

>> Sorry, what percent?

Matthew Scullion

>> So Matillion customers typically spend 70% less.

>> 70.

Matthew Scullion

>> Yeah, it's huge.

>> I thought I heard 17. 70.

Matthew Scullion

>> So six-zero percent faster to build a pipeline in the first place. Seven-zero, 70% less time maintaining it over time.

>> Thank you.

Matthew Scullion

>> That adds up to a 270% ROI using Matillion build data pipelines historically. We'll get that better over time as well, but 3x is pretty good. The real exciting stuff perhaps is if you think about this thesis that the world's always needed more data but can't get it because we don't have enough data engineers, well, that's true, but more urgent and more pronounced for generative AI. It's more urgent because generative AI is so disruptive. We can completely revolutionize our business with it. So if we don't and our competitors do, that is a real problem. And we have less abilities to do anything about it because there are even fewer engineers in the world that know how to build out GenAI pipelines. So over the last 12 months since we last met, what we've been doing is quickly imbuing the data productivity cloud with the capability of building generative AI pipelines, rag pipelines in the visual composable way for which we've always been famous. So Jeremy's taught through how he can pick from the palette and bring in APIs. Well, now you can pick from the palette and bring in data from the hundreds of sources we support already. You can then chunk that data, vectorize that data into the common vector data sources like Pinecone and Postgres . You can then prompt into LLMs like Arctic, like OpenAI, or the LLM of your choice. You can take that data, and then embellish it into your Snowflake scaler data in the Snowflake warehouse. Or in fact, you can send it onto the destination systems, like Salesforce or Zendesk or some internally built system that you're using. Very quickly, using the skills that we all have in our businesses today, kind of common old garden data engineering skills. We can compose these generative AI pipelines. And we think this is a huge need. We're really proud of the progress we've made on it. Really thrilled this week at Snowflake Summit to be announcing the cortex support and the container services support, which also allow you to bring in those custom large language models if you need to use those as well. That's the direction we're taking it, Dave.

>> Jeremy, for you, this just drives more automation into your business, takes away some of that toil that Rebecca was talking about.

Matthew Scullion

>> The things that really excite me about GenAI, and I'm seeing more of it here, I saw it at Tableau a few years back, is that everybody wants to be an analyst, and we talked about that democratization of data. And now, one of the biggest barriers to that is convincing people that, "Hey, you have to understand this data very well to really truly analyze it." Now, they're able to ask large language models, plain English questions to get stuff out of their data. And then the other thing, I think most engineers have been in the existential dread phase of all this, right? Because data infrastructure can be really bad. And I think one thing that this is going to drive that's going to make everybody's lives better is AI and GenAI is not going to work, and it's only going to be as good as the data you feed it. And it's going to force these CDOs and CIO's and CTOs to get with their people and say, we have to get this right because there's too much upside to be left on the table if we don't. And it's going to help solve that problem for us and get us better data. And then our users are going to be able to get at their data without having to be data experts, right? And it's that democratization making things easy for people that aren't really technically skilled because eventually, hey, data's the most single most valuable commodity on this planet, and it's only going to get more valuable.

Rebecca Knight

>> And you're a living, breathing example of it. Well, Jeremy and Matthew, thank you both so much for coming on theCUBE. Really great conversation.

Matthew Scullion

>> Thank you for having us, Rebecca.

>> 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're watching theCUBE, the leader in enterprise tech news and analysis.