Ep #154: The Future of Data Warehousing in the Age of AI ft. Jacqueline Cheong (CEO, Artie)

Taylor Kenerson: [00:00:00] Welcome to the Hyperengage Podcast. We are so happy to have you along our journey. Here, we uncover bits of knowledge from some of the greatest minds in tech. We unearthed the hows, whys, and whats that drive the tech of today. Welcome to the movement. Adil Saleh: [00:00:01] Hey, mornings everybody, this is Adil we are doing this Happening Ed podcast for quite a long, and a lot of times I, I'm, I feel I look myself in the camera and see like it's, I'm growing all with this podcast. And, but, all of these guests that come in and share these stories, in really interesting kind of technology they're building and an AI that they're building on those technologies, making a huge impact. It makes me feels younger and younger with every episode. So today we have, the CEO and co-founder at Artie that is a, a CDC streaming platform. And it's unique in a thousand, million different ways. the way it's, it is making it, a no, a no time, like it's, when it comes to data. When it comes to storage, when it comes to streaming, you think about, hey, loads of time, and it's resources and, all of that. So that's what, rt, doesn't do. They actually optimize your time and they actually, help you, manage your data and infrastructure and all those storage in, in, in no time from, basically capture places to, even production. So thank you very much Jacqueline, for taking out, time and having this conversation with us. Yeah. Yeah. Very happy to be here. Love that. So now, and looking at your category has been, not because it's not so much credit, that's the best part about it. there's not a lot of platform inside it. And, looking at some of the founders and finding you're solving even, having slightly different, experience and I was just thinking that how did all this, all of this come together when. Started, from YC by the way, and, how did all, this inception came through from someone like your background in finance and all, what was, that, that thought process entirely of building, an RT in the initial days, for good or bad? I'm making you remember those stories. yeah. Jacqueline Cheong: [00:01:37] No, it actually started a couple years before yc, but, my co-founder and CTO, his name is Robin. he's also my husband, but he was working and, he's an incredible person. He, is one of the most uniquely positioned people to actually solve this problem because, one, he has a decade of experience with like databases and distributed systems, but he felt the pain of this problem both at, his last two jobs at Zendesk and at Opendoor. so at Zendesk he actually worked on CDC. So there's, two open source frameworks in the world that helped with CDC. He worked on one of them. And then at Opendoor, he actually architected this whole system. They needed to stream data from Postgres into Snowflake, and he architected the whole thing. got, executive approval, got the teams aligned, and then they started building it and 10 months later, it never actually shipped to production. And that was his insight, wow, this, actually could unlock for a, lot for our company, but it's just really hard to build and to build it in a way that's like production, re ready, real time and reliable. and while he was doing this, I was at my job, I was a hedge fund analyst, at Valley asme, and I was at Bank of America before, but spent, six plus years studying how great enterprise software businesses are built. and at the time I was actually investing, looking to invest into Snowflake, I was also looking at, Databricks. This was many years ago. and I, could understand what, what's bad, like the benefits of streaming versus batch based off of learning about those companies. and, I got really excited by this idea. and we were talking to a couple of, engineering managers at, tech companies, and it happened to be that they all, were they, all faced this kind of like problem and they all were contemplating Wow, to get this in, It is a pretty big project. We, we're, let's design like how to actually build this in-house and that was the insight and that, that's where we were like, okay, this is very interesting. It seems like every enterprise has to deal with this problem and there's no easy button. What if we built the easy button? and that's how we got started and got into yc. Adil Saleh: [00:03:51] Interesting. And, a lot of this has been of course, been validated in the beginning, with advisee, like hypothesis might be different when you go in Brown and Ed with actual consumers. How was that experience when you started heading over to, these enterprise clients? What kind of challenges you might have faced in the beginning? Jacqueline Cheong: [00:04:09] yeah. Like how the vision kind of changed, how the product changed as we actually went into market. Adil Saleh: [00:04:16] Yes, absolutely. Like during the validation process, like you need to validate it and see like how, it is helping, there are some nuances, some, roadblocks, some friction points that came through in the beginning. Jacqueline Cheong: [00:04:26] Yeah, So luckily, we were solving a problem that Robin had. So in the beginning, really this first, six plus months, it was really him building the exact solution that he wanted that would've worked for. Open Door and Zendesk. so there, there wasn't validation in the sense that you talk to your customers, potential customers and you build the product with them. We, our MVP was built exactly based off of Robin's own experience, and of course that has changed over time. As we onboard more customers, we realized there are. Funny enough, we realized there are way more complexities than we thought originally. Way more edge cases. And then once you, once we started onboarding customers with a lot of data, it was complexities with scale as well, or complexities with how their database is designed. just to give you one example, a lot of enterprises have like single tenant database designs. Where you're left with, instead of a normal situation where you sink a couple hundred tables, the design of the database means that you're sinking 10,000, 20,000, 30,000 tables. And that becomes like very, complex from a streaming architecture perspective. And so those are the complexities and challenges that we face over time that we solve for now. Adil Saleh: [00:05:48] very interesting. And I was, this brought me to, thi this next question, there are like, infrastructure changes like from industry to industry, segment to segment. And then within those segments there are different users. 'cause let's say 10 different customers in one segment might be using Artie in a slightly different way, and they're perceiving value in a different way. So how do you guys like, I know that it's not. Or early, like it's been like two years, almost like three years. I'm not talking about like product market fit, but how you guys are trying to have a fine balance between what to standardize, what do you know, white glove and have the architecture like solution engineer or implementation managers and to especially post-sales. Jacqueline Cheong: [00:06:25] Yeah, and I think in the beginning it's like you really don't want to do any of that stuff. It's really finding out what are all the common pieces. So there isn't a lot of variability between, syncing data from a, transactional BA database into a data warehouse like Snowflake. And once you built it, once, the pattern is very, similar across all your customers. Even in the case of, The very sharded databases or the single schema, like single tenant database designs, that's actually very, common. Like the first time you see it, it's okay, you have to build a bunch of stuff. it's custom. but once you build it within a couple months, you have four or five, six customers using that same functionality. And I think that's the way that. At least our philosophy is like that's the way we're going to build. If there's something that's super, custom where we have conviction that nobody, else is gonna ask us that. The answer is you just don't build it because it's not, repeatable, it's not scalable. Adil Saleh: [00:07:28] Yes. And it doesn't align with your product vision. You gotta make sure that you do something that, for at least, addressable segment of customers, even though if they're segmented, that's fine, but you need to have like good amounts of, customers. So are you also thinking about, I know that it's, you started off, up market lot, these startups that go like with SMBs and mid-market and they try to go up market, but what, do you see from a Business standpoint, like commercial standpoint, thinking about the segmentation of the customers, are you, like now tools are going to come, if not, building, a open search model on top of these CDC platforms and you're gonna have more and more competition over time and you might be thinking about it. So how you target this and B two, slightly mid-market, segment of, of your customer base. Jacqueline Cheong: [00:08:08] Yeah, so our go to market motion has evolved over time. it went full circle. So when my co-founder who was working at Zens and Opendoor, obviously those are enterprises processing high volumes of data. When we went through yc that first year, it was. Let's validate that someone will pay money for this. And so we were selling to everybody, anybody who, needed the product, and we observed how they used it, how happy they were with us. expansion of usage is really, important. And over time we've realized that, really what is the segment that feels the most pain? It actually is these enterprises because when you have high volumes of data. The complexity and the sophistication that comes with that requires a different set of challenges to be solved versus a smaller company with less data, it's a very, different problem. And so when we solve that problem for enterprises, the unlock is just so much bigger. And so that's why we've actually started focusing. Entirely there. And, there are obviously edge cases in terms of, there are segments of the industry where it can be a mid-market or even SMB, but you're processing the same amount of data as like some of the enterprises that we, serve. That is ICP. For us, it's really about data volume and the complexity of your, needs. Adil Saleh: [00:09:36] interesting. Because, it doesn't matter whether it's enterprise or SMB, it has to be, that level of threshold of data and issues to make it, to make, this product more sense for them. So now thinking about, you talk about streaming CDC, what kind of, industries that you can account, the use case is border up and this is for my education a lot. Some of, them I've found on the internet. So if, you are to go wide, especially as you mentioned, that there's no segment, like in terms of enter enterprise at SMB, it's just about. And, use cases and all of those. So what kind of industries that you can count as, for our audiences like, pharmaceutical companies, or these sort manufacturing companies, or let's say just your media houses. You talk about Warden Brothers. You talk about what kind of, industries, Jacqueline Cheong: [00:10:20] Yeah, funny enough, this is not an industry specific problem. there are like industries are powering, roughly the same use cases with, realtime streaming into the warehouse, but it's not an industry specific problem. So just, to give you like, just to back out a bit, the reason people need this is, over the last. Two to three, the number of use cases and workloads that is built on top of the data warehouse has exploded. So like, it used to be like really static BI dashboards that like executives would look at once a month, and then they started layer on like marketing and product analytics. And then they started layer on, even like today there's like personalization, experimentation, customer facing data products. And so as the. The number of use cases, but also like the sophistication of it has increased, the demand for the infrastructure to power these systems. reliability, real time, flexibility has also increased with it. And so that's the dynamic we're seeing. So enterprises with more and more parts of their business being dependent on their warehouse, like sales teams, marketing teams, customer support, finance, the more you have. the more data needs to be moved in. And, that's why it's every industry is going through this pattern. Adil Saleh: [00:11:49] Yes. So this calls me like, I just, spoke to our customer last week and they were like, they were making sensors for labs. they, the, sensors were transmitting data onto a BI dashboard where customer success team was actually using our product and they were actually getting those data, capturing data from that BI tool to be able to measure success for those products that were hardware. to me the success, and essentially it's about it, you a hundred percent. that it's, more about, now every company in, the industry that are, least, tech enabled, they are trying to be data driven and they're trying to more rely on their warehouses and stack more tools and more integrations and more, I would say, gateways, Integr integrating segments, with Segment is a big, tool that pipes data into it, that. And they're integrating with directly there to their warehouses to send it to these tools that will enable marketing, sales, support problem, all of this. So now, I know that, you guys, got into YC that, you guys are still funding from a, customer success standpoint, what makes you excited in terms of ai? I know that a lot of these tools, teams are building their own agents, custom GPS and all that. A lot of that has been, transformed and automated and optimized the bandwidth doing more with less and not, why didn't you go, about, doing some sort of a specialized agent flows within for, your own teams? If not, or. for your customers, a lot of these companies, they're doing like building agents, co-pilots internally. having, to mention that, you also mentioned that you have, like in one segment, you have definitive patterns, that you guys can follow and you don't have to do anything customer, a lot of custom if you have to, you just get rid of that, that kind of, I mean you to more align with your product region. So how do you see a gentech workflows all of these evolving, in the recent times? Jacqueline Cheong: [00:13:25] Yeah, I guess when we were going through, I see there were a lot of these, like AI agents, AI companies that were building around us. the reason we decided to, to solve this problem is really from a personal standpoint, really saying, thinking about what did I wish I had or what did my co-founder wish he had when he was back at Sunde and Open Door? And it wasn't, at the time it was like this was the most painful thing. So we were like, let's just solve that problem and forget about everything else. That's happening around us. That being said, though, I think AI is pretty exciting in the sense that that, that whole, use case being built on top of the warehouse, that I was talking about. AI is the next wave, right? AI feeds off of data. Your data warehouse is your centralized source of truth. So there are a lot of these like AI initiatives being built. On top of the warehouse. And with that, it's just yet another use case That's, bringing, up the sophistication and making real time even more important than it ever was before. I would argue as these enterprises actually figure out what these AI initiatives will look like and push 'em into production, that is going to make the sophistication of these like data ingestion solutions even more important in the future. Adil Saleh: [00:14:40] That's a green light for you. AI evolving people, building agents for their own teams, for the customers. okay. So now thinking about, I was looking at, trying to find the right, meaning for the pricing and also how challenging was it for you guys, build the right price points for, for all of your plans. And of course, there has to be some expansion and to, different transiting into different plans based on the usage and, size and all of those. you internally a lot of teams do. So how challenging was it, price, what kind of GTM frameworks or g TM measures you taken while finalizing those prices? This is the question came from my team. Jacqueline Cheong: [00:15:13] Yeah, I think it depends on, you have to think from the customer perspective, like what do they value and how do they want to pay for value? And so we have actually two different pricing models. There's one for the lower end of the market where you can flex up and down. They really value being able to pay more when they use more and pay less when they use less. Obviously with. Automatic discounts. So like it still scales with them. if they increase their volume by 10 x, their costs don't 10 x. So that was important for the lower end of the market. And then on the upper end of the market, what they really care about is not how much they're paying, but predictability like CFOs need to be able to project what they're going to spend over the next one, two, or three years. And. Creating a pricing model that is built on predictability with the same caveat that it has to scale with your business. Like it has to be, it has to continue to be profitable to serve that segment as they grow. So that's how we've designed it. So there's two pricing plans for two very different, like personas and values. And we flex on both depending on what the customers want. But it was, it was a evolving process. It required. Talking to your customers, talking to other founders with usage based businesses that are on the infrastructure side. Like how are they thinking about it? So it, actually requires quite a bit of thought. Adil Saleh: [00:16:44] Yeah. Usage based becomes tricky. Starting off it becomes really, tricky. I've seen products changing every three months to pricing and all that because they're still figuring out. Okay, so now another question, from the team. there, there's, I know that there are tooling that is coming, that is more modular and there's some modern data stacks built on modern data stacks and they're shipping really fast, maybe, getting lots of funds as well, capital as well. So how, what kind of plans you have in terms of marketing position in a longer term view? Not in five, 10 years, but like two to three years. Jacqueline Cheong: [00:17:15] Yeah, I think, I think this one's really important. When we think about where, database replication is going, there's that core market that we are entering with, and we entered this market because it, was a no brainer in the sense that like every company needs to solve this problem to unlock any other types of workloads other than like transactional workloads. That's number one. So it was, a very, a very important like beachhead for us. And then it's thinking about, in the next two to three years, like what are the other, like real time streaming use cases that enterprises care a lot about. It may not touch just the database, for example. it can be, hey, I wanna stream data into Elasticsearch to power product search. And because we are good at moving data in real time. With like data consistency, reliability, those primitives still matter a lot when you're streaming for like product search use cases. So expanding into those sort of use cases in that sense, make a lot of it, it's like a very clear adjacent step for us to make. Adil Saleh: [00:18:28] I think, yeah, you guys have, it's, you learn a lot with customers to be very honest, especially enterprise customers. They teach you a lot. Like how do you guys manage with the sales cycle? I know that it's a big red tape state in the US with this AI and all, it's more about relationship getting coffee table talks and all those. So yeah. How do you see, will you, would you be guys, be better off like having some sort of a. Less, white glove sales motion or, maybe some have a digital touch into it or maybe, are you guys still feeling that it's hard to get let's say five or six enterprise customers every year? still. So how is that motion going on? Jacqueline Cheong: [00:19:03] five or six? No. I think it'll be hard to get 50 or 60. A year. Year, and I think you're right, like the enterprise sales cycle is much longer than SMB and mid-market. There's a lot of relationship building. Like you might meet someone at a conference, you grab a coffee and then, six months later you meet them at another conference or at a talk. And then one year later, they read a blog and they're having this problem. And then they come in and then you start a conversation, and then from there, maybe six to nine months down the line, they're like, oh, okay, now I'm ready to do a POC. And then you do it and you close. So like depending on you, when you measure it, Obviously when they're ready to go and they're ready to POC and then things like move very quickly. But the kind of nurture cycle takes a, really long time. But I think it's like you get, you, pick your battles because the beauty of what the enterprises is once you get in, They give us the most insights. they're like, oh, like we have this other use case. Could you do this? Could you do that? And that's how we're like thinking about our two to three year roadmap. I wonder how many enterprises are you gonna ask us the same questions? When we hear it enough? It's okay, this is like common across. These are five customers across three different industries that are asking for the exact same thing. Now we have conviction to actually build it and go to market and market it. So like I think there's a lot of like pros and cons that you just have to play with. Like with SMV, there are pros. The sales cycle is much lower, but they have their concept they have to deal with. So I think it's, you just go where the customer pain is and then you deal with the. You deal with the cons, however you have to, make it happen. Absolutely. Yeah. You know that you need to solve the subject. Adil Saleh: [00:20:45] It's that simple. You need to solve the problem. And, it may ever take, play the hard, knocks. Perfect. So now thinking about like your, like what makes you, guys excited for this year product-wise? what you guys are planning on the roadmap? I saw this. is it public? the roadmap? No, it's not public. Yeah, I was about to, look at it. So, what, will you like, I know whatever you can share, you can, what makes you excited product wise in terms of, some of the top features that you're building? Any initiative, you guys taking product wise? Jacqueline Cheong: [00:21:17] Yeah. Yeah. I think it's, a combination of two things over the next year that I'm really excited about. One is, broadening our base of databases that we support in the sense that, of going deeper into, databases that are really complex and more enterprise. Focus like a, our next version of our SQL Server database connector, next version of Oracle. Next, our new IBM, database connector that we, don't have yet today. but then the secondarily, which is actually the more exciting part is Building out those, like building really, in depth the sophisticated features to, for, these enterprises. So the, one example I gave about like the singleton of databases and needing to fan in to be able to stream more efficiently to their destination. You have to do that per source. And so it's okay, we have it for a handful of databases, that we support. Now let's bring it on so that all of the databases that we support have that capability. So it's bringing on more like sophistication in every single pipeline that we have. Adil Saleh: [00:22:26] Yes, Jacqueline, this makes it even more interesting, thinking about having a longer term view on your customer use 'cause of that fitting to and having to forecast and like what they're gonna be needing tomorrow. And how I can align, our product, roadmap, for the future of the problems that they're facing. Because when you sign up an enterprise, you think of serving a minimum of five years. that's great. So now I'm all about Jacquelin, being, about, thinking about women and leadership and so much, excited to, my wife, she's also, in tech. She's, leading a team like quite junior, the, the new guys. But, I'm so big about it. We have like around 30 plus, women leadership here onto the podcast out of these 150 episodes. And, I really appreciate that you've been, inspiring a lot of, other women as well. So what is that one, number one message you have for them? Jacqueline Cheong: [00:23:12] Ooh, I, think this happens to men too, to be fair. But I think, imposter syndrome is very real. especially, when you're doing new things. Again and again. And I think being a founder, by definition, the vast majority of the things that you're tackling are probably new to you or very foreign concepts. And I think the message is everyone feels imposter syndrome. I think as you're doing every, everybody that's doing something new probably feels imposter syndrome. So knowing that the whole world feels this and you're not unique and, I think, that helps a lot, just knowing that it's not weird. It's, what everybody feels. But then also knowing that, you're probably in a really good position as long as you have a good problem solving framework. You can knock out anything that's like in your way. Adil Saleh: [00:24:03] Yes. Amazing. And any one book or inspirational leader that you have, that you've always admired, that you always followed, it could be anybody in your, in your family, in your circle, friends, anybody that is celebrity, sportsman. Any, anybody. And one book that it inspired you. Okay. You can take all the time. let me think about this for a moment. a lot of these folks have said, like, when you say one, that makes it complicated. So you gotta come up with one, one book and one inspiration figure, inspirational figure. Jacqueline Cheong: [00:24:43] I'll talk about the inspirational figure. I am, I've actually called him out before, but I find my coach, Roz, very. Inspirational. I met him many, years ago, but only started working with him two years ago, and he's been very helpful and instrumental in guiding me through the founder journey as I've grown. Really turning me into. From a, like every founder needs to do this, but you go from a very strong IC to a manager to a Eventually A CEO, like a true CEO and I think He's really inspired me with the way that he's been able to allow me to think about my situation or working with people in a very, different way. So he was very helpful in terms of like my own personal growth, what means in terms of wow, amazing. In terms of a book I'm really struggling to find, it could be a movie, it could be a short film, it could be any story, anything. this has nothing to do with, this has nothing to do with entrepreneurship, but, a recent book that I read that I really, actually no, this one is really good. The book is, the score takes care of itself. I actually have the score takes care of itself. Yes. Yes. it's Bill Walsh's book. he used to coach the, San Francisco 49 ERs. he was a very, famous coach and his, he distilled his philosophy on. How to lead into this book. Okay. But he's leading a sports team, but obviously a lot of his skills were transferable into business. That was a really good book. I actually have a quote of his, taped on my mirror. Okay. What's that quote? it's actually the sport takes care of itself. Adil Saleh: [00:26:33] Score takes care of itself. Wonderful. Love that, Jacqueline. It was, such an amazing conversation with you. A lot of that was, some more education for me in terms of your product market positioning, all of that. And, you did an amazing job, educating me in audiences. And of course, this is not just, it's just a beginning and we'll keep on sharing more stories associated with you. thank you very much for your time and all this knowledge that you shared. The energy was infectious. Jacqueline Cheong: [00:26:57] Thank you. Yeah. Very happy to have, shared my story and shared a little bit about art. Adil Saleh: [00:27:03] Amazing. Take care. Have a good rest of your day. Thank you so very much for staying with us on the episode. Please share your feedback at adil@hyperengage.io. We definitely need it. We will see you next time and another guest on the stage with some concrete tips on how to operate better as a customer success leader and how you can empower engagements with some building, some meaningful relationships. We qualify people for the episode just to make sure we bring the value to the listeners. Do reach us out if you want to refer any CS leader. Until next time, goodbye and have a good rest of your day.