 Everyone, welcome to this CUBE conversation here in Palo Alto, California. I'm John Furrier, cohost of theCUBE. We're in the CUBE studios. Our next guest is Adam Weinstein, who's the CEO of a company called Cursor. So introducing Cursor, it's a hot startup growing in the data analytics space, doing something unique, very innovative around changing the game on data, data catalogs, but more importantly, how data is being used and consumed and also kind of revitalized. So Adam, welcome to the CUBE conversation. Thanks for joining us. No, thanks for having me. I'm excited to be here. You guys are a young startup. You're in a really good wave right now, obviously cloud data, the changing nature of data, taking me to explain what Cursor does, what's the company, what's the focus, how big, you raise money, you get the update. Yeah, yeah, so I'll give you a quick background on me that sort of leads into that, right? So spent most of my career as an analyst, you might say, right? So working with data, living in data, good, the bad, the ugly, right? And spent the last couple of years prior to this at LinkedIn, working in analytics team there. And one of the challenges we had as an organization was finding what was where and who worked on what. So when you had literally 1,000 people across the company that was 10% of the business, touching data on a daily basis, one thing we struggled with was knowing who was working on what, what was where, what was accurate, what was maybe outdated. Data was getting created at insane velocities. Talking earlier, literally we were creating a trillion events a day inside the business. And so as an analytics practitioner, if you will, it became increasingly difficult to get to a quick answer. There was no search to go and say, okay, I want to look for this question as I've been asked before, and if so, where is the data? So there was this new space called data cataloging at the time that seemed interesting, but the cataloging was really only looking at how do we create like a yellow pages of data? Not necessarily how do you put it in the workflow of a person that's then taking that and acting on it and then feeding that insight that they may have created back into that sort of cataloging field, right? So saw an opportunity to create something new that really supported an analyst and really was mindful of how their day-to-day job existed and that was cursor, right? What's the role of the analyst now? Because one of the things that's challenging for the industry was this idea of, let me just go back five years. Data science is the next big thing. There are more open jobs in data science than there are people. But then this also trend came on around humanizing data science and not requiring you to know hardcore C++ or Python or having all this wrangling environments doing all this provisioning of stuff to get started. It was an idea of, okay, can we level up that and also can it make it easier, almost like using Excel? That's been kind of the trend. What's your thought on the current state of the data analyst role? No, so I think that there is a lot of analytics work that maybe five years ago was being done and there was no automation around it and in the next five years it'll get, I wouldn't say automated away, but it'll be heavily automated away. Call it 80% of the workload. But that 20% of data that's really difficult to understand and may not be able to get an answer out of it automatically, that's not, you know, that needs people. And someone that understands the business that's technical enough to go dive into the data. And even though that may not be the 100% that existed before, the amount of effort that's needed to decipher it I think is maybe even greater than it used to be just because the rate of data getting created is so much greater too. Hence the demand for more solutions. Tell us about Cursor. How big are you guys? Who's on the team? What's the product? Is it SaaS? Is it software? Give a quick overview. No, great. So we're a small, we're a seven person team right now. We started the company a little over a year and a half ago. You know, the idea was to get a solution on market that was lightweight enough that someone could come and download it and try it very quickly without having to go through a long enterprise sale cycle. They could get something on that computer, literally stand it up in five minutes, start pointing at a data and having it, you know, identify and help with their day to day job. The team is all engineering minus me, right? So, you know, there's, we have folks from Salesforce where, you know, I came from a company called Exact Target that Salesforce bought, Pandora, Thumbtack, basically we've tried to bring people together that have all, you know, seen company scale and data scale and, you know, bring those insights alongside them. So first generation data scale, the classic, you know, web scale, build it out on open source, grow it, have things break, rebuild it. Yeah, yeah, I mean, we certainly levered some open source, but I think for us right now, it's really just how do we get something that's unique to market as quickly as possible, right? So if there's things that we can use that are out there, that are available, that are, you know, especially if they're standardized, right, we'll make use of them. But other times we'll, you know, we've built quite a bit of stuff on our own. And our solution lives, it can live in the cloud, it can also live on-premise and actually see a lot of customers deploy it in a hybrid manner. So they may have this sort of collaboration layer live in the cloud, but it's pointing at data that's both cloud-based and on-prem. And even though that data make it migrated to the cloud over the next several years, it, you know, a lot of large enterprises are still. So are you guys going to market by selling a product, a freemium? What's the, and is it software they download on-premise? Is it SaaS in the cloud? Talk about the go-to-market and how people engage with the product. No, it's heavily SaaS in the cloud, right? So I think sort of companies that are in a heavily regulated industry that really haven't yet figured out that cloud model, you know, our products SaaS delivered, there is a client that lives on the user's local machine and the reason that exists is just for security purposes because data is still often behind the firewall. So like, you can ask your security guy, hey, poke a hole in the firewall for this company I've never heard of or you can have a tool that lives on their machine that sort of brokers that in a thoughtful way. So you guys are flexible. We're flexible, right? You don't necessarily need that, right? If you deploy it in your own infrastructure, obviously there's no need to then have that client. It can handle things. So why cursor? What are the market drivers for you guys? What's driving your business? Yeah, I think we saw this need and I felt this need very acutely at LinkedIn, which is, you know, analysts are getting, you know, hundreds or thousands of questions as a team on a daily or weekly basis if they're in a large organization. How do you address some meaningful portion of those with automation? So if a question's been asked before and you've got, you know, great solutions like a Tableau or a Looker or a ThoughtSpot or a Power BI, like you've got tons of reporting solutions around the business, but there's no place to go and say, hey, where's the answer to this question? Which one of those is it in? Is it a Salesforce report? Is it a Tableau dashboard? And so you'd ask your friendly analysts who'd be happy to help, but like that's taking them away from doing things that are new. And so I kind of became that switchboard unfortunately and so I saw an opportunity to create a solution that would sort of automate me. And that's really why- Basically index all the questions, kind of see what the frequency was, the behavior of the analytics, kind of packaging it up in the catalog. Yeah, and taking it even a step further, right? Like what are the topics? How do you map topics and understand, okay, there's a fire on aisle seven and that fire happens to be churn and it's Q3 and why is there a fire on churn and how do we dig into the data behind churn and get some automated insights around it. And then, you know, but yes, certainly the step one is being able to direct people to the right place. Once you get beyond that, though, to understanding where our company data is and what the sort of size and shape and characteristics of it are, you can actually take it a step further and really sort of recommend things, which is what we want to do. So the alternative to not having like a data catalog and cursor is to go ask your resident analyst or hope that someone posted something on Slack and then you have to search through Slack. I mean, all kinds of, I mean, really not a viable- No, it's a hodgepodge of solutions, right? So one of the things we saw and it's interesting having been at LinkedIn is that, you know, more and more teams around the organization are hiring analyst talent. They may not call it an analyst, they might call it like a citizen data scientist, they might call it a researcher, they might even call it an engineer, like a data engineer. A lot of overlapping skills and what the real need is, like, someone to be on that team that knows their data inside and out, but yet can help answer, like you said, sort of the ad hoc question that comes up, you know, every day. And so for that, like, you know, they can use cursor to answer 80% of those or as many as possible, right? We think we've got a pretty good opportunity. It's interesting, I do see the same kind of knee-jerk reaction. Hire the data of analyst or whatever. When LinkedIn and other clients that have a similar profile where they have a lot of data, I certainly see that. When they get hired, what's the marching orders? Go jump into the data and figure it out. Is there, I mean, because this is kind of an evolving new position that's growing very, very fast. What are they directed to do? I mean, what's the job responsibility? It's a great question. So I think one of the challenges is how do you onboard people when there is no place to start, right? Like it's, okay, here are the hundred places where we store data, go figure it out, go learn on your own. We had built a little bit of a training and onboarding, I have to call those, really it started as a PowerPoint deck and then it expanded into some code and some additional training. But, you know, there is no solution for that, right? I think internally we had this notion that somewhere between three and six months the person was ramped enough to begin to be productive. It was like how do you measure our life on a person when you hire them, right? And that was LinkedIn where I think we were pretty, you know, we were out here, we have, you know, quite a few nerds, right? Like I think we're pretty good at organizing things relatively speaking. I can't imagine what that's like. And I- Chilly, just write some Python code, spit out some answers. Is that good enough? Like guessing or sink or swim kind of mentality. And then, you know, then get someone else in there. Yeah, and the nuance of the data has gotten just because everyone's mindset is record everything, right? Like it becomes harder and harder to actually get a quick answer. So I'm going to give you an example. Like, you know, looking at data, do you know if something's, you know, test data, if it's, you know, fake data, if it's, you know, if there's something you need to be mindful of, like in e-commerce, how do you account for returns? How do you account for, you know, fraud? How do you account for things that, you know, if you look at the data and say, I just want to add up all my orders and get some total amount of receipts. Like you would think, oh, that's my sales for the day. But then you forget, like there are all these things that if you don't know the data really well, that you miss out on. And so multiply that by, you know, large corporates. What's a phrasing needle in a stack of needles that I'm trying to find like everything in there. I mean, data structures, data cleanliness, these are huge issues. Huge. And, you know, we will address every single one of them, right? I think what cursor wants to sit is in between a lot of best-of-breed solutions, right? So we're not building a new Hadoop. We think Hadoop does a great job of storing data, whether you want to call it a lake or, you know, something beyond a lake, right? Like, you know, there are plenty of data stores in an organization that do a great job of storing data. You know, on the opposite of the spectrum, like in terms of visualizing data or actually generating, you know, insights, there are great BI solutions in the market. But in between those two sort of, you know, ends of the spectrum, there's a lot of work that gets done and that's what we want to live. Adam, talk about the innovation and the tech behind cursor. And then just innovation in general, the way you see it and the team sees it because you're on the front range of a new trend, bleeding edge, cutting edge, whatever you want to call it. Certainly you're pushing the envelope. Yeah. What's the core tech for cursor? Where's the innovation lie? How does it all tie together? Sure, so we have a couple different deployment models but our most common one is we have a cloud layer that enables collaboration. So anytime a company is using our product, you know, metadata, we don't ever look at company data. That's one promise we've made because we want to work in regulated industries. We want to be in places where, you know, there are high security environments but we never push actual data to the cloud but metadata about a company's data. So, you know, what's the name of a column? You know, what's the name of a database? Who's used it? How often have they used it? What dashboard names are using? All those kinds of things get pushed to our cloud. You know, we use a language called Kotlin which is a Java derivative to write most of our backend code. Mostly because a lot of legacy data stores are designed to interoperate with Java. And then, you know, we have a client component that lives on a user's machine and that's what facilitates a lot of the day-to-day work and we do that just for security purposes because, you know, because most data is behind a firewall whether it's cloud-based or not is, you know, it gets independent of that. It's oftentimes not publicly accessible. We can't expect our cloud will be able to get directly to it, right? Whether we're at AWS, GCP or Azure we can work with any of them. You know, we expect that the company's security policies require some sort of, you know, local connectivity. And so that's, you know, that client is actually just a product called Electron that wraps, you know, our React front-end. So very, very common, you know, paradigms, you know, I think we try to pick packages that we think have some staying power because, you know, every time the wind blows there's a new framework that's, you know, the latest and greatest. So that's awesome. Talk about the marketplace and customer interactions you've had, I'll show you guys are a year and a half into this or so. What's the feedback? What are you seeing? What are you learning? What are the key signals from the marketplace that you're seeing that's supporting your company? The direction you're going? Share some anecdotes and data around what you're seeing and hearing. So when we launched the first version of the product it was last May. And what we were trying to do was get something out there in the wild that anybody could try and get value out of without having to go through, because it's sort of a long enterprise sales cycle. So download it, you can use it, you can share it with the guy next to you. Think of like an Evernote or a Google Drive style approach to actually being able to do something. And so that had some great success, right? When we went out with an announcement, we announced we had fun with the company. We got about 1500 users in the first four months just that we're trying it. It was across about four to 500 companies of four-ish, five-ish users of a company. And that will let us get a bunch of feedback, which was great, right? Some of it was, hey, we don't like this. And other was, hey, double down here. And the key thing that we learned was there are sort of three audiences that we're serving, right? One is that traditional analyst, which hopefully that was the case because that's where I came from and that was the goal. There's also two other audiences I didn't expect as much of, one being software engineers. And software engineers that are constantly pulled in to like, you know, like you said, find the needle on the pile of needles. And they don't want that to be their day job, but they do want to like do it once and then share it with the rest of the organization. And they don't have a place to do that today. So there's a great audience of software engineers. And then the last one is actually business leaders that are the ones asking the questions and they want to find a place that they can go to that will answer the majority of them. And so the feedback we've gotten is that there's probably three skins of the product that we're going to have to build. One's for that analyst. The second's a little bit more technical for an engineer. And the third is actually very business friendly, which is just, you know, you don't care about SQL code. You don't want Python code. You don't want any code at all. You just want to know the reports here or if it's not Ask Danny. That's interesting. So the feedback for the marketplace is kind of lays out the workflow stakeholders. Yeah. You know, the analyst has got to do their job and doesn't want to be coded. So they bring the coder in. Coders wants to get pulled into the project. So they're doing their thing. And they certainly want to get back to their coding but get pulled in for business reasons. Then the business wants to search and discover. Kind of all kind of coming together. That seems to be the stakeholders. It's the stakeholders exactly right. I mean, I think it's, it almost lines up probably engineer, analyst, business leader, right? Like in the engineer oftentimes is the one that has to go build a pipeline if that's what's needed, right? And the analyst is the one that consumes from it. And the business leader is the one that looks the report every morning and says, oh, that's bad. And really what you're getting down to is classic software development. Kind of thinking of DevOps and cloud computing, which is you want to automate repetitive tests and you don't want one-offs, right? So engineers want to do one-offs, a constant one-off pipelining. Yep, yep, no, you hit the nail on the head. Like I think the whole notion of like self-service BI or self-service data, like it's aspirational, I think it will be forever, right? Even as you get into AI and yes, automated AI and a certain percentage of problems will always be able to be automated but a certain percentage won't be, right? We'll just get... Well, your point about the reporting is it's only good as the data being reported. So you might feel good that you're looking at a dashboard that's underlying data. It's not good. It sucks and you're like, you're dead in the water. That's a very true thing. Fortunately, we saw that, you know, not just like every company feels that, right? All right, talk about the environment and customer base, okay? As you worked at LinkedIn, which I think is a very acute example because LinkedIn is one of those magical companies where they really hit the data equation really well. Obviously it's like a resume for recruiters and it turned into a social network and then they got a treasure trove of data, all kinds of gesture data. They got great metadata on profiles. Now they got a feed. So again, it's like Facebook, they got all this data. So the unknowns probably came piling in. So it's a great proxy for as enterprises and businesses start thinking about how to think about the tsunami of new kinds of data, not just growth in data, but like, hey, there's all kinds of new data, mobile, touch point, gesture data, all those kinds of stuff's coming together. How should they think about setting up a plan? So if I'm a customer and say, hey, you know what, I got all this data, I got CUBE interview data, I got consumption data, all these new things, and what do I do? How do I create a holistic architecture to take advantage of the different data silos or data sets, but yet not screw up the operations of those data? What's the- Because it can't stop, right? What's your advice on that? Because that seems to be a core problem. It is, and one of the things I think I've come to believe is that companies will get together and they'll spend months or even years coming up with like an architecture of the future, right? And I don't believe that you can come up and sit in a room no matter how many days it takes and come up with something that's going to be all things to all people. Like you're going to basically need solutions that are nimble enough to be installed and get value very, very quickly, even if it's just a small amount of value, and then grow with you over time. So Cursor, that's sort of the way we're set up, right? You can come have a small team, so take a marketing operations team. They work with advertising data they're dealing with. How do you get a lead and convert them into a sale? They can use a product like Cursor, or I think any other good product in the marketplace should be designed this way, where you nibble on it, you get some value, and then you deploy it to other teams once you've learned how to best do that. I think that like big bang approach of like, hey, this is our solution that's going to work for everyone is really tough. So pick an area we can get time to value quicker. And is it like a data lake up in a wall where you just kind of throw some data into one corpus? So we connect to the database. Data doesn't actually live ever within Cursor, right? We may, if you're actually operating on it, say you're an analyst and you're writing some Python and you're writing some SQL, like yes, for the sake of seeing it in the UI, it will temporarily be cash that encrypted there, but we never actually store any company data. We just point to it, and what we've built are these really intelligent connectors that can go mind what's there. So if we're looking at a Tableau instance, we can say, okay, here are all the dashboards there, here are all the code behind those dashboards, here are the table and the data stores those dashboards are hitting. Here's how often they're consumed. Oh, every Monday morning at 9 a.m., 250 people in New York hit this dashboard, and how do we learn from that? And then ultimately make recommendations on it. Like what happens when data underlying a dashboard changes every Monday morning and all of a sudden it doesn't? Should there be a red flag somewhere that we should tell somebody that, hey, there's probably an issue with this. So we're trying to really learn from things that are already there today, as opposed to having you create new things. What's next? What's going on now? How are you going forward? What's the key objective for you guys? Yeah, so I think there's two things, right? As an early stage business, like you can get sort of pulled into this, hey, we want to be a generic solution for everything. What we found is that there's probably a couple industries that are really, they feel this problem very acutely, and some of its financial services actually retail, surprisingly, just given the dispersion of data inside retail. So we've had pretty good success in both of those areas, and I think our next step will be to actually probably formalize some playbooks, if you will, and continue down that path. And then integrations are the next thing, right? Like, we integrate with a bunch of stuff, but we definitely don't integrate with everything, and there's an infinite amount of tools out there, right? So we want to continue to partner with companies that have best-of-breed solutions, work with them to create deep integrations. We're not trying to displace them, we're just trying to compliment them and help drive the traffic to them that's looking for what's in there. So that integration work is really never-ending. Why should a company give up the old way to bring in the new way? What's your think that is by some end? I don't think they're necessarily having to give up the old way. I think it's, there are some things that you're going to naturally be transitioning off of, right? There's always going to be a BI solution that transitions from legacy to new, whatever legacy may be defined as. And as you're doing that, that there's this missing ingredient, I feel like, of how do I track what's where when? You could say that that was sort of solved by data catalogs, but I think the old data catalog is kind of dead, and I think what's really happening is that you need something that works with where you are every day, whether you're an analyst, a business leader, or an engineer, right? It can follow you along, not disrupt you from your day-to-day workflow, and also be intelligent about what's where, and that's sort of what we're trying to build. Well, great to chat. Thanks for coming and spending the time, talking about Cursor, congratulations on the venture. Thanks. Looking forward to seeing that B round coming soon. Yeah, thanks for having me very much. It's coming soon, B round? A round, A round, C round, and yeah, it will definitely be on the near-term horizon. Adam Weinstein, CEO, Cursor, serial entrepreneur here inside theCUBE. Innovating around the data, this is the new model, this is what's going on, this is the new wave that they're riding. I'm John Furrier with theCUBE. Thanks for watching.