theCUBE + NYSE Wired: Physical AI & Robotics Leaders | Kirk Dunn, Kurrent

Clips
More from theCUBE + NYSE Wired: Physical AI & Robotics Leaders

Kirk Dunn

CEO

Kurrent

play_circle_outline Introduction to theCUBE and its history, focusing on AI and robotics innovations.

play_circle_outline Navigating the Shift: From Structured to Unstructured Data and Centralized to Distributed Computing in Modern Data Management

play_circle_outline Kirk Dunn on Kurrent's Innovative Append-Only Log-Based Database: Revolutionizing Event Recording in Business Operations

play_circle_outline Industry insights on evolving operational states versus analytic states in data management.

Info
Transcript

Kirk Dunn, Kurrent

Kirk Dunn

CEO Kurrent

Kirk Dunn, chief executive officer of Kurrent Inc., joins theCUBE’s John Furrier and Dave Vellante during theCUBE + NYSE Wired: Robotics & AI Infrastructure Leaders 2025 event to explore the evolution of data infrastructure in support of AI and robotics. Drawing on decades of experience, Dunn discusses Kurrent’s focus on event-driven architecture and stream processing.

With roots in Cloudera and the Hadoop era, Dunn outlines how Kurrent captures data as a continuous series of events to power real-time analytics and intelligent automation. The company... Read more

explore Keep Exploring

What recent events are being discussed in relation to theCUBE and the Wired community? add

What are the key developments and challenges in the evolution of data platforms? add

What is Kurrent and how does it address the challenges of capturing business events in data modeling? add

What developments and focuses have occurred in the cloud product strategy over the last year and a half in relation to financial services? add

bolt Powered by CUBE AI

Kirk Dunn, Kurrent

search

>> Welcome back everyone to theCUBE here in Palo Alto. I'm John Furrier, your host of theCUBE with Dave Vellante. 16 years of theCUBE we've been at it, watching all the big waves. This is our featured series on robotics and AI leaders from physical AI all the way down to the data layer. We have investors. We've got entrepreneurs coming through. It's a great day. Of course, tonight big party with the NYSE and theCUBE, which has formed the Wired community, the NYSE Wired. Been a great friend of theCUBE going back to actually our early days. Kirk Dunn is now the CEO of Kurrent. He's leading an innovative company around data and in perfect position for this wave. Kirk, great to see you. And we go way back, the old days of Cloudera when we started theCUBE, you were in charge of that organization.

Kirk Dunn

>> Yep.

>> You look great and you've got a new venture.

Kirk Dunn

>> Well, the good news is we're going to agree we all look like we have an aged a day, right?

>> Yes, I'm with you on that one.

Kirk Dunn

>> Yeah, exactly.

>> I had black hair. .

Kirk Dunn

>> Well, first of all, I would just say congratulations to you guys for... And I always claim you guys are the innovators of the podcast. You're the ones that did it before anybody believed that it was actually a media outlet. And the fact that you guys have been doing it not just for 16 years but successfully, man, hats off to you.

Dave Vellante

>> Appreciate it.

>> Appreciate it.

Kirk Dunn

>> Exceptional.

>> Kirk, and same to, you've got a great career. I want to ask you, Dave and I always talk about this as being in the arena for so long. You become kind of a historian while the current events are happening. This is kind of a generational wave here. You've got chip software, geography challenges that are transforming the role of data, which is your wheelhouse is the central ingredient right now where all the innovation, all the infrastructure players are innovating to manage the scale for large scale data. Of course, token demand is feeding in the applications and the agents and then the applications themselves are going to have databases. So every app's going to still need a database. Now there's different kinds of database. So the whole data world is now right, front and center in the value proposition. This was the promise of big data 15 years ago when you pioneered the Hadoop movement, which set the table and became kind of where the flowers grew now and all that's happening. So now you're back where if we can go back to 15 years and wave the magic wand, this is what it would look like. Right now, the big data revolution is agents, it's horizontal data. What's your reflection-

Dave Vellante

>> Real time. Yeah....

>> on this?

Kirk Dunn

>> Well, it's interesting you say that because there's been a lot of moves in the data platform area for years. I mean, it started with the operational database, right? Rights, reads. That was it. And now look what we've done since Oracle, In-formix, DB2, and now we have companies like Snowflake, Databricks, and now we're into the AI world, which is using data in different ways. There's two major movements, and one of the reasons I came back to go run a company was the movement from centralized computing to distributed computing, everybody knows that. There's nothing interesting about that other than when you start to distribute compute the way it's distributed, you now put other pressure that you didn't have in monolithic systems. Now you're distributing systems across a portfolio. And guess what you're doing? Exactly what you say. You're distributing data with those systems. So now what do you need to do? You got to bring it back together again because you have to correlate data. The thing that we talked a lot about at Cloudera years ago was this movement from structured to basically everything was structured data. And all of a sudden, guess what showed up, unstructured. And Hadoop was about schema on read, not schema on write. Imagine that you're creating the structure when you're reading it. Well, that was the beginning of all sorts of things that then evolved since then, kind of motherhood and apple pie, and it exploded what we saw as the data warehousing space into the data lake, into the lake house. And of course, as we say, Databricks, Snowflake are having amazing success. But what has that whole platform done is that one of the things you mentioned, Dave, is really right. It is taken how do you take origin data and deliver results in real time. And back in the day, real time was close the books in a month, close the books in a quarter.

Dave Vellante

>> Before you lose the customer.

Kirk Dunn

>> Before you lose the customer. And now it's while you're swiping the credit card. If that transaction doesn't go through, you're going to pull out the other credit card and you're going to lose that transaction forever. And so that's point number one. Point number two is most of what we've done thus far has been human generated. You have a business owner that has a data team, that has an analytic team, and they all look to tell the business owner what happened. We are right there where it is going to be non-human oriented. It'll be agents that'll be making decisions based on listening for streams of events and taking action with no humans being involved. And of course, what an agent says is I already know the protocols of the data that's coming to me if it's time series or graph or relational. Don't do any data integration for me. When I see it, I know what it is. So here's what I want. Give me native data at Origin and give it to me fast because I'm capable of making decisions based on what my agentic program does to act. And then the thing that makes this like turbocharged is agents will then stimulate other events that will stimulate other events and other agents and other agents. And so this move from monolithic application environment to distributed microservices to an explosion of distributed agentic workflows, we're barely seeing it right now.

Dave Vellante

>> Yeah. And those microservices are large. I mean, basically hard-coded.

Kirk Dunn

>> Yes.

Dave Vellante

>> Right. So they can't reform workflows on the fly.

Kirk Dunn

>> Precisely.

Dave Vellante

>> That's a huge change that's coming.

Kirk Dunn

>> And you guys talked about when you talk about 2D to 4D. So typically when you think of a workflow, you think of a graph kind of oriented data shape. If you think about the operational database, and this has come out recently that one of the really only advancements in operational data has been the shape of the data. Oh, okay, we need time series. Oh, we need columnar. Oh, we need graph. Oh, we need vector. So the shape of that data was changed to deliver a result downstream for a particular use case or application. Well, how about you have an agent that can read any shape of data because in the workflow you're talking about, "Oh, it's a graph workflow, so we're restricted to that." Oh, it's time series. We're restricted to that." How about a workflow that says, "I can take time series graph, columnar, relational. I'll make the decisions and return the analytic result and I'll do it in real time."

>> Kirk, talk about the company you're leading. You ended up briefing with us and it's fascinating the position you're in.

Kirk Dunn

>> Yeah.

>> Kurrent, what's it about? What is the core value proposition? How do you see the market you're going after and what's the value proposition?

Kirk Dunn

>> Yeah. So Kurrent is an OLT database that deals with streams of events. This goes way back to the beginning. I mean, this is why I got really interested and turned on by it. A business occurs as a series of structured events, of ordered events, every business, a yogurt shop, Lululemon, whatever, a stock brokerage firm. And so those events happen in a series of order. Well, what happens in the application environment is a business person says to a data architect, "Here's my business. I want you to build a data platform that captures my business." So great, data model, business model. The first area of dissonance in the world is the dissonance between the business owner deriving the model and the data model. Why do I say that? Because databases capture events off of an application as states. So a bunch of things occur, they write state. A bunch of other things occur, they write state. But a bunch of the things that occur don't get captured. And so what Kurrent is Kurrent is an append-only log-based database where we capture every single event that occurs. Harrison Chase, CEO of LangChain the other day was on a podcast and he was talking about what agents will do, and this is really important to hear it this way. They will listen for events. They're not listening for data. They're listening for the events that derive the data. And so the business model is how the business runs. The data model is the architect's interpretation of how the business runs. And then the state that gets written to the database is again another step removed. Our argument is if a business happens as a series of ordered events, why not record it as a series of ordered events? Keep it all. Here's the kind of the key point. I always ask people, "Why did change data capture occur? Why did it even happen?" $3 billion a year business growing at about 20% a year. It happened for this reason, an application environment, right state. And then somebody goes, "Hey, Dave, how did that state get to be that state?" And they go, "I don't really know." So let's invent CDC. And what we will do is we will capture those state changes. Here's what's interesting about that, those state changes are still puzzle pieces. It's not a puzzle yet. And so what happens is then some data engineer takes current state, takes CDC streams, runs it downstream, and puts the puzzle back together and says, "Oh, it's a mountain scene. Oh, great. Well, how about if you didn't have to do any of that? How about if you recorded everything exactly the way it happened?" It's like the difference between in a basketball game looking at the stat sheet versus watching replay the replay of the game. Oh, the Warriors won and Draymond Green had 20 rebounds and blah, blah, blah, blah, but it doesn't tell you the story. Well, you can actually literally replay from the events the entire story of the game, so you know exactly what happened.

>> I do like YouTube's key plays feature actually, by the way.

Kirk Dunn

>> Exactly. Yeah. It's a good one.

Dave Vellante

>> That sets up digital representation of the enterprise-

Kirk Dunn

>> Totally....

Dave Vellante

>> and everything that who did what where-

Kirk Dunn

>> Totally....

Dave Vellante

>> and the lineage of all that.

Kirk Dunn

>> Yeah. And there's another startup company, Chad, who is a founder of a company called Gable. And Gable basically is around to do this kind of data governance data lineage thing. And the argument is from origin of data, from let's say it's time series or relational, whatever, it doesn't matter. From the day it gets originated, all the data transformations it goes through, all the different organizations it traverses, all the different Parquet files it ends up in, and then the BI tool that represents it back to the business. Do you know how many steps we call that lost in translation? And what Gable is all about is we're going to give you the historical lineage of that data so you actually know where it derived from and where it's headed. I'll give you one other example of one of our big customers. Interesting. Not a very sexy company, large automobile parts manufacturing company in Europe. They have all this application data on one side and they have Databricks on the other side. And what they realized was in this translation era, they were taking all this data and they were trying to recompose it to tell the story of the business. And they realized the problem is when we get down to the Databricks layer, we have no way to find our way home because it's been stepped on and changed so much. And so what they did was they stuck current in as what they call the event operating system. Because again, if a business is a series of events, then capture the data as those events and you'll derive whatever state it is.

>> So who's the buyer? Who's the principal persona that you want to target?

Kirk Dunn

>> So we have two that we go after. So because we're a database, one is we go after application development environments and we have people that have... It was an open source project when it started. And we have customers who are that literally say... And I always say there's more than one way to do anything. So gray hair proves that there's no one way, but they literally tell me they will never, ever build another application any other way than this way because they capture the events, they capture the entire history of their business and they can always go back. So app dev is one area we, go very natural, but you guys know the database world can be challenging if they're used to Postgres or SQL Server. The other area is this data integration layer where we know that when you've originated data and you're not moving it, then data is stale. Why originate it if you don't move it? And so the whole streaming world has come up, which is how do you move data in real time. And so the data engineer where we're taking data in off the operational tier, organizing it in topics, and then delivering it to downstream applications, agents, BI tools, whatever you want. And again, the key is you can find your way home. We know your source of data, and so we'll deliver those read models to others. So app dev as well as data engineering integration play.

>> Well, Kirk, we're super excited you came on and super glad that we could connect. I know you got a lot going on. We're going to see you at the event tonight.

Kirk Dunn

>> Yup.

>> What are you guys doing right now? Give us status of the company, levels of people. What's your focus? Are you funding?

Kirk Dunn

>> Yup.

>> How's things going? How's the business model? Give a quick plug on what's happening.

Kirk Dunn

>> Business is good. I like to tell us because we're a database that streams, everybody's attaching AI on to it. Although I will say two weeks ago we just released our MCP server.

Dave Vellante

>> Okay.

Kirk Dunn

>> So you'll be able to use natural language to maybe to say, "Build me an application that does this, that and the other," and it'll build it for you. So that's very exciting. We have a number of really significant product announcements coming in the next month. So I'd love to come back and share those.

Dave Vellante

>> Terrific.

>> Definitely.

Kirk Dunn

>> A lot of those are interesting things in this ability to curate data closer to the operational state. So you don't have to wait until it gets to the analytic state to inquire that you could to query the data, you might say, if I'll give you a little bit of a forewarning on what's coming. We're keeping the company lean. We've got about 140 customers. We have a 95% renewal rate. The product just works and people like it. And so this year we've spent the last year and a half when I joined to kind of build a very advanced cloud product. We're about 50% on-prem, 50% cloud. 50% of our business is financial services. So if you think about financial services is a ledger business, well, events are a ledger business, so that's why it fits. And so this year is about getting those products fully into the market and then investing in the go-to market and expanding that customer base.

>> Well, you definitely have been an AI infrastructure leader and continue to be. Obviously now, databases joins network compute and storage as critical infrastructure.

Kirk Dunn

>> Huge.

>> And databases are tied to every application. You got graph, you got all kinds of databases. So again, the world has spun back in the front door of your world.

Kirk Dunn

>> And I say this too because we've been around long enough to know what Oracle did was they ended up with Exadata, a data warehouse product, an analytics product. What Snowflake and Databricks are doing as analytics companies, they're coming back to the operational state. And so the right answer is you actually have to have a portfolio of products. But here's what I would say, is it more likely that agents will be fed from the analytic tier or they'll be fed from the source of data?

Dave Vellante

>> The latter, for sure.

Kirk Dunn

>> Yeah, totally.

>> Origination.

Kirk Dunn

>> And events too. Track everything.

Dave Vellante

>> It's a true source of truth. It's through where the process logic lives.

Kirk Dunn

>> Exactly. Yeah.

Dave Vellante

>> The 4D map.

Kirk Dunn

>> It's a 4D map. We love what you guys are doing in this area and, man, we're small but mighty, but we are running with the cubes, 2D versus 4D. So we're big fans of George's work and your work there.

Dave Vellante

>> Thank you.

>> Yeah, it's great. , of course, we've been tracking the programming models here and data. The layers change, the game is still the same. Innovation, real-time, GenAI has new requirements, new architectures, and of course that's changing all aspects of the stack. theCUBE's bring you all the action here in our studios at Palo. I'm John Furrier with Dave Vellante. Thanks for watching.