He quit Google, launched Rubrik—then grew to $1B ARR & a $16B market cap. | Soham Mazumdar, Co-Founder Rubrik & Wisdom AI

Soham Mazumdar (00:00:00):
I think one of the things to realize is early customers, they feel great about making you successful. So they can take a lot of pain, because early products are terrible. When you pitch to VCs, you have industry reports. And, oh look, this trend happening here, there, all of that is nice. But there is only one thing. Which is rubber meeting the road, which is the customer feedback. Otherwise, there is no PMF.

Pablo Srugo (00:00:22):
How fast do you hit a million ARR?

Soham Mazumdar (00:00:25):
Few months? Not few months, I would say. Like, two-ish quarters I would say.

Previous Guests (00:00:28):
That's product market fit. Product market fit. Product market fit. I called it the product market fit question. Product market fit. Product market fit. Product market fit. Product market fit. I mean the name of the show is product market fit.

Pablo Srugo (00:00:41):
Do you think the product market fit show, has product market fit? Because if you do, then there's something you just have to do. You have to take out your phone. You have to leave the show five stars. It lets us reach more founders and it lets us get better guests, thank you. Soham welcome to the show, man.

Soham Mazumdar (00:00:56):
Thanks for having me, Pablo.

Pablo Srugo (00:00:57):
Dude, so you're one of the founders of Rubrik. Quickly becoming a legendary company. One of the few, successful IPOs of the last few years, and now a $16 billion company, up more than 2x since the web public last year. Which is a pretty epic journey. We had another one of the founders before, Arvind, who's now at Glean on the show. But we'd love to hear the story from your end. Maybe take us back, to before 2014. What was going on and what led to the start of Rubrik?

Soham Mazumdar (00:01:33):
Yeah. So the, I mean, the story of Rubrik. It's a little, it's funny, when my CEO Bipul Sinha, first pitched the idea of Rubrik. I was a little bit mystified. He was talking about how you go about bringing some cool technology to enterprise customers. Who ran their own data center and my background, I had spent a lot of my time with Google. I had spent a lot of time in Facebook. This whole scenario of enterprise didn't quite fully settle with me and the more I dug into it. What I realized was what these hyperscalers, like Google, and now obviously after that Amazon, Microsoft, all of them. Have done with their own data centers is like truly magical. The ability to scale to petabytes to kind of like just run massive amounts of compute. That technology was simply absent, from enterprise data centers. So the very basic idea behind Rubrik was, could you take the same thought process technologies that had been perfected in the public cloud and essentially bring that into the context of these enterprise data centers. So that was kind of the very basic, you know, kernel of an idea behind.

Pablo Srugo (00:02:56):
And you were at Google, at the time?

Soham Mazumdar (00:02:57):
So, I mean, I spent like six of my first years. First working years at Google. I was part of their web search infrastructure team. The whole objective was, how do you scale? You know, initially when Google had designed their web search index, it was designed for 4 billion documents. Eventually they realize it actually needs to work for a trillion documents. So I was kind of part of that scaling effort at Google to kind of, and with the project's name was in fact at Google that was about scaling Google to trillions of documents. So again, it was a, anyway. That's kind of just my original like formative journey and kind of realized doing that same architecture in the context of businesses. Who don't have the means of a Google. That was the opportunity and I think, we definitely hit a nerve. And, the company kind of just came out really strong off the blocks. In fact, got to this product market fit, quite quick.

Pablo Srugo (00:03:58):
How do you think about? One of the things I want to explore with you is just taking that leap. I mean, it's something that people, whether Google, Meta, or even these days, Stripe is a big company. You know what I mean? You're at a very successful business and you're thinking. And you've got this maybe idea that you want to jump into, you're thinking of jumping into. How did you think about leaving Google? Which is probably a high-paying job that was going really well. To start something that ultimately worked out, but at the time I'm sure was risky?

Soham Mazumdar (00:04:26):
Yeah, so my journey into Rubrik actually was more tortuous. My first company I did right after Google was something in mobile payments. This is before mobile payments were a thing. That didn't go anywhere, luckily, it landed me at Meta. So I spent a year and a half in Meta as well along the way. But again, in terms of jumping, I kind of credit the tech ecosystem in Silicon Valley for it. There is literally no other place in the world. Where risk-taking is admired and appreciated, like it is in the Bay Area. So in fact, people recognize that even if you start a company, it doesn't go anywhere. The learnings that you get in an accelerated timeframe is like, one year of a startup is like easily 10 years of a Google, right? I think that's a very, very special atmosphere to be in, which makes it all okay. Again, the other thing is the power of the network. Arvind Jain, you mentioned the founder of Glean. Actually, I met him the first day of my Google life. We were all having lunch and he just walked by. So I knew him from back then.

Pablo Srugo (00:05:46):
Did you work in similar departments? Or completely different parts of Google?

Soham Mazumdar (00:05:49):
It's like sibling departments. I wouldn't say I worked directly with him. It's just like pretty adjacent and then. You know, Bipul, he was an investor, right? He was an investor at Lightspeed. I'm the CEO of Rubrik and I tried to raise money from him unsuccessfully for my previous venture. So that's kind of how I stayed in touch with him. So again, this, you know, the power of the network. Working with other people in the ecosystem. The ecosystem itself wanting you to kind of take risks. So I think, yeah, it's a special place, to be honest.

Pablo Srugo (00:06:21):
When he came to you with the idea you were at Meta? What stage was, like, had he already? Because he was from Lightspeed, did he already kind of like have a round together? Do you have funding together? Where was the business kind of at? Where was the idea at?

Soham Mazumdar (00:06:32):
I think he had, you know, identified the idea. Because, again, his leaving Lightspeed was a big deal, right? You know, it tends to be a one-way journey. 

Pablo Srugo (00:06:43):
Yeah. 

Soham Mazumdar (00:06:44):
Like, coast and make money off other startups. But then, you know, like he wanted to jump back into the sort mines. So, at that point, I think Lightspeed was sort of like very, very curious. They were sort of in that, you know, they just had to kind of, go after this. Whatever Bipul was doing. For me, it was like, I was kind of looking to leave Meta. Because, once you've kind of tasted startup life, nothing quite cuts it. So I was sort of not fully there and then, all the other people connections kind of worked out.

Pablo Srugo (00:07:20):
And tell me, maybe we can take a tangent into the world of data centers. Obviously a world that you know well. Walk me through, the high level story is Google, Meta, these hyperscalers. They really know what they're doing there. The rest of enterprises don't, right? So maybe for like a late person that, obviously knows what a data center is. But not, you know, workings of it at all. What were some of the things that's happening in the data center that have happened over that period of time? That the hyperscalers got so right that you were then going to bring to the rest of companies?

Soham Mazumdar (00:07:51):
Yeah, so I think the story of the data center, at least back then. Was the story of, I would say, power to software, right? So there are actually, there are two ways in which you build technologies. One is, like it's a scale-up technology. Where your hardware just becomes more and more powerful. And even basic software has the ability to work on it. The second is your hardware does not become powerful, but you can put many units of it. So that you have the ability to kind of scale. Sort of like scaling infinitely, because you can keep putting smaller units of hardware, as many of them as you want. But what you need is the software needs to be much, much more sophisticated. To be able to harness the power across so many machines. So I would say that is the one piece that Google truly mastered and all of the technologies that are commonplace today, like Kubernetes. That was pretty much how its journey was inside of Google, or you hear about things like Hadoop. This came much later and that was very much birthed inside Google. That was a way to handle large amounts of data. So that was the key thing. That you move the power of how you harness this complicated hardware into software. Which has the ability to work across. So that's, I would say, the very basic nutshell and there are other things. Like you're going to think about how you separate compute capability from storage capability. There are a bunch of, I would say, associated technologies that were co-developed. I would again say give Google a lot of credit for kind of like really making these things mainstream.

Pablo Srugo (00:09:41):
And these other enterprises that had data centers. They just didn't have, their software just wasn't at this level. The things just weren't as optimized?

Soham Mazumdar (00:09:47):
No, no, no, no. It was just very, you know, the general philosophy was just buy like a even bigger piece of hardware and then you're going to be in good shape. That was just very wasteful. So again, we were able to kind of, you know, one of the. Again, from a product market fit perspective, one of the superpowers of like this scaling technology is that you can start small. So, when we go into a new organization as a small company, and you say, okay, give me a million dollars. Even if you believe that you should get a million dollars. The company is not willing to give you that. Because it's like, well, you're new. I don't know whether it works or not. But then you can say, okay, well, it doesn't matter. Why don't you start small? And you know that the data is going to grow, right? That's just like a, that's a general rule. But you can start small and you can kind of keep growing to it. So, you know, over a period of time. You kind of like accumulate a lot of value from a customer and have the ability to expand. So that is sort of like baked in. That was baked into the Rubrik architecture from day one. Which can really help them eventually get great scale.

Pablo Srugo (00:10:54):
Walk me through those early days. So you decide to leave, you decide to join. What happens? What are some of the first things you guys do? 

Soham Mazumdar (00:11:00):
We had a general idea about what we're doing, right? I mean, and I would say one of the beauties of Rubrik was the original idea had, very significant legs. But we knew that we needed a team. So the first thing that you do is to kind of think about who else you want on the ship. Because it was a very complex product. Again, we had a strong, I would say, core founding technical team. But we realized that, to build what we're trying to build. We needed more, right? So we kind of.

Pablo Srugo (00:11:28):
How many co-founders were there?

Soham Mazumdar (00:11:30):
There were four of us together. Three being technical and Bipul being the more GTM focused. But again, building out the team, high quality team. We got the cream of the crop from Google. Who had actually built some super large scale systems there. That was very important. So that was one class of things that we did.

Pablo Srugo (00:11:55):
Tell me more about that. How do you think about hiring? Was it all just people you knew already?

Soham Mazumdar (00:12:01):
I would say that was certainly a pretty big source of hiring. I think that, I mean, it was sort of advantageous for us that we had strong networks. We were able to kind of lean on it. But we did a lot of other things. So for one of the guys who is currently a Rubrik fellow, he was about to have a kid and then his wife was very concerned. How is the startup going to work? So we threw a baby shower for him. We got some flowers and champagne and chocolates and whatnot. So you kind of go all out, and once you kind of identify who you want. You have to kind of go crazy.

Pablo Srugo (00:12:47):
Do what it takes to get them all, yeah.

Soham Mazumdar (00:12:49):
Yeah, and that guy was not from our network. It was just like, we got a reference that is the best guy you can ever find. So you should just get him. So we then were like, okay, what is the best thing we can do to get him? So we realized that it was through his wife and making her feel good about the company. So we went all out trying to do that.

Pablo Srugo (00:13:05):
How many people did you kind of build a team to, before you really went out?

Soham Mazumdar (00:13:10):
So the first year, I think we. Maybe I would say we had got to 25 people totals. We actually add like a pretty fast. We knew that what we're building, we realized how complex it is going to be. I think Bipul had like great sense of the founding GTM team as well. So we kind of went after it. You know, VCs give you money to get to PMF quickly. Not to just, you know, this coast along. So again, so we were aggressive from the very beginning.

Pablo Srugo (00:13:44):
And how much had you raised? What was the seed round for Rubrik? $10 million?

Soham Mazumdar (00:13:49):
Yeah, so that was one investment. I would say on the product side, one of the good things that we did was we hired this SDR. Which got us in front of a lot of customers, which was great. So from all of that feedback, what we realized was. You know, as much as it is about bringing modernization to people's data centers. People wanted to connect to the cloud and so they simply did not know how they would do it, right? And cloud was a thing. AWS would say, hey, you get these free credits and so on, just come over. But, like, it's not very easy to come over.

Pablo Srugo (00:14:26):
Who is the first customer?

Soham Mazumdar (00:14:27):
It's weird. Our first customer is this like, think of them as a local hardware chain in New South Wales, Australia.

Pablo Srugo (00:14:42):
No way.

Soham Mazumdar (00:14:43):
Just like as far out as you can fly from here. That's where our first customer was. Yeah, I mean, I remember the name Lang's Building Supplies. I mean, they have like five or six stores. The CTO was like this enthusiast. I mean, we found him in some conference and he was like, yeah, let's do it. So one-year-old company, he says, yeah, just deploy this. Protect my data, no big deal.

Pablo Srugo (00:15:11):
And that was, that first product was this kind of new idea of just being an extension to whatever they already had?

Soham Mazumdar (00:15:16):
Yeah, I think. Yeah, pretty much. I think one of the things to realize is early customers, they tend to kind of root for you. So I mean, they almost like, I don't know. They feel great about making you successful. So they kind of take a lot of pain. Because early products are like terrible, right? I mean, I'm like a terrible art user. I'm just not happy with early products. But these guys, some of these guys really are, right? They give you a lot of runway leeway. So from your side, you have to take extra care of them. I mean, their feedback is cold. You have to, really pay attention,  really value their time. Give them like a super warm embrace.

Pablo Srugo (00:15:57):
Do you have to go to Australia to do any of this stuff?

Soham Mazumdar (00:15:59):
Yeah, of course. I mean, this guy, Eric, our first head of sales engineering. So he joins us I think, around December, January, something. About a year into the company and then first week, second week. We say, all right, Eric, you're going to Australia. All right, great. Let's do it and this thing that we had. We were selling, it was like 70 pounds in weight. So it was like, okay, how is that 70 pounds going to go there? Okay, don't worry about it. We'll let you ship it over, but then. But what happens in Australia, you figure it out. So anyways, he made it happen. That was quite a baptism by fire.

Pablo Srugo (00:16:43):
And was that like? Did you do like a free pilot or was it already paid? What kind of scale we're talking about at the beginning with that customer, for example?

Soham Mazumdar (00:16:50):
So we always did paid pilots at Rubrik. I think the rationale was that you want some skin in the game and then. There is this data center hardware component. So it's a lot of effort to get a POC going. So in some ways, if you're not at all vested in it. I think it's, it is not, you know, just press a button and you're, like a trial just starts. This was a significant amount of effort on the customer's part as well. So we want to make sure that there was skin in the game and deal. These deals were, you know, like $60K-ish. They're not like, this is not a million dollar deal or anything. But it was a great starting point for us.

Pablo Srugo (00:17:34):
Is that their, big expansion opportunity or not really? It's just a small customer, but to prove it out. So you could go do the big customers later.

Soham Mazumdar (00:17:40):
Yeah, I mean. I think, I mean, I would say we were not thinking expansion with the first 10 customers. I think it was more about, you know. I mean, going to an early customer or rather large customer too quickly is problematic. You know, their large customer needs are so peculiar and I'll give you an example. Facebook ended up being a customer. I would say in year two and they said, our data centers run on something known as IPv6. It's a particular way in which, servers are identified. It was all the rage. The idea was that the whole world would move to IPv6. Turns out to date, it's literally our one and only one customer. And we spent, you know, like months trying to make that happen. So I think, yeah, early customers. I mean, there's so many other things you're trying to prove. I would say expansion was not the most important thing that we were trying to prove out at that point.

Pablo Srugo (00:18:40):
And did you purposefully not go after these large enterprise customers at the beginning? Or you tried, but just they just wouldn't onboard. Because you weren't ready? 

Soham Mazumdar (00:18:50):
Yeah, I mean, it was both. I think we kind of realized that we wanted to start off with mid-market. We knew enterprise would come. It was in our destiny, but we did not think we were ready for them. You know, generally, it's like a completely no-name startup. It's hard for them to take a bet on you. So we kind of proved it out with mid-market and then we started realizing that there is a genuine pull that we were. I mean, naturally some of them would come in, but we did not start chasing them early on.

Pablo Srugo (00:19:22):
What was the, like, I get the value prop, but did they think about the ROI of it? Take a shop that's spending $60K or $100K on Rubrik at the very early days. Did they think about, okay, I'm going to get half a million from this somehow? Was there kind of an ROI math they did or not really?

Soham Mazumdar (00:19:37):
So there was a PCO math. Which is not quite ROI, it's like total cost of ownership. And, you know, you had to match or you had to beat the total cost of ownership any other way, right? And any other way would be like old way of doing things. The additional people expenditure and all of that, right? So we absolutely had that argument going, but there was a lot of pain. You know, I think one of the early studies that we encountered. This was like something from Gartner, it was something like sixty percent of people would be willing to replace this category of software within the next three years. So there was just those, the, the solutions back then were just terrible. They were extraordinarily fragile, extraordinarily, again. You know, a lot of effort needed in kind of keeping these things up and running reliably. Or you think you know you have been hit with ransomware and then you want to kind of recover. And then you realize oops, I can't. So this was a very common occurrence back then. So the software back there were not great, right? So I think and again not a whole lot of, so again unlike many other things what Rubrik started off solving was a very unsexy problem. You know, Silicon Valley startups were not going after unsexy problems. In that sense, we were able to capture a ton of latent demand that people had for change. You know, a lot of dollars here, a lot of demand, but for some reason it was kind of flying under the radar, in Silicon Valley. So that was one big aspect of what made Rubrik scale really fast.

Pablo Srugo (00:21:24):
Because these were people in general that were already trying to move to the cloud for a variety of different reasons. They didn't need you to convince them of that. They just needed a way to make that transition happen.

Soham Mazumdar (00:21:34):
Yeah, that's a great way to put it, right? It's like they wanted it. They didn't know how to. They were stuck with their legacy solutions. Which, you know, didn't really give them the opportunity. They wanted to change, they didn't know how to, they had the budgets. That was another great thing. It's like, hey, anytime you have data, data has gravity and data also tends to have budgets. So yeah, we were able to tap into it.

Pablo Srugo (00:22:00):
And so how quickly do you get? This first customer takes how long to land?

Soham Mazumdar (00:22:04):
It's about a year. I think this is a difference between. I mean, there are people who claim to create companies within a few weeks these days. That was certainly not the case. I think it took us a whole year to get the first customer. It took us, I would say, a year and a half to feel like this is a product that can, you know. I would not feel ashamed about giving it to people. With the first customer, I'm like, dude, things can.

Pablo Srugo (00:22:33):
Are you sure?

Soham Mazumdar (00:22:36):
Yeah, why are you using us, right?

Pablo Srugo (00:22:38):
And how fast did the ramp happen? Once you're talking, let's say the 2015 year. How fast do you hit a million ARR?

Soham Mazumdar (00:22:44):
Few months, not few months, I would say like two quarters. Two-ish quarters, I would say. It was quick, again, look, don't compare us to Cursor. I mean, that's a different.

Pablo Srugo (00:22:55):
Yeah, these days, quick is taking on a whole new meaning.

Soham Mazumdar (00:23:00):
Think about it, this is like an enterprise solution. Which includes deploying a piece of hardware within your data center. For a product of that complexity. It's absolutely astonishing to get to a million ARR in like within two quarters. I think that's the friction that is inherently present in putting a solution like Rubrik's. Despite that, to have that ramp. I think was quite extraordinary and it just showed that from a product market fit perspective. I mean, I would say Rubrik is one of the, you know, it's quite incredible how solid a fit it was able to find early.

Pablo Srugo (00:23:41):
Well, and this is part of the thing, like, the friction cuts both ways. I mean, in the lack of friction to. These days a lot of these apps, These AI apps are so easy to use and onboard. That everybody wants to give them a try and it's great. At least insane revenue growth. But it also, I'm sure in a lot of cases, leads to crazy churn. Whereas I would assume with Rubrik for the ones that installed it, you know, retention. You tell me, but what did retention look like?

Soham Mazumdar (00:24:09):
Yeah, sub one percent churn. I'm glad I'm not there yet, because I don't know the exact numbers. But churn was just not a thing and that was quite remarkable. Churn was not a thing. NRR was like, even at IPO time, it was like a hundred and thirty percent plus. So again, people expanded, people did not churn, that was the rule. That was great.

Pablo Srugo (00:24:35):
Do you remember how fast you got to$ 10 million from one? Like a couple years later?

Soham Mazumdar (00:24:38):
I think less than a couple years later is my guess. I would say year two, certainly we were, we had $10 million. That's well before year two ended. Year two of selling, right? So there is, so sorry. I think, there is a timeline. Which is, I'm not counting. Which is the first year when we didn't sell anything. We were simply building enterprise products. The people who had worked in enterprise previously. They came in and said, enterprise products don't get built in a year. It takes two years to build. We said, okay, we'll build it in a year and then from a scaling perspective. Certainly in two quarters of selling, you don't get to a million ARR, but we got there. And then I would say early in the second year of selling. Which is like, let's say two and a half years into the company.

Pablo Srugo (00:25:25):
And, that go-to-market at the beginning. I mean, the first few customers are kind of serendipity. After that, it's what? is it conferences? Is it just like, people just calling? What was the go-to-market in those, early years to get so many customers, so fast. That those level of customers too?

Soham Mazumdar (00:25:42):
I mean, it's a. I mean, it's combination. I mean, certainly like outbound played quite a bit of a role. But I would say one of the most memorable things I remember from, you know, early Rubrik days was what we used to do in conferences. So VMworld used to be like this, ultra mega massive conference that everybody in the industry used to go to. So the very first year we bought this massive premium booth and bought like this, DeLorean machine. I don't know if you've seen it, the Back to the Future car. 

Pablo Srugo (00:26:16):
Okay.

Soham Mazumdar (00:26:17):
The car with wings before the model X came along. I remember there were long lines of people trying to get a picture with the DeLorean machine. Again, that was brand building. I think, again, we realized kind of getting the word out. Making more people know about us was important. So again, for a few years, we were like known as the company, oh, the DeLorean company.

Pablo Srugo (00:26:41):
Would you go in? Would you guys do a lot of talks? Talking about migration to cloud and stuff like that? Or was it just a matter of showing up and doing some kind of stunts just to get people in the room?

Soham Mazumdar (00:26:50):
Don't call it a stunt, it was such a big deal. You have to, I think it's like a full court press, right? I think you have to, absolutely, evangelism was a huge deal. One of our key early hires was a person who was one of the top evangelists in the data VMware ecosystem. And he used to kind of talk as a, you know, as a free radical. Unassociated with any company, any vendor and we kind of convinced him that this is the future, you know, you should like really join us. And he was like, his name was Chris Wall. He was crucial to building up the, you know, our brand. Make people understand our value proposition. Likewise, I think, again, like Gartner. That if you're building an enterprise company, getting these analysts who are like the gatekeepers to the larger enterprise. You have to kind of get them excited about what you're trying to do. And then for them, it's also a lot of education. Because we are fundamentally saying that the way things were done earlier, or like the 10 companies you used to talk about earlier. They make no sense, you should do it our way. So for them, it's a pretty radical shift as well. It's like, hey, you this, you know, young punk abstract. You're telling me what I used to talk about earlier doesn't make sense. But you have to bring them along, right? I mean, you have to. So again, it's again, a lot of different things need to come together to make this. This sort of like enterprise GTM successful. I think that was something that we were able to execute on, really well.

Pablo Srugo (00:28:29):
Another question I have is, you know, you mentioned that between Google and Meta. You had another startup that you did. That didn't, you know, truly work out. When did you realize that Rubrik was just gonna be very different?

Soham Mazumdar (00:28:40):
I would say the first few customers is a very, big change. Like I think the, you know, building tech is something that I was trained to do and I was good at doing. My first startups, my main experience was that you can build tech that sounds cool and all of that, but it doesn't matter if nobody cares. Which ended up being the case, so. 

Pablo Srugo (00:29:07):
What was it, by the way?

Soham Mazumdar (00:29:08):
It was called Tactile. It was a small mobile loyalty payments company.

Pablo Srugo (00:29:13):
Got it.

Soham Mazumdar (00:29:14):
It was good. It was, you know, we built some like, before NFC chips were like ubiquitous phones. We used the microphone and the speaker to build a communication between a phone and this like device. And that's how you would you know communicate payment information and stuff. Anyway, it was a fun project but you know to scale up a business was hard. I think here I would say after first five customer conversations to me it became very real. That first of all I mean these guys are willing to pay so much money for this thing that barely works. You know once you kind of learn the level of pain your customers are facing and are willing to kind of go through to kind of make your product a success. That gives you a lot of validation that you're onto something and there actually isn't a proxy for it, right? I mean, everything else is theoretical, right? You can, you know, when you pitch to VCs. You have, you know, I don't know. Industry reports and, oh, look, this trend happening here, there. All of that is nice, but there is only one thing. Which is rubber meeting the road. Which is the customer feedback. Otherwise, there is no PMF.

Pablo Srugo (00:30:17):
Well, that's why I asked, right? Because the first time you start a startup, you just assume it's going to work. Like you kind of know nothing going into it. You're like, it's going to be huge, it's going to be amazing, and then if you have one that doesn't really work, like you did. Like I have, then you realize, okay, wow, there's so many other pieces of the puzzle and then you do it again. You probably less, you probably more jaded, more skeptical, but then all of a sudden. You're like, wait a second. This was real.

Soham Mazumdar (00:30:40):
Yeah, yeah, yeah. 100 percent. 100 percent.

Pablo Srugo (00:30:43):
And then maybe fast forward through that story a little bit. Just another point I want to talk about really quickly is when Arvind. I'm just curious on your take, you know, 2019. Rubrik is absolutely crushing it. You know one of the co-founders Arvind decides he's gonna go start Glean. I asked him, when he was on. How he thought about that, why he did that. I'm curious  just from your perspective as another co-founder, how you saw what he was doing.

Soham Mazumdar (00:31:06):
I mean, that was an agonizing moment for us, for sure. You know, when like Arvind. He'll probably not say it himself, but one of his superpowers is hiring engineers. You know, like he shakes hands with engineers and they just want to work with him. So that was.

Pablo Srugo (00:31:23): 
I can see that.

Soham Mazumdar (00:31:25):
Right? So we're a little bit worried about. Like, you know, we have this T-pop behind engineers . How are you going to keep it together.

Pablo Srugo (00:31:33):
He does have a bit of that superhuman type of feel. I mean, I've met him a few times now and you talk to him. And you're like, this guy's kind of on another level. I mean, that's for me to him. I don't know how you look at it, but for me. I'm like, this guy is.

Soham Mazumdar (00:31:48):
No, no, I admire him a lot. I think, his thing was. I think this search was something that he was just. That was his problem, you know, he really had to work on it. So ultimately, I think, when somebody decides to do. Pursue a certain avenue passionately. All you can do is support them as opposed to trying to fight it, right? It's just never gonna work. So, I mean, it was certainly a moment. I would say again, like, this is kind of where building a high quality team matters, right? I mean, one of the aspects of high quality teams is resilience. You know, I was there. My other co-founder, who also happens to be an Arvind, you know, between us. We kind of figured out how you would run the engineering team. I think, there was a, we had also hired, very well. So again, like overall team was in a great shape. I would say from a customer perspective, you know, we were. I would say like, I don't know if you know this. So Microsoft was like, eventually became an investor in us and the reason for that was that we were the biggest driver of data into Microsoft Azure. Even to date, we have driven a few hundred petabytes of data.

Pablo Srugo (00:33:03):
Was that the cloud that you tied into specifically? Or did you tie to all the clouds?

Soham Mazumdar (00:33:07):
I mean, we were agnostic, but we were actually the biggest to Azure. Second biggest to AWS, something like that. So I guess where I'm going with this is that, it was clear that the company is doing well, right? So that's number one. If a co-founder leaves and you have to answer two things. First of all, again, are you doing something wrong? And that's the reason why the person is leaving. And the second is, if you have answered that. And in my case, it was clear that every business metric, every usage metric. Whichever way you define the success of the company. It was all there and the second is how do you manage? Which is, he leaves, like, do you have a team that has the ability to keep going? And, you know, we had. Luckily for us, both of those metrics held up and we were able to kind of keep going. But yes, it was, you know, personally, I was pretty close to him. So it was quite, a bummer to kind of see him leave.

Pablo Srugo (00:34:05):
Well it's funny, I've asked him now a couple times. Like man, you start something it's absolutely crushing it. Why would you leave? You know what I mean? But anyways seems he had a strong pull and obviously Glean has worked out, so that's exactly. So then that was like 2019-ish or whatever. Obviously you stayed there. Did you, like? How do you decide? I mean yours is kind of post going public. Had you thought for a while that you were going to start something else? Or how does the idea for your new startup come about?

Soham Mazumdar (00:34:36):
Yeah, I mean for me. I would say eight years into the company, from a personal perspective. I had kind of realized that company has reached a certain stage. Where I could take a step back and it would be fine. Maybe I could have done that earlier too, like AJ did. But I didn't quite feel that back then. So that was one and the second thing was. What do I want to do? In my next chapter. I think certainly one option was, you know, Rubrik has taken a lot of leaps. I was actually part of the security pivot that Rubrik did. So that was certainly an option. But even for me, my own personal tech journey sort of started with search as well. So kind of like search, being able to extract information from data. Being able to do, you know, get insights from data. This is sort of what I was doing in my, very formative years. So that was a problem that was always at the back of my mind. The second thing was, again, like within Rubrik as it kind of was making his journey from this small startup to like a much more bigger one. I tried becoming this data guy for Rubrik engineering, unsuccessfully. Because I can realize it's not an easy problem to solve. So, again, it was at the back of my mind that I care about finding information. It's very hard to wrangle information. Can we do something about it? And then fast forward to end of 2022 with ChatGPT and so on. This one kind of realized that. Well, if there is one moment in the sort of history of technology. Where there is this quantum jump in terms of like what you could do. This was the moment and again. This problem was kind of like, very close to me. Which is, you know, how do you? How do you dramatically simplify the ability to work with enterprise data. Do analytics with enterprise data? And I realized that my passion and what I was excited about, or what I had been noticing. And this technology moment came together. That's when I made the jump. But again, it did line up with my own personal realization that, at least within Rubrik. I had delivered what I could and this was an opportunity to move on to something different.

Pablo Srugo (00:37:03):
And did you? As you're seeing, for example, Glean, but others maybe. But maybe Glean kind of go and grow. Did you think to your back of mind. Maybe one day I want to just kind of do my own thing or you kind of just were doing Rubrik and then as you saw the ChatGPT moment. You kind of jumped on it right place, right time.

Soham Mazumdar (00:37:19):
Ah, no, it was at the back of my mind. I mean, I think, once you're, you know, like, this WisdomAI is my third startup, right? So I would say after two, you become sort of unemployable. 

Pablo Srugo (00:37:30):
Hooked, yeah. 

Soham Mazumdar (00:37:32):
Right? There was like, I knew I couldn't really just work. So that was just out of the question. I think, venture and so on are potential things. But, you know, venture capitalists have to look at ten ideas and eight, nine of them back one of them. I tend to be a guy who just ends up liking nine out of ten ideas. Because I'm like, oh, that's cool. So, you know, I'm just an optimist by nature. So like, for me, I realized that I had to build rather than pontificate or judge others building. So anyway,

Pablo Srugo (00:38:07):
Tell me more about WisdomAI. You have Looker, Tableau, is it in that vein? Sigma Computing is one that we had on the show recently. Is that kind of the space you're playing or is it different?

Soham Mazumdar (00:38:18):
Yeah, I mean, that's certainly one way to think about it. So the way data analytics has been done, or if you look at the tools. You mentioned a bunch of tools. The way I look at it is, they are sort of analogous to Photoshop. Photoshop is something that graphic designers used to create their, whatever cool artifacts that they wanted to create. They are simply not, you know, if an end user has ever tried to use Photoshop without the training. They would like die. It's just not going to work. It's just way too hard.

Pablo Srugo (00:38:56):
And is that true of? You would say, is that true of Sigma computing as well, specifically?

Soham Mazumdar (00:39:00):
So it's a spreadsheet based solution. So a spreadsheet is great for a certain class of applications. I think financial folks really love using spreadsheets and it's great. But again, the power of chat is just.

Pablo Srugo (00:39:17):
Yes, everyone can speak.

Soham Mazumdar (00:39:19):
Exactly, everybody can speak and the second thing is, I think. Again, analytics is actually more. Your data is more diverse than something that just fits into a spreadsheet. We have customers who have tabular data, who have PDF data, and we have JSON data. There's just such a big variety in the modalities of data that ultimately to build a solution that does sort of like universalizes analytics. You need the ability to work with a variety of different form factors. I mean, the other big thing is, because these tools were designed for specialists in the past. The prep time to kind of get data into like perfect analyzable shape is enormous. In fact, you would have these six-month long consulting agreements before you can even deploy some of these solutions. And that simply is kind of, in my mind, it's quite ridiculous. You should be able to just start working immediately. You know, our value proposition is how quickly can you get started working? How simple can you make it? Can you make AI do work for you behind the scenes? You know, why is it an active process? Why don't you think about proactive? You know, every morning you should know what is it that you should be looking at. So there is just the opportunity to kind of reimagine analytics. It's quite extreme when you have a powerful AI friend to kind of lean on.

Pablo Srugo (00:40:52):
So tell me the story of this of WisdomAI. Like you leave, what? 2023? Do you raise money right away?

Soham Mazumdar (00:40:57):
Yeah, so we've been doing this for about two years. So we raised, you know, we used to seed round as well.

Pablo Srugo (00:41:06):
How much was it?

Soham Mazumdar (00:41:08):
It was $15 million total. Again, like my co-founders, they're essentially folks I have kind of worked with within Rubrik. So there's a strong bit of a Rubrik mafia here within the founding team. And I would say for the next year or so, we also focused a lot on just building the product. Not thinking about customers necessarily. I would say the last, towards the end of last year. Towards the end of 2024 is kind of when we in earnest kind of started getting in front of customers. Got a bunch of very exciting, you know, like even fortune 500 companies. I kind of realized that the data problem becomes more extreme as you start working with more complex the organization. The more pained you are by your data problems. It's just a general rule. So you're gonna realize that we are getting a lot of pull even amongst like larger organizations. There is also like, I would say between 2023 and 2025. 2023 AI was a curiosity. I was like, okay, sounds cool, don't know. I would say now people have kind of realized that, okay, this is it. This is going to change everything. There's a lot of embrace that's happening now, it was not there a year back. So again, seeing, you know, pretty significant demand. We have been able to get into literally our first customer was like. Our second customer was this Fortune 100 company. Where we started off with like a few hundred users and now the goal is to kind of move into like a couple thousand. So yeah, it's exciting times.

Pablo Srugo (00:42:50):
How many users do you have more or less across kind of all the, using the product?

Soham Mazumdar (00:42:55):
So again, like in terms of a total number of customers as in like businesses who are at this point. Either, you know, paying or POCs. It's in the, you know, about 30, 40 range. These range all the way from, you know, a fortunate companies to some smaller startups as well. The number of users within this, it's kind of that also varies. I guess, you know, like large company have the opportunity to kind of sell to potentially thousands. Smaller companies they will forever be within like, 20-30 users. So it's a there's a full spectrum for us 

Pablo Srugo (00:43:31):
And what's the hardest part of this product? Like, you know, the chat piece, a chat interface. I mean, obviously that just kind of comes for free sort of thing. But I would presume the hard piece is for the AI to really understand what you're getting at and being able to build visualizations, and grab the right data. Anyways, you tell me.

Soham Mazumdar (00:43:49):
Yeah, so I mean, number one is accuracy, right? So I mean, when you say, okay, tell me how revenue is going, right? So it turns out revenue is a piece of math that needs to be done and it needs to be done with a great amount of precision. You cannot make up a number, you cannot, like two people ask for revenue, they can't get different numbers, don't make assumptions, don't hallucinate. So all the things that you hear about  AI's challenges. They get dramatically amplified when it comes to analytics and this is probably the single biggest difference. Between what we are doing and something like Glean or, you know, many of the other. Let's say, consumer products that are trying to build data chat. So that's one, high, tremendous burden on accuracy and trustworthiness. And what you need to do, and to be able to kind of do this. You need to understand the bespoke semantics of particular organization's data. Because, again, even if conceptually everybody makes money and everybody has revenue. How exactly the revenue is calculated and where exactly bits and pieces of it is stored, is different from organization to organization. So again, capturing semantics, removing ambiguities, not having hallucinations, making it trustworthy. So this is kind of, I would say, the core of what we need to solve and then when you're talking about more from a usability perspective. How do you kind of like, you had to think about visualizations. There is a lot of like this visual storytelling that happens within analytics. So that is something that you have to be really excellent at. And, again, like lastly, the ability to kind of. Again, work across different data sources. There's a bunch of complexities that come in, but again, it's complex kinds of data. High accuracy, understand bespoke semantics, appification. Which is visualization, being able to do things proactively even before ASP. So it's a big gamut of problems that you need to solve and it's kind of like. It's different from, I mean some are common but a lot of it is very unique to this space versus problems that let's say a ChatGPT is solving.

Pablo Srugo (00:46:20):
And, you mentioned before, in one year of doing a startup. you learn like ten times what you would in a normal company. So, you know, eight years at Rubrik. I'm sure you learned a lot. What would you say are some of the parts with WisdomAI that are harder than Rubrik? And which parts are easier?

Soham Mazumdar (00:46:35):
I think so, okay, a few things. The easy part is easy to describe, you know, there's a lot of chatter about AI. Everybody's excited, everybody wants to do something, right? So, versus, like, it's easy to say what you're doing and people say, oh, yeah, that's nice and cool. The flip side to that is that, there is a tremendous amount of competition as well. I think one of the things that Rubrik got right was picking like an old problem. Which nobody was, ignored problem, right? When you ignore a problem, you get to be the only person kind of doing it. That was like one secret sauce that Rubrik had. We don't have it. On the flip side, there's this general enthusiasm for people to understand what you're doing. It's much, much, much higher. In Rubrik, we didn't get a chance, right at this point. We have about 10 Fortune 500 companies that are either a customer or a POC or about to start a POC in the next few weeks. That was not the case with GetThroughBricks. So that's something pro, something con. I would say on the technology side, they're both hard problems. I would say AI introduces this quality engineering as a huge requirement. How do you make something trustworthy? This is very different from Rubrik's brand of engineering. Which was much more deterministic. So again, as an engineering problem, this working with non-deterministic systems is a pretty big change. I would not call it easier or harder. Because again, Rubrik also builds a fit. It's just very different in terms of how you approach it and again. And going after a place where there is no competition, you are bringing in new technology versus a lot of competition. But there is also a lot of enthusiasm. So it's quite different between the two.

Pablo Srugo (00:48:29):
Would you say, now knowing and having seen. And felt what Product Market Fit really is. Would you say you have that at Wisdom.ai?

Soham Mazumdar (00:48:35):
I would say we have strong early signs of it. A few things that give me a lot of, it was quite exciting was we. When we land into some of these large organizations. The ability for us to be able to spread, that's quite remarkable. Few of the things, so that's one. The thing that kind of like bothers me about what we are doing is, the moment you're talking about business data. The security and legal scrutiny is, very high. So I cannot just say, here, try it. It'll just solve all your data problems. Because the guy will be like, hey, my legal guy is going to kill me. So I think I'm still going to have crack open that little bit more of this quicker time to try out the product. I think that's the one piece that I want to crack out. That's something that we are hoping to launch within the next couple of months. But again, what's exciting and which makes me feel great is customers and the POCs that we have, we are in. It's amazing to kind of see the customer's engagement and the desire for them to bring their own friends. Once you are kind of in. So, yes, it's early signs, but feeling good.

Pablo Srugo (00:49:54):
Was there ever a time, whether at Rubrik or WisdomAI. Where you actually thought maybe things just aren't going to work? Maybe you're just going to fail?

Soham Mazumdar (00:50:03):
I think, I mean. In Rubrik, I would say like the COVID three, four months were quite interesting. It's a very high touch motion, right? So, you know, you have to deploy this like hardware in somebody's data center and you can't. And COVID says you can't go anywhere, you can't meet anybody. But we are certainly like worried, okay, how the heck is this business gonna work, right? So that was certainly one moment. So that was like a required a bit of, re-figuring out how we do the GTM. In the case of WisdomAI, I think, you know, early on, this whole, like, working with a non-deterministic engine? How do you make this reliably work on enterprise data? It took us, I would say a whole year to kind of crack the nut on. What is this magic incantation training that we need to do that gets us the level of accuracy that we really need. Until then it was a lot of trial and error. I wouldn't say we were feeling that it wouldn't work. It was just like, took a while to kind of get to that point. So I would say those are two moments that come to mind.

Pablo Srugo (00:51:10):
And then last question. What would be like, you know, I'm sure you meet a lot of early stage founders just on a day-to-day basis. What are some of the most common pieces of advice you give? What would be your top advice to an early stage founder?

Soham Mazumdar (00:51:22):
A little bit of first principles thinking is one thing that I do talk about. Which is like, okay, everybody's GTM is different. So, you know, Rubrik GTM, not the same as Wisdom GTM, not the same as green GTM. So kind of take that, like, you know, completely fresh mindset as you think about. Don't just say I want to do that, may not make any sense. So that's like one thing. The second piece is there is a tendency to kind of overbuild, amongst first time founders. Because they're all like engineers typically and they're like, okay, we'll just build and build, and build, and build and people will come. Don't think about the GTM enough or what makes it really easy to get in front of people. So I think that's one thing that I always kind of advise. The third main advice I usually give is that you should just do it. Don't let people detract from it much. Because people don't know, that's number one. Number two, you have to live your life. If somebody has made a billion dollars and you're asking their feedback. And it's like, okay, I made it this way and yours seems doesn't quite fit the bill. It doesn't matter, you have to do your own thing. So I think that's probably the single most advice that I give all the time is that, don't be detracted. You have decided to do this. Go ahead with it and it's going to be okay. There are many paths to being great. Many of them go through failures, but so what? You just got it.

Pablo Srugo (00:52:57):
I love that, because I mean, startups are fundamentally a game of action. You just you got to take decisions. You got to move forward. You got to get stuff done and some of you right, some of you wrong. But you can't, there's no theory that will lead you to the promised land.

Soham Mazumdar (00:53:08):
Yeah, exactly.

Pablo Srugo (00:53:11):
Cool, man. Well, Soham it's been great having you here. Appreciate you sharing the story with us. 

Soham Mazumdar (00:53:14):
Awesome, thank you so much for having me.

Pablo Srugo (00:53:18):
So picture this, it's months from now, years from now, and one of your founder friends. A really close founder friends of yours, guess what? Their startup went bankrupt and it turns out, if you had just shared the product market fit show with them. They would have learned everything they needed to, to find product market fit and to create a huge success. But instead, their startup has completely failed. You have blood on your hands. Don't let that happen. You don't want to live like that. It is terrible. So do what you need to do. Tell them about the show. Send it to them. Put it on WhatsApp. Put it on Slack. Put it where you need to put it. Just make sure they know about it and they check it out.