fbpx
until LLM Bootcamp: In-Person (Seattle) and Online Learn more
Future of Data and AI / Hosted by Data Science Dojo

Building a Multi-million Dollar AI Business - AI Founders Reveal their Success Hacks

Vector embeddings are not just a feature, they are going to be the future. Initiate over the things that make you unique. That starts at the heart with vector embeddings.
Bob van Luijt
Co-founder and CEO at Weaviate

Listen on your favourite podcast app

Bob van Lujit - Future of Data and AI Podcast

This episode features Bob van Luijt, Co-founder and CEO of Weaviate—a real-time vector database that can help you build and scale AI applications. At just 15 years of age, Bob started his own software company. He went on to study music at ArtEZ University of the Arts and Berklee College of Music, and completed the Harvard Business School Program of Management Excellence. Bob is also a TEDx speaker, exploring the fascinating relationship between software and language.

Bob shares his journey, from early childhood and a fascination for tech to launching his own entrepreneurial venture at a very young age. They also discuss the importance of standing out and continuously evolving and adapting in a highly competitive landscape, and the potential of vector embeddings.

But it’s not all sunshine and robots. Bob and Raja also discuss the triumphs and tribulations of leading their own start-ups, fostering a healthy culture, and the key decisions that can make or break a young company. They also explore how to impress investors and build their trust so that funding isn’t a constant worry. Packed with practical advice and valuable insights, this video is a must-watch for aspiring as well as seasoned AI entrepreneurs aiming to make their mark on the industry.

Next Recommended Podcast: CEO of LlamaIndex, Jerry Liu on Generative AI, LLMs, LlamaIndex, RAG, Fine-tuning, Entrepreneurship

Transcript

(AI-generated)

Raja:

Welcome to the show, everyone. I’m Raja Iqbal. My guest today is Bob van Lujit. Bob is the founder of Weaviate, which I consider one of the foremost databases in industry. It is one of the foremost vector databases, in fact. Welcome to the show, Bob.

Bob:

Well, thanks so much for having me and your kind works in the introduction.

Raja:

So let’s get started with your early life. Bob tell us a bit about yourself. Your first introduction to computers. You know, high school. What did you do? What are you doing? Any programing earlier in your life? Really? How did it all get started?

Bob:

Yeah. So when I was like, I think around ten or nine, I’m born in 85, so that gives you a little bit of context and it’s relevant in the story. By then I brought home IBM XT right. So like a green and black screen computer and that computer came with Q basic. Q basic was a huge eye opener for me and born and raised in the Netherlands and where I grew up, there was like a library and you had these books with like programing for kids and there was a book.

Bob:

Q Basic for kids or basic or something like that. I don’t know exactly, but something like that. And I, I still remember that the first assignment was that you just had an input assemblies, just like input and then what, your name and then dollar sign name. And then on the next line it said like Echo High and then dollar sign name.

Bob:

And I remember that I tried that out. And at that I was like, mesmerized immediately. That was like, holy cow, this is amazing. I mean, I had to go a long way from there, but that was like my first introduction. And then when I was 15, a little bit before I was 15, so around like 12, 13, we got an Internet connection at home and I was like kind of able to connect two and two together.

Bob:

So I was starting to build websites just static websites, and there was a big Dragon Ball Z website that I made. So it was like that that kind of, that kind of started. But you know, the stuff you build at that time at that age and simultaneously at that time I fell in love with what music and with specifically a jazz which is, you know, at that age, I now know that that age that’s maybe a little bit unconventional.

Bob:

But for me that came very natural and I started to play an instrument and, you know, I was like kind of at playing that. So when it was time for me to study. So now fast forward a couple of years, I decided to go to college or to study music. But in the meantime, when I was studying at the college, I started my first software business.

Bob:

So I was basically writing software and making money off websites because everybody was asking like, who can make any website? So by the end of, you know, of and I finished my, you know, code story, so I have like a music degree or a degree music. So when I was done, I was like I was making a lot of e-commerce related stuff.

Bob:

So now it was like, 1920. And then I got a Fulbright to grant a scholarship to, to go to Boston. And in Boston I started at Berklee College of Music, and then I continued writing software. So that was like remote work of all kinds basically. And then when I was early twenties, I made a lot of you can still find other lines, a lot of art projects, also some art projects with software, but all that.

Bob:

In the meantime, I was writing software to basically pay for these art projects, but then I noticed how fast and rapid software, especially online, was happening and you know, and I was like that 24, 25 I was kind of done. I was like, you know, I’m proud of what I’ve created and I want to focus on software. An interest for me started also to slowly arise in business and like, why do we create stuff of value?

Bob:

And we’ll probably talk about that later. And I was hired as a sort of my now by software consultancy business, and I was hired by a publisher and I very well remember when I was because it was in 2015 and it was my first introduction to word embeddings. And because out of Stanford you got glove and I was introduced a true that publisher because they of course had a lot of unstructured data in the form of published documents and they were looking for ways to, you know, build new stuff.

Bob:

And while working on it, it was like I slowly the idea started to emerge like, hey, we can do search with this. And then I was at Google I/O and a year later, 2016 and back then already and I fact check this this so this is actually I double check that this is actually true. And it is.

Bob:

And so Sundar Pichai said we’re going to move from mobile first to AI first. And I knew.

Raja:

And what year was it?

Bob:

2016.

Raja:

Okay, Google.

Bob:

Google I/O 2016. And so people can build and Google that will find that. So it’s like that he makes that change from mobile first to AI was like, I think I know what they’re doing. They’re using vector embeddings in some way, shape or form, and that was how Weaviate was born. It was not a database yet.

Bob:

It was more like a, a way to store, you know, embeddings to we calculated centroid based on individual words, start to centroid, try to collect them, make data, connections and those kind of things. Funny story is like you can actually find an old video of me on the Google Cloud YouTube channel where I’m interviewed on something called Stack Chat, where I still refer to Weaviate as a knowledge graph because the term vector database didn’t exist yet, but it was all about these vector embeddings and talk about vector embeddings, showing how we make relations in the data.

Bob:

And then two things happened. So one, of course it transformer paper was released and that used to impact on us in the sense that all of a sudden because of an external factor, the way the just became way better. Right? And by then I also met my co-founder Etienne and we were like, wait a second, this is not just a feature. These vector embeddings, They are going be the future. So does a database exist that specifically focuses on vector embeddings? No.

Bob:

And then when we started the company and the company was officially founded in 2019, so we were working already since, I believe, July or June 2019, under vector embeddings. And then of course, yeah, the whole techy thing happened, right?

Bob:

So and then in the slipstream of that, all of a sudden these vector databases came after. But this year it’s this group that I’m saying, no, sorry, next year marks my, my 10th year like of working in the with vector embeddings.

Raja:

Okay So that’s a great overview. So just out of curiosity, right, So you said that you were working in embeddings so of pre vector database era or maybe vector database era, but not as mainstream. So what was your storage layer back then? Just out of curiosity, the technical side of me is just curious to the world, what were you doing?

Raja:

You mentioned that you were taking the centroid off, call them text encoding and not technically not semantic emitting. So how were they stored back then?

Bob:

We did that in the elastic with a separate index and because we always had the assumption. So it was very important for context to have in mind here. There was no business idea yet. This was just a you know, you know how to ghostwrite open source project. You start working on it, you’re excited about it, you’re not thinking business yet.

Bob:

So we’ll probably talk about that later because it has a huge implication on what we finished today. But back then it was just a project and we thought like, that’s probably going to happen in Lucene, Right? So that’s what’s the thing that sits on the elastic or seller. And what happened was and here the kudos go to my co-founder.

Bob:

He said, okay, wait a second, Lucene is somewhat suboptimal to do this. So and it has to do with the way that what the sharding mechanism and how to graphing start and out it’s merged together, blah, blah, blah, etc. But the point was that we were thinking like, Hey, wait a second, we believe that there is room in the market for a database that specifically focuses on a vector search.

Bob:

And what is important to bear in mind here as context is that the problem with brute force vector search is that it’s like it’s skills in a linear fashion. So that means that the more you add that, the slower it becomes, right? So if you have like 10,000 data objects, you’re not going to notice it. But if you have like million or more, it’s problematic.

Bob:

So they’re the role of these approximate nearest neighbors started to happen and there’s a website and benchmarks dot com and what you see in that website you see all these different approximate there’s neighbors algorithms and now the algorithm that was adopted for example we’ve yeah it was NSW and now one of your listeners might want a Yeah but hey wait a second, if I go to Asia and benchmarks.com and I look at what’s the fastest way in an algorithm that’s not a NSW, That is correct.

Bob:

But NSW had some very interesting important properties because if you have, if you create a database, you want to have crop support, create, read, update, delete. It turns out that the faster approximate needs they were algorithms did not support that. So then every time that you wanted to add something to the database, you had to rebuild the vector index from scratch.

Bob:

If you have like an e-commerce. Well, I can list many use case, but if you had use cases that want a database, you need to have an Asian library that supports that. So also in a library, let’s keep my nomenclature straight. And in an algorithm, then you need an algorithm that supports that. And that’s what was that’s built into the core of each edge, and that’s like what sits at the heart.

Bob:

And then we learn thanks to the open source nature. Appreciate that. An end is great and fact search is great, but just like to search is often not enough. And that’s where the impact of database starts started to differentiate itself because it’s got all these features like hybrid search, filtered search, those kind of things as certain types of aggregation.

Bob:

Certain types of integrations were where the community was saying, Hey, we need this to, you know, to build our vector search cases. So and that’s why the open source community plays a role, because that’s something I, I don’t know how we could have done that without the open source community. And.

 

Raja:

And do you think back then when you started, probably you were one of the earliest, if not the first one in all or how about because this is a long known problem, Do you think Yahoo! And Bing a long time ago, Yahoo writes in the Yahoo! Bing and Google, did they have any internal implementations of similar ideas or.

Raja:

Yeah, and but what did it look like? I mean, what did the landscape look like when Weaviate started and open source and some of the giants that already were in this space.

Bob:

Yes. So one thing that completely surprised me because we have an advisor in the company and we’ve hired who was working on the early effective search capabilities, leading the team, training Google ads. That’s the fact, the search use case. Right. So I was kind of assuming that, you know, if they do this at Google, if it’s so important, then somebody has got to figure this out.

Bob:

Then they got to productize this. But, you know, Google’s big company, right? So it’s like by the time, of course, like of smart people. So figuring out that this was something that people in the market wanted was not they figured it out. But by the time that you and actually have something and position that in a market that might take years.

Bob:

Right. That’s one thing. The second thing I think Yahoo! I think they open source that right. So at the risk of now saying something that’s from an historical context, not 100% accurate, but that’s what nowadays I think is fair spot right. So that is what that was open source and that’s our business as well. And they also mark all that like a I believe they call it a big data service serving engine.

Bob:

And I think the first open source project that started to really focus on vector embeddings, I think that was obvious. And I mean this we’re not talking years apart right? We’re talking sometimes months apart. And but the approach there was a similar approach to Lucene and Elastic and solar. We take face Facebooks or Meta’s vector search library and wrap that around the database but that it the challenge was that asserted earlier described that these ancient algorithms that always are suited to be a database that became challenging right so our thinking was like, hey, we can build something from scratch that is really like, that’s a product, right?

Bob:

So that is something. And there also the idea for the business started to emerge like, Hey, this is a product. And I really learned that randomly open sourcing stuff and then hoping to sell it, that just doesn’t work. That well. You really need to focus on building it as a product. And related to what I said earlier, that my interest in business was starting to grow, I started to connect a couple of things that I was like, Hey, wait a second, that this is happening from a, you know, a theoretical, a market, you know, perspective.

Bob:

Then this is how we can align it with we’ve shaped as a product. So that was like why we went the way, you know, to the path that we took. But this is all we’re talking months apart, right? So it’s all it’s all yeah. The point is like a couple of people at different places in the world figured out, Hey, wait a second, this fact, the search thing, that this is going to become a thing because all these machine learning models, they output vector embeddings, probably somebody’s going to want to do something from more infrastructural perspective.

Raja:

Yeah. So you mentioned Milvus and I was going to actually ask you to we hear about Milvus as we hear about Pinecone Quadrant and now I believe other mainstream relational and nautical players, they are also getting into this vector database market, right? So how do you differentiate Weaviate from, well, any other, any of the mainstream players out there?

Bob:

Yeah, I think what’s important to bear in mind is that the vector embedding itself is just a data type. So storing vector embedding is something we can already do for ages because it’s just a you know, it’s just an array of floating points or integers nowadays. But the point is like it’s just an array of numbers, right?

Bob:

Worrying it. So retrieving it, that’s a different story. That’s why these indices start to play a role, right? So that’s the uniqueness in these and in these indices. So the fact that we now start to see that almost any database under the sun has support for vector embeddings is, in my opinion, like a great thing that shows that there’s a need for it.

Bob:

Somebody sent me a video of Larry Ellison this week, even using the terms, the words vector database about the new upcoming release of Oracle. So I think it’s safe to assume that if it’s in the Oracle database, it’s established. Right. And I think that’s a good thing. Correct. So now the question becomes about I know you mentioned also new players, right?

Bob:

So also where we sit. So now the question is like, okay, so what’s differentiating? And if you look at that history, that’s like an age old thing, right? So and we did an interview with Andy Bellow from Carnegie Mellon the famous database professor. And I asked him the question that I said, like, you know, why is it that just one database, just in general, why are two so many?

Bob:

And one of the conclusions that we got there is like, that also has to do a lot with developer experience. And so what is a developer building, What does developer want, what does a developer meet? And that’s very different, right? So that is in the interaction which the database are very different. So for example, in a no SQL time, when we start to see these Jason, these document databases, right?

Bob:

The post press people said, you know, you can sorry Jason object in Postgres, but of course the developer experience was very different. So talking about affected databases, it has a lot to do with the quality of the database, which gives a certain type of performance and those kind of things. But that’s kind of the low hanging fruit because you just put dollars in that can and great developers work on that, but it’s the developer experience, how you educate people, how your clients work, the features that you have, you get to these features that is very differentiating.

Bob:

So one thing we say, for example, we’ve yet is that we like to use a terminology, a native, right? So I was saying like we view it as a native and that’s different from just the search that we support way more things than just fact, the search to build a native applications. And so that is the big difference in fact so model integrations, hybrid search filtered search storage of multiple embeddings on data objects and so on and so forth.

Bob:

And we’re going to have a lot of stuff coming up in the pipeline. So that’s the big differentiating factor.

Raja:

And this is the differentiating factor compared to, you know, your traditional mainstream, you know, relational and no physical databases. I know Mongo also introduced better embedding. I was not aware of the Oracle introducing that, so that must be a moment of pride. But you’re right. So, you know, here finally they acknowledge that you know better embedding matter database.

Bob:

Yeah this is also what I always tell tell my team right I said, okay, we have two options or nobody adopts it. So yes, then you’re alone or like with a very tiny group of companies doing something. But if it’s not adopted, then, you know, developers don’t want it right. But the other end, if developers really want it, a lot of people adopted.

Bob:

So then you differentiate over the thing that makes you unique. And it starts at the heart with the fact of embedding. But that’s just a starting point. And then there’s like a whole thing around that. I often give it as the metaphor. I often give this an avocado, they say an avocado, the center of the avocados, the pit that’s like the vector index, right?

Bob:

That is the heart’s part. You need that. But the flesh that sits around it, that’s what you eat. And that comes with documentation, that comes with client libraries, that comes with performance, which scalability to call the ease of use, the education videos, tutorials and so on and so forth. And that is very different in, in an AI native company than a more traditional company.

Raja:

Okay, So but if I phrase it differently, it’s about the focus that you have, right? So basically you’re focusing on AI applications then, yes, that’s your sole focus. If, if I want to build an API application, maybe it is my choice. Yeah. That case, that’s that’s great. So, you know, being a founder myself, I would like to actually understand, I mean, your perspective on what is the hardest part of being a founder and especially in a space that is which I believe is getting hyper competitive nowadays.

Raja:

Everyone is pouring money. Google claims should be opposed. And so there’s a lot of competition. So in general, I mean, not from a business standpoint. You know, it could be in the people side of it could be the technology side of it really end to end startup. What is the hardest part of being a founder?

Bob:

Yeah, so ironically, so let me first tell you what is for me personally, what is not right? So it’s like a because you mentioned this like outer space is evolving and those kind of things, the competitive nature, I have to say, I enjoyed a lot. It just, you know, I’m very proud of the technology that we built. I’m very proud of the community that we built.

Bob:

So it’s okay and I enjoy it. It’s like, that’s fun to me, right? That is that that is the and I once I forgot to set this, but I once had a like I was I was a founder of a I forgot to set this, but there was like everyone’s in a space like where we and Right and and you compare it with the Olympics it’s not like the Bobsleigh team, right?

Bob:

It’s like the 100 meter sprint. That’s and then you need to win from Usain Bolt. And that’s fun, right? That is. I love that. That’s fantastic. So the thing to ask your question so that that’s what is not for me. But to answer your question, we are a remote company, fully remote company. And every year, at least once a year, we get together with the whole company at some in the world because we have people literally all over the world, we get together somewhere.

Bob:

And right now, what is it, 80 people or something? And then they see all these people, right? And these people, they get married, they get kits, they have houses, they have families, and they put so much effort, like literally like, you know, blood, sweat and tears in the company. And I try my best in my role to make sure that I can provide for these people.

Bob:

Right. And I’m providing these, like three things, right? So that is in my from my opinion, that’s efficient. So, you know, where are we going culture. So who are the people we’re working with and enough capital to do that, right? So that’s related to any type of capital revenue investments, whatever. So those three things, that’s what I’m responsible for.

Bob:

And you know, sometimes you try to figure something out or sometimes something some very difficult, and that’s what I find difficult. So that I think about the people, the people that spend so much time and effort helping me and my co-founder built a company that I want to make sure that you can be part of this journey for a long time.

Bob:

So that is what I what I sometimes find if I see all these people that I find too difficult. But, you know, it’s a catch too because on the other side, I enjoy that too. That’s how I would answer your question. It’s like it’s very people focused, right? So that’s where my concerns come from. Not so much tech focused.

Raja:

Now that’s created. So I’m glad that you mentioned culture. So at Data Science Dojo, one of the things I think I mentioned this several times a day when I’m interacting with my engineer is, you know, marketing everyone and pretty much everyone, not just myself, right? So we have it is our culture to actually keep one of the foremost value is that we focus on our culture.

Raja:

That’s part of our culture to write in. It is so important. So how did you build a good culture as a startup? I mean, was it difficult to actually build a culture? And, you know, sometimes there is this fallacy about culture and, you know, those vacations and trips to Hawaii and then, you know, those latte machines and massage chairs, I mean, in my opinion, that’s not culture, right?

Raja:

So I would like to hear from you. What do you think is culture and how hard it is to build and what are some challenges?

Bob:

Yeah, I think the this is this is an excellent question, by the way. So let me let me think about giving you a proper answer. So let’s first start with a cliche. So I think that the cliche of the the founders set the culture that is true. And that let me give you two examples from how we do that.

Bob:

So and I think that for those listening to this, who might want to become founders, the nice thing about is it can be anything you want, anything that is for you, important. You can pick. So one thing that was important to me was that I and I had another advisor in our company and a conversation about this very early on in the journey, and I was thinking about these kind of things, right, How do I define this?

Bob:

And at some point I got this, this insight that I was like, you know, I want people to see the fun in the companies so people should and enjoy that. They should also be full of urgency. I don’t like nasty, backstabbing behavior to reach our goals. So the word that I got to that was like, I want I want to be kind.

Bob:

So we want to be kind to each other, our community, our customers, but also our competitors, etc. Right. And I had these words on a screen and I was looking at these words, and all of a sudden I had this moment of epiphany because I had like, joy, happiness, kindness, all these words. And something that that came to mind was like, there’s one word that stands out, and the word that stands out is kindness.

Bob:

And I showed that to the team back then. The team was way small. And I said, the reason that it stands out because words like joy, happiness, those kind of that’s what you feel. But kindness is something you don’t feel. That’s something you give. So you can always make that decision to give it and this same advisor told me beautiful story because he said, like, this is a building startup, his heart, right?

Bob:

So if you if we like it or not, it’s going to be heart and told me this beautiful story about you said like every mammal one thing that every mammal has in common is that by the time from birth to death, we have like a billion heartbeats. So our heartbeats around a billion time. So that’s why an elephant gets older than the mouse, right?

Bob:

And he said, So as a human, we know that on average it’s a billion heartbeats. We can’t change that. That’s just nature. But what we can do is that we can decide what we want to spend those heartbeats on. And that’s something that I tied together to this story about kindness. And, you know, so being, being kind and I’m no saint, right?

Bob:

So I also make mistakes. And then also I can I can get angry. Right. But that trying to be kind was something important. And one way how we know what we know doing the company that is like became a little tradition that was actually born for me studying music was that when the company was very small, like eight people, my co-founder, H.A. and I, we started to cook for everybody.

Bob:

So every time that we got together as a company, there’s a co-founder dinner that’s just us cooking for everybody. And back then when we were like eight people was like, We will do this forever. So every time we get together, this whole company, the co-founders will cook a meal. And so the last get together was like 70 people or something.

Bob:

So this whole company, like we had like candles and music and table and still us cooking for the people as a token of appreciation for people helping us out. So that is how I decided to do that together with Asian. But that’s not a silver bullet, right? So it is like people can do whatever they want, but that’s what we decided to do.

Bob:

And now to your question, what’s hard about that? That sometimes you hire somebody that can deal with that, right? And then you need to let that person go. And that has happened that we’ve had that not I mean, I can count that literally on one hand, right? So it is like just just one or two times. But that’s heart, because then you then you’ve got a grip.

Bob:

This is not working. And then it’s very hard to decide come to the decision like this person does not fit the company. So that’s the hard part of it. It’s like it’s nice to have those soft skill words and fluffy words, but that means that on the other end of the spectrum, you need to be just, you know, brutally honest as, okay, this doesn’t work, this person, you know, because the downside of doing that and companies and I, I like to borrow this word from Ed Catmull from Pixar like they have the concept of like the circle of trust.

Bob:

So people are in the circle, but if you have like a bad apple in a circle that’s toxic, so you need to take it out again. It hasn’t happened of like one or two times, but that’s difficult. If you want to maintain such, such a culture.

Raja:

And you you don’t have to convince me because I have my own fair share of these things. I mean, as you were you were mentioning, I love the honest conversation. Right. So, you know, yes, we try to define the culture, but at the end of the day, founders are humans, right? So, you know, we tried to set standards and we try to live up to those standards.

Raja:

But at the same time, we are humans, right? So at the end of the day, we’ll make mistakes, too, as long as we are open and we acknowledge that we made mistakes.

Bob:

Yeah.

Raja:

And and move on. And also you mentioned this example where you once in a while you will bring someone in, was not a cultural fit and you know, you have to fix that. And so we have a lot of emphasis actually within Data Science Dojo Right. So we emphasize a lot. We have hired people who did not actually meet the technical bar, but to the fullest.

Raja:

But we would actually hire them because of, I mean, being a very good cultural poet and we have had a lot of success, right? So I think we can coach, we can teach, we can upskill people, but I think fundamentally changing people and making them a fit, it may or may not work. And all cases. So I love you know, that that honest answer here.

Bob:

So no, I, I appreciate you saying that. And I think the the important thing here to it’s something you might find interesting is I actually have a little theory about it, and I’m still planning to I will once write this on the blog post, but the reason I haven’t done is because I haven’t figured it out in the details yet.

Bob:

But I’d say and it’s related to what you just mentioned about when you’re hiring a data science dojo that you think about, you know, is is the 100% technical fit. But do we then need to accept that this person does not fit culturally those kind of things? And this is something that I call the long tail of kindness.

Bob:

And it’s this If you build a culture that’s very direct and very harsh, then success might be right. So you hire somebody just for the technical expertise they might like, very unkind to their coworkers. That gets an A, and then something might be a problem. It is very hard to maintain that because people don’t like to work with this person.

Bob:

You might not like to work with this person and so on and so forth. So then it drops down and then you have like very short tenure at the company. However, if you take time and you know that somebody fits, the company might take a little bit longer to get people up to speed. But then this is a great addition to the company does great work.

Bob:

And because everybody likes each other and they do better work together, then there’s a huge long tail. So if you go on our LinkedIn, for example, just we’ve just LinkedIn and you see how long people stay at the company, that’s pretty long. And that’s hey, I mean, if people the company that are still here from day one, right?

Bob:

And that’s that long delay. So depends a bit on what kind of culture you want to build. But if you also want to go down that path of having values like, you know, for example, you also means like kindness, then you can keep in mind that you build a culture with a long tail rather than a culture that peaks.

Bob:

And I mean, I know founders are very successful at doing the first thing, so and they just they just burn through people very fast. But, you know, these people also do work and their products are also, you know, rapidly grow. So it’s like it’s one or the other. It’s whatever you want to do. But I don’t want to do that.

Bob:

I want to do that, the one that I just described. So that’s kind of how that’s different.

Raja:

And that’s a more sustainable way of doing things right? So it’s not about bringing someone in for a short stint at the expense of the cost of the culture, right? So, so I totally get that right. I mean, we are definitely on the same page on that.

Bob:

So yeah, but, but one more thing that because this is so interesting, I don’t have a problem with people doing it the other way. If that works for them and that’s how they want to do it, absolutely they should it. But I also just have my own moral and ethical standards as a founder, and this is how I want to do it right?

Bob:

So it’s a it’s more that it’s more just there are many ways that you can do it, but this is how I am. And then apparently you as well, how we want to do this, right. So and that is just you know, that’s just how we do it and how we’ve decided to build the companies. And also for you guys, I mean, you guys are pretty successful, right?

Bob:

So it seems to work.

Raja:

It seems to work. And I think there have been many false starts. I mean, I have to admit that. I mean, there have been false starts. I mean, because as a no one teaches you how to be your parents don’t teach you how to be a founder. Right. So universities don’t teach it. You know, it’s it is something that you learn.

Raja:

And one of my mentors in early days, he told me, I’m going to share things with you. I will tell you. And 50% of the things that I mentioned to you, they may not be actually applicable to you. Right. So they may you may end up in a very different situation. So because everyone has their own journey, every startup, every, you know, the timing but juncture, they are at, you know, personalities, you know, said there’s there are so many variables.

Raja:

So there is no fixed advice. I mean, you do this or you do that, right? So you have to be actually very mindful, considering your circumstances and how you want to actually set the tone and what decisions you want to take. Thank you, Bob. This was very insightful. So I know we have been talking it’s close to 2 hours.

Raja:

I mean, I could talk to you for the entire day, but I want to be respectful of your time. Maybe one last thing about coming back to your website and your entrepreneurship and what’s the road ahead. So you raised $50 million in seed equity. you raised $70 million dollars, right, in Series B last year or combined last year this year. Right. So and this is significant. So what did you tell the investors? What are you going to do?

Bob:

So I’ve been I’ve been in a very fortunate situation that except for my seed round, I have not been actively fundraising in the sense like, you know, I’m running out of money, I need help. What started to happen was that new pattern started to emerge. So what people had seen from the know cycle waif, the people like, Hey, there’s this new type of database that’s emerging which has not happened for years, right?

Bob:

Or maybe even a decade. And that especially funds with knowledge about this base. So like, hey, we want to help you, you know, to build because we have a lot of knowledge, but we’ve known from the best that you might be able to apply what you’re doing right now. And that was very attractive to me. So I’m very, you know, blessed with some big investors that are really good at that.

Bob:

So and they’re helping us. And for those listening, raising a lot of funds does not equal spending it. Yeah, I mean, for some people it does, but that’s not for me.

Raja:

Not for the sensible founders. Right. So you don’t go on a spending spree after you raise money, Right. So yeah.

Bob:

Exactly. I’ve raised one month of Sam Altman burns right but that’s a joke so what you try to do is that you we are currently in this we made this transition from an open source company to an open source company to, sorry an open source project company to an open source business company. Right. That’s a transition we’re making.

Bob:

So and we started that transition like about a year ago. And it means like that goes to the customers that we’re onboarding and the customers that we’re working with ranging from startups to big enterprises, right? And building the team to do that, the go to market team to do that and those kind of things, keeping R&D going to keep working on all the exciting stuff that we’ve just been discussing with generate the feedback loops and those kind of things.

Bob:

And that is how you, how you use these kind of funds moving forward. And the, the thinking is and then at the risk of oversimplifying a little bit from the many problems you will encounter building the business, you do not want lack of money to be one of them, right? That’s such a that’s such an unfortunate situation, if that’s the situation.

Bob:

Right. So and that is kind of what but but you get that right So so you get the opportunity to truly build a great company. And that’s why there’s so much excitement, because people have seen like, hey, we’ve seen we have done pattern matching from what we’ve seen in the past and maybe with AI right now that’s repeating itself.

Bob:

And but we raised our series A. That was definitely maybe, but we raised a series B, It was not that maybe anymore and especially today is happening, right? So it’s like it’s a people are bringing this stuff to production, these AI native applications, and that’s where the market sits, right where the market is growing. So that’s how the funds are related to what we’re doing and the role that it plays in, you know, capturing, you know, markets tell you if you up.

Raja:

Okay. And one last question. So are done after that. So what are some of the most exciting things that you think on the technical side and the business side? Any launches, anything that you would like to share with that us? Right. So what is new in the next 6 to 12 months?

Bob:

Yeah, so we have a lot of stuff coming down the pike when it comes to ops from an operational perspective. So managing different indices and indices pertaining to like in memory on this compression algorithm, scalability, new consensus algorithms, all that kind of good stuff to run this stuff in production. But if I’m going to be honest the thing that I’m also like really excited about is what we discussed, like with the Gen AI feedback loops and the fact that we’ve hit becoming the context window, the work that’s happening there to really weave the models and database together.

Bob:

That is something that I’m like really excited about. I mean, the first thing to because that just that’s just what our customers need to have the customers ask for to scale their production cases. And that’s super important. And I love that too. But I get so excited about the new stuff that’s coming to true native stuff, because that’s what will mark this new era of a native infrastructure and you know, the businesses that are built on top of that.

Bob:

So I guess that would be my answer.

Raja:

Yeah. Thank you so much, Bob. It has been a great discussion. Thank you for being here and taking the time to talk to us.

Bob:

Of course. And thank you for the wonderful questions. I really enjoyed it.

Raja:

Thank you.

Subscribe to our Podcast

Subscribe to our podcast for the latest insights in data science, AI and technology. Let us know if you would like to know about our webinars, tutorials, newsletters and more!

Join our growing community of 900K+