E681 | Emil Eifrem, Neo4j: Building the AI Infrastructure Layer: Neo4j’s $100M Bet

0:00Today we are joined by Emil Eifrem, the founder and CEO of Neo4j. He recently launched a$100 million push to support startups building with graph databases and Gen.AI. We'll dig into why graph thinking matters now, how the program is designed, and what this means for the next generation and wave of AI infrastructure.

0:35This show is not investment advice, and the hosts of this episode may be invested in the funds and companies featured. Emil, great to have you with us today. Great to be here. Yeah, thank you. Emil, we met many years ago, right? Tell us and tell people listening in who and what is Neo4j. Yeah, so back when we met 138 years ago or something like that, Neo4j was a database company or wanted to be a database company, right? Obviously, your prior firm led the seed round in us, right? So we're now a grown-up database company and we're broader than a database. It's a data platform with a graph database at its core and we can talk more about what a graph database is.

1:23But it enables our customers, which are primarily global 2000 customers, to write what we call intelligent applications and to reduce hallucinations in AI. And we have every single one of the 20 biggest banks in the US are now customers, 20 out of 20, nine of the 10 biggest pharma companies in the world, all of the 10 biggest car companies, and so on and so forth. So deep, deep adoption horizontally across all verticals that I know of and all geographies on the planet. Yeah. So one of the things that I do remember is something around the Panama Papers. Could you disclose that a little bit for the listeners so they actually know what Neo4j can do?

2:09Yeah, this is like one of our coming out parties, right? Like so we have a developer go-to-market motion, right? So we sell, and I put sell within air quotes, to developers, which means that to normal people, and I say this affectionately as an ex-developer myself, but to normal human beings, we were very unknown. But then in 2016, the Panama Papers happened. And the Panama Papers, for those of you who won't remember, because it's almost 10 years ago now, it was the biggest news story worldwide in all of 2016 until Brexit happened and then Trump. They won the Pulitzer Prize like the following year.

2:45So it's like a massive story. And what it was, was there's a law firm in the Panamas called Mossack Fonseca, who specialized in tax planning, but it turns out, which is legal, but it turns out that it was also used for tax evasion, which is obviously illegal. and there was a data leak. It was the biggest leak in journalistic history at the time and it was leaked to this organization ostensibly based in DC called ICIJ, the International Consortium of Investigative Journalists. But they were really small, like 11, 12 people worldwide and they got this big leak and they wanted to try to make sense of it.

3:26It was 1.6 terabytes of data and this is like scanned passports and order forms. It's just like, what do you do with that? Like, how do you make sense of it, right? And they had found Neo4j in a prior investigation that they'd done. So they took the data, they loaded it into our graph database, and it came to life. And so then now let's talk about what the technology actually does. So instead of storing information in tables, which everyone here in this call will be familiar with Excel, so we know the power of a tabular representation. And it is really powerful. And that is how all databases were, like since the 70s, that's how all databases worked.

4:05And it's really good. But when you have data that is really messy and connected and deeply chaotic, a tabular representation is not the best form. And so what we invented was this network model. So instead of storing data in rows and columns, you store it in nodes, so think circles, that are connected to other nodes through relationships. So then back to the Panama Papers, Sigmundur Gunnlaugsan was the Prime Minister of Iceland at the time. He did not have a bank account in an offshore tax haven directly. But what they uncovered by taking all that data and loading it into Neo4j was that Sigmundur lived at an address.

4:48At that address, someone else lived, who subsequently turned out to be his wife. That person was an officer in a company that had a subsidiary, which had a subsidiary. That subsidiary had a bank account and an offshore tax saving. So you can see here, this is like six, seven levels of indirection. You would never have found a normal investigation, right? Exactly. If you just scroll through a table and you look at it, it's impossible to find that. But with this network graph visualization, it just came to life. And so, so much tax fraud was uncovered in this way. And that in some ways was our coming out party, which ended up being really meaningful from a business perspective because that happened.

5:32Literally the next day, all the IRS equivalents or the tax authorities of the world started calling us saying, we want this. We want this. What is this technology? That must have been wild. That must have been wild. Exactly, right. But not just that. All the big banks called us as well. right because maybe there's this narrative that the banks would allow it to happen i mean i just told you we work with all the big banks i'll tell you those guys don't want to break the law right and so they call us up it's like wait how come this you know rag shag band of independent journalists underfunded with like grassroots technology how come they know more about our customers than we do.

6:16And of course, to me, that was like, are you kidding me? That's like open goal, right? I was like, you're using the wrong technology and let me tell you. No, but now I'm going to direct you into the topic, right? Because two things has happened since then, right? A decade passed by. Neo4j has grown up. You are a real company today. But then we have seen more tech developing coming, right? Now we have Gen AI, right? So you now have this$100 million initiative. And what is driving that? And how do you see Graph AI becoming a core infrastructure layer for startups in the Gen AI area? Yeah. Let me start with the second part of your question, which is kind of the end goal.

7:04And then let me use that as framing for why we've launched this startup program and initiative, right? Yeah. But it turns out that this way of representing information is deeply intrinsically related to knowledge, right? So if you think about what knowledge is, when you try to learn something, the people who listen to this podcast right now, they're trying to understand who I am. And they do that by taking the unknown concept, in this case, ML or Neo4j, and they relate that to previously known concepts. Like you organize it into the concepts you already understand. That is what knowledge is. And how do you organize it?

7:42You relate it. So relationships, the lines between the dots, are absolutely crucial for understanding things. By the way, in English, you can even hear it in the language. You say, connect the dots. That's a euphemism for understanding something, right? And so it turns out that our way of representing data is deeply knowledgeable. And it's a perfect fit for AI. So if you put your organization's information in what people call a knowledge graph, That's the perfect way of giving an LLM, giving an AI model, access to that information. Right? So this we saw. And the backdrop here is for about a decade or so, when you and I first met, then we sold it to anyone who had a graph problem.

8:25Big, small, as long as they had money, we took it. But for about a decade, we've been exclusively focused on the global 2000. Really the enterprise, like so a billion and above in revenue. And that's been phenomenal for us. It's driven all of our commercial success. And I love it. I wouldn't change it for a bit, right? But if you now think of this massive tectonic platform shift that is going on in AI, where the entire world is replatforming their tech stacks, right? We have this core technology that is a super good fit for this. But then you think about when will this happen first? Well, it is likely to happen at the earliest amongst startups and tech native companies.

9:10So if we keep selling only into the global 2000, only into the biggest banks in the world, biggest telcos, biggest pharma companies, and so on and so forth, right? We will miss out on the early adopters, right? I agree. And so that's what we said, like, we got to get back to our roots. We got to target startups again. And the problem right now is that if you were a startup like six months ago, right, and you wanted to talk to Neo4j, first of all, we barely even pick up the phone. Second of all, when we come in, I don't know if this is going to be a visual podcast, but I sit here in a t-shirt. It is.

9:45We usually don't show up in t-shirts. We show up in suits with a tie. And our entire body language is, if you can't pay me$250 ,000 just to start, I'm not interested. And that's great for Bank of America and Maersk and Novo Nordisk and Verizon, right? But the startup crowd is like, go away. I'm going to find some cheaper option, right? You know, and so even though we have like our core technology is a great fit for this next generation of startups, like our price point and our packaging was not right for it. So those two things were the driver. Great promise and potential, but a form factor and packaging that wasn't right.

10:35Yeah. So that is kind of the driver. Should I go into like the program? But I think for me, this is great, right? Because the way I view you now is like any other that I have on the podcast, right? you're almost big corporate now, right? You have all your processes, you have all of that, right? You're not that agile Emil that I met in 2010, right? You know, this is a different story, right? So maybe, you know, can you walk us through how the program is structured? You said it's for the startups. What do they get and what outcomes do you hope to see from it? Yeah, so the outcomes we hope to see for it, and then let's talk about the more important thing, like why is it valuable to them?

11:15But what I'm hoping to get out of it is the next generation of AI-native startups should be built on Neo4j. That's kind of what I'm hoping for. And it's already, despite what I just said, kind of our body language and how we show up and the form factor and the pricing, it's already happening. Despite that, right? And so when we launched this program, I can tell you a little bit of stats about what's happened since we launched it, I guess, six weeks ago or something. But it's taking off. So that's the goal from our perspective. From their perspective, what do we get? We have an on-prem version of our product, which is open source, but the main form factor is a cloud service.

11:54And that cloud service is called Neo4j Aura. And it has all the things that you would expect, which is, we operate it for you, you just sign up, and you can focus on building your application, running your company, rather than how to manage a database, right? And so there are three pieces of it, of the startup program that we offer. The first one is free credits on Neo4j Aura. It's our cloud service, right? This is just easy to get up and running, don't have to pay, you know, that kind of stuff, right? And as you can imagine, part of that is coming from us, but then we also partner with the big cloud providers underneath us, right?

12:29Like the Google Clouds and the AWSs and the Asher's of the world, right? So that's the first piece. The second piece, honestly, is probably the most valuable one, which is we have a dedicated team built out with graph expert engineers who will work with these startups for free. They will look at the architecture. They will review. They will help them with design patterns. They will engage. But even as far as brainstorming ideas or what are other features you could build now that you have this graph power, you know, that kind of stuff, right? And I think that's generally good for new technology.

13:04But man, are you kidding me? AI, everything is moving. I mean, none of us have ever seen technology move at this pace before. Like the half-life of your knowledge is just insane, right? I know. We have a linear thought line, right? And this exponential stuff is really hard to grasp. It's crazy, right? And so having a dedicated team that they live and they breathe this, available at your fingertips for the startups, that's the second piece. And then the third piece is kind of co-marketing or joint go-to-market, right? And so we have a platform now with some reach, right? And we offer blogs, PR, speak at our conferences, those kind of things, right?

13:50So those are the three things, like free aura credits, dedicated engineers that help you design your application, and then three, joint go-to-market. So from my background, right, which is very corporate now, even though that I was a VC in the bag, right? This, to me, looks like an accelerator, right? So is it a structured, like, six, eight-week program, and then you get some training in that, and then you leave, and then you take a new cohort, or how does it work? No, it's completely self-paced, so to speak. It's like a first-come, first-served kind of thing, right? We do have an approval process, right?

14:29We don't want any random, you know, I claim to be a startup, and I'm just some independent person who don't want to pay for it or something like that. Right. We have a free tier of our cloud service for those people, which of course is capacity constrained and stuff like that. So there's a bar and there's an approval process and then you get into it and there, but there's no programmatic approach. So once you're in it, there's a kickoff call, you get introduced to the dedicated team that you have. Right. And then it's on like reactive, like you reach out to them and those kind of things. So Emil, I know you're only six, eight weeks into the program.

15:11Have you seen anything yet that excites you from in there? Yeah. So we launched it very, very recently. We already have over 300 startups in the program, which is pretty phenomenal. Our goal was to get to a thousand in a year, which would make it one of the largest cohorts of dedicated AI companies on the planet. I now think we're going to get there even faster, which is very cool. And I see all kinds of cool startups in there. And Aaron can give you some anecdotes around that, but I'm very happy so far with the program. But that is great, right? But Emil, you have this ecosystem approach, right?

15:54So, So, you know, what is this, you know, the community as a strategic mode for Neo4j? How do you see that? Yeah, it's interesting, right? Like, so we were always very focused on the developer. And when you and I first met, it was only the developer, right? And that built this big community. I remember even, you know, the pandemic shifted some things, but like even back in 2019, so the last year before the pandemic, right, there were over 500 Neo4j events in all of 2019, right? So that's basically two events per working day, right? And that's meetups and conferences and stuff like that. Now, many of them have shifted to become digital, right?

16:40But the pace has, of course, just increased even since then, right? And this, when I talk to, you know, again, and developers love them. But today I talked to a lot of CIOs. This is one of the key features of Neo4j. That's how they look at it. Oh, I have skill availability, right? Because you built this new weird thing called a graph database. Like, what do you mean? Like, when I went to school, people only talked about relational databases, the SQL databases with tables, right? What is this? The fact that we have the biggest community is a massive, massive feature for them. and I think there will be a lot of the corporate listeners out there that's going to be really eager to see what they can learn from you in this community building because you have scaled your own company.

17:27Now you have your own corporate customers. You mentioned the top 2 ,000 companies now. And then you're going back and you're adding the startup ecosystem, which is what all the 100-year-old European corporates are looking for, right? Are you going to be successful in that, or is the corporate structure going to trouble you a little bit? Yeah, well, I mean, I guess it's TBD, right? I guess what we have going for ourselves is exactly what you just asked about, which is we were always developer-focused. And so when you have a developer-based go-to-market, it truly is focused on a persona, right? So our developer marketing programs, even as our commercial engine, was focused on the global 2000, basically North America and Europe.

18:11Our developer top of funnel was focused just on a persona. That means anyone who is a developer, be they a hobbyist, be they a student, be they in India, Pakistan, Boston, London, right? All the way up to, hey, I'm a 20-year veteran, right? Working at a big bank, but I'm a developer. We targeted everyone, just as long as you're a developer persona, right? And so that means there was always this latent demand of startups. And again, even before we launched the program, we had startups. It's just that there was a lot of friction for them to adopt, right? So I think that's on the plus side. On the minus side, everyone is focused on hitting numbers and revenue growth is what the company's valued on and that kind of things.

19:02And we're not doing this to make money out of the startups. So we will make some of the kind of breakaway ones, but it really is to capture the next generation of applications, AI applications. And that's more of a strategic investment than anything else. And it's just super interesting that you do it now. You don't as such need it now, but you're kind of securing your future. And I think that's an excellent strategic move. In Miu, we are here. The podcast is called EU VC Corporate, right? So there's an EU element in it that I would like to spend the last couple of minutes on here with you, right?

19:38Because Neo4j was born out of Europe. You have scaled globally. It's a rare path in deep tech. So could you share a little bit about, you know, what advantages and challenges do you see for European founders in this new AI platform shift? Yeah, this is interesting, right? Because, of course, when I got started over here building like a deep tech company before that term even existed, developer focused, you know, out of Europe was very, very rare, right? And it just felt that there's obvious to me that I would have to go to Silicon Valley, right? And, you know, I've always been very Silicon Valley oriented.

20:22And I did move over there for six, seven years, right? And built the HQ over there and built the leadership team. moved back before the pandemic, right? For basically for family reasons, but we always kept engineering in Europe. And now we're truly a global company. Most of the leadership team is in Silicon Valley. I'm based here, right? And so we span at least North America and Europe and we have like, I don't know, a hundred people in APAC or something like that. So that's growing fast for us, but it's primarily centered in North America and Europe. What I observe is that it is a world of a difference today.

20:58like the European ecosystem is much stronger, much more robust throughout the entire arc, right? Like the life journey, pre-seed, seed, like even angels, right? Pre-seed, seed, going all the way up, right? And then in AI, we have some of these breakout success stories, right? That are proudly European, right? And that was like, that's even their marketing. And that didn't used to happen, right? And I'm thinking about Mistral on the model side, for example. And the obvious example is Lovable, right out of Stockholm, right? And Paris and Stockholm, respectively, where they make a thing out of it.

21:38Like Anton from Lovable, he says, look, no, I don't want to go to YC. Why would I do that? The opportunity cost is not worth it. I want to stay in Europe and I want to build. And I think that's the best way to build my company, right? That kind of pride we just didn't used to see here before. And Emil, you're deep into it, right? I came from a conference yesterday with a lot of corporate investors. One of the conclusions based on that was that, you know, we would like to see less regulation and please deregulate in Europe. How do you see the future for AI in Europe? Can we compete against, you know, the Chinese and the Americans?

22:16Yeah, it's a great question. So I 100 % am in the we should not regulate more camp. I think the intent is good, but the unintended consequences, the knock-on effects are huge and mostly negative, right? So that's kind of the first one. I think the second one, can we be successful in AI in Europe? I 100 % believe that we can, but our success will vary in the different layers, right? Like we just talked about Mistral. Yes, Mistral is our one shot at this point in terms of being successful in models. And I love the team over there and know them well, but they're not the leader right now in terms of models.

23:02And so can we be successful in such a capital intensive? It's not clear to me that we can. Applications, the infrastructure between the model and the applications, right? where a company like us that we sit, 100 % up for a grab. 100 % we can be successful there. Sounds great. Emil, we're running out of time. Any last comments to the people listening in of the future of AI? Future of AI? Well, there's lots of things to say there. No, but maybe just kind of weaving together a couple of things we talked about. This massive platform shift is a huge opportunity. it's also a huge threat, right? Like for those of us sitting there believing that we have a business to protect, right?

23:50We may not have a business to protect. You were very generous to me when you said that we are doing like strategic wise choice because investing in the future and we don't have to. I don't look at it that way. I look at it as I have to win this next thing because if I don't, whatever I've built to date, I may not have left to protect, right? And so it's a massive opportunity, but we have to move. So that's my kind of final question. Thank you for that, Emil. Thank you for joining the show. Everybody listening in, I hope you'll enjoy it. Reach out to Emil if you're out there building for the future.

24:26Thank you so much, Emil. Awesome, Jeppe. Thanks.

24:33Tear down this wall. It's more than just an alliance. This is a union of values. Let's start acting.