Our annual Graph Technology Landscape maps what the graph and data ecosystem looks like today. Alongside it, we wanted to understand how it got there and where it's heading next, so we sat down with the practitioners, strategists, and builders shaping that shift firsthand.
Among the experts we discussed with is Marko Budiselić, CTO and Co-founder of Memgraph, an in-memory, real-time graph database engine. Marko and his co-founder Dominik Tomicevic started Memgraph roughly ten years ago with a mission to bring real-time graph analytics to the market.
In this chat with Giulia Rotondo, Head of Sales at Linkurious, Marko explains what the biggest barriers to graph adoption are, what Memgraph Zero's "zero ETL" approach means for querying data that can never be moved, and why GQL is set to become the natural language for that shift. He also shares why a centralized semantic layer, with a feedback loop agents can update as they learn, is becoming essential infrastructure for agentic AI.
Read the full interview transcript below and watch the full video to listen to the whole discussion.
Giulia Rotondo
We're going to talk about AI, we're going to talk about Memgraph Zero, we're going to talk about trends. But first and foremost, let's talk about you. Can you introduce yourself for our audience? And also, I'd be very curious to hear you talk about your own experience with graph, and how you got to work in the world of graph.
Marko Budiselić
I'm Marko, CTO and co-founder at Memgraph. Basically, Dominik and I, my other co-founder, started Memgraph about ten years ago. Just to introduce it briefly: Memgraph is an in-memory, real-time graph database engine. We wanted to bring the real-time graph analytics to the market, and roughly ten years ago we started building the product.
Giulia Rotondo
So, this is part of a series of interviews we've done with other graph experts, and almost all of them have highlighted an emerging trend, of the barrier to graph getting lower. Would you agree with that? Is that something you've observed yourself?
Marko Budiselić
I would say so. I mean, I think the biggest barrier to entry about graphs is the first part, the education. Because people are mostly not aware of it if they are not familiar from, like, the educational system.
And with the whole recent trends, especially with AI, it's much easier to generate some code, or AI will actually pick the graphs as a natural solution. Still a lot of education is needed to actually get started, and then take it from there. Once there is some software or something that does some complex task, it actually takes some time to understand the stuff and figure out what actually to do next.
Giulia Rotondo
What are some of the most exciting things happening in graph today?
Marko Budiselić
There's multiple projects going on, and I guess the most exciting thing is the diversity of tools and all the new things that are happening under AI.
Because, that's maybe a bit overstatement, but with AI graphs are almost the only way to model data and go about it. And then, a lot of those tools started incorporating graphs and that way of thinking, like linking data, building those networks, and running some analytics on top of that. And that's, I think, super exciting because it's getting much more closer to the user and much more easier than before.
Giulia Rotondo
And since you're speaking about AI, what role do you see graph playing in AI today, but also moving forward?
Marko Budiselić
It's both things, like solving a particular problem, and usually graphs are there to do some analytics. Historically, it was all about some analytics, maybe the most famous example is Google's Page Rank and such algorithms. It's very detailed and very specific.
But, more and more, it's also about connecting different data sources and bringing all the data together, and making those connections. And breaking those data silos, and having data on multiple places, and then how to combine them. That's, I think, definitely emerging recently.
So, that's where graphs also can massively help. It's usually not one big knowledge graph, it's a lot of small ones. But they have to be connected and wired up somehow.
Giulia Rotondo
Now let's talk about Memgraph Zero.
Marko Budiselić
That's actually a perfect segue, because that's one of the reasons why we started Memgraph Zero. That's actually a bigger umbrella under the Zero ETL move, because of all these data, there is a lot of large data sets which can't be moved in any system.
The idea was "let's just figure out a zero ETL solution to execute the graphs", because, again, graph data model is very convenient for modeling and for data access. But the challenge is: how do you access a petabyte of data and do that efficiently?
Then, there is also the GQL aspect because GQL is getting the standard declarative graph query language. And it feels very natural to actually build such zero ETL layer with GQL, because the assumption is that over years that will grow and expand and become much bigger. And plus, it's actual ISO standard, so people will follow that, because it's actual very well thought and designed and has all the stuff built in.
And then, one big thing is also the semantic layer for agents, because agents also want centralized data access and having data or at least illusion that data is at one place. That's very convenient for agents, because they can only have one endpoint and do all data access through that one endpoint.
And also, it's not just about data access. It's also building the, let's say, I call it "semantic layer". But an important aspect is the feedback loop, so, establishing the feedback loop where agents can incrementally update new knowledge as they figure it out. Because in large systems there is no one person or one agent that knows everything, that's a very iterative process to figure things out as you go. And then, if the feedback loop is centralized, then agents have a much bigger chance of success and figure out the right schema or the right knowledge graph in the end.
Giulia Rotondo
Indeed, it sounds like you chose the perfect time to launch a product like Memgraph Zero. We see a lot of interest for a product like this in the market, and we're very excited about it as well. Congratulations.
Giulia Rotondo
One last question: we have talked about the great things happening in graph right now, but what do we still have to do? What challenges do you think the graph industry still has to solve to achieve wider adoption?
Marko Budiselić
In terms of the challenges, I guess it's very similar as 10 years ago. I think the main thing is education, because people are just not aware of the concept and especially not all the capabilities that are available.
And then what Memgraph focuses on a lot are those three pillars that we have: ease of use, speed and scale, because those three things are usually not very well supported. It's either too hard to get started, you need some education, and that's super hard for people in general. But the main job is to actually get that much easier, much faster, much more scalable.
Giulia Rotondo
Thank you, Marko. It was a pleasure having you.
Marko Budiselić
Thank you.
A spotlight on graph technology directly in your inbox.
