Every year, the graph technology ecosystem evolves, new tools emerge, categories expand, and organizations across industries continue to explore how connected data can unlock deeper insights.
With the release of the Graph Landscape 2026, Linkurious is once again mapping the key players and trends shaping this fast-moving space. To complement this year’s edition, we interviewed seven leading voices from across the graph community — practitioners, researchers, and industry builders — to share their perspectives on where graph technology is heading, what challenges remain, and what the next wave of adoption might look like.
One of those experts is Amy Hodler, Founder & Executive Director of GraphGeeks. With over 20 years of experience in B2B technology and deep expertise in graph analytics, network science, and responsible AI, Amy has played a key role in helping organizations understand and adopt graph approaches. She is also the co-author of O’Reilly books on Graph Algorithms and Knowledge Graphs, and a recognized community leader in the space.
In this conversation with Matthieu Besozzi, Head of North America at Linkurious, Amy shares her personal journey into graphs, the mission behind GraphGeeks, and her view that we are now entering a “third wave” of graph awareness, defined by hybrid approaches, growing interoperability, and a renewed focus on the graph model itself.
Read the full interview transcript below and watch the complete video:
Matthieu Besozzi
Maybe we can start just by having you introduce yourself. So I remember I knew about you when you were first working at Neo4j. Now you're some kind of the pillar of the graph community. So where are you from? What happened? How did you become this leader in the graph community?
Amy Hodler
Well, thank you. Sometimes I call it the mother node of the community for at least the GraphGeeks community. So I've been in the graph space for, I don't know, I've been saying 10 years for several years. So I don't know, maybe we're at 15 now. But I first got interested in graphs because of network science.
And looking at odd IoT device behavior and didn't really know that I was thinking about network science complexity theory until I had a friend kind of introduce me to some classes on it. And I then kind of pivoted into that area and then had another friend, so it's always about connections, right, that brought me into and said, you know, I think you're really talking about graphs and there's this thing called
You know, graph, in this case it was RDF, so it was a graph compute engine that you might be interested in. I get seriously excited about it. Started working at Cray actually on the Cray graph engine, then got introduced to Neo4j, so very different types of graphs, but could see just the potential and the excitement of users on being able to take something in a totally different data model and do a new type of analysis. And so I got into graphs when I was at Neo4j for I want to say almost five years or five years somewhere around there.
Left Neo4j just to pursue some other areas of interest and realized that very quickly I really love the graph community and wanted to get back to graph and eventually ended up outside of the vendor space started a little community called GraphGeeks about two-ish maybe two and a half years ago. And I wasn't sure if it would take off and it has. And so what that really showed me is that people want, especially graph people, they love connections and they're often the only handful of people at their company actually thinking about graphs seriously.
And so having a space where they can connect and kind of toss around ideas that they have, look at new innovations, struggle, you know, problems they might be struggling with—only another graph practitioner really can help. And so having a vendor-neutral space just was a desire that people had that hadn't been fulfilled yet. That's my graph journey in a nutshell.
Matthieu Besozzi
And so concretely at GraphGeeks, I know there's a Discord channel, there's also events, so what else do you do at GraphGeeks?
Amy Hodler
Yeah, so on GraphGeeks, we have an active Discord channel, which is wonderful. We also do podcasts, on-site interviews, webinars. We do a small amount of training, which I'm hoping will expand as we get more people who are interested in helping out with some training. But we also do a very small amount of advisory as well. So because a couple of us core GraphGeeks members are able to just talk to and I would say hang out with a lot of the graph community, it gives us a broader kind of market landscape view of what's going on.
And so we also help out with companies that either are trying to figure out their graph path, they're getting, they're new into it, they're looking at different technologies, want to see where graphs fit, where graphs don't fit, and want to hear from people who are talking to multiple different vendors and solution providers. So that's in a nutshell what we do. We also have a newsletter. We occasionally do in-person events, which I'm hoping to do more of in 2026 as well.
Matthieu Besozzi
So previously you said something interesting. You said that you created GraphGeeks, you created this community because you realize in a given company, you have just a handful of people who are experts in graphs or interested in graphs. So you and I are both convinced that this is a wonderful technology and that there is a need for more graphs to solve the issues of any companies. How do we make that happen, you know, to expand the awareness of graphs?
Amy Hodler
Yeah, I think that's already happening. But I think the more you can, get out of our own bubble. So if you think, okay, I'm gonna go graphiania. If you think about betweenness centrality, if you think about what that is really about, it's the bridges to other communities. So I think, and that's about being an ambassador to other communities that we're not currently prevalent in.
So if you're an ontologist, you know, bridge to people who aren't familiar with that term. Go ahead and kind of speak in the terms where people are right now. One of my favorite events right now—I love the Knowledge Graph Conference. I love the Connected Data London Conference, but I also really like the Open Data Science Conference. I also really like some other conferences and part of that is it helps me bridge to people who maybe are thinking about connections and relationships, but graph isn't a terminology they're familiar with.
So I think that's a big part of making the awareness and also doing education that slips into other technologies that they're already doing. Somebody's already thinking about entity resolution. My goodness. I mean, that's a graph problem and graphs are really prevalent. So, speaking to people who are doing things in that area. Same, neuro symbolic. Call it neuro symbolic. I could call it, you know, graph plus probabilistic. We can speak in those other terms.
So I think that's probably the best thing we can do as practitioners is to be ambassadors outside of our own bubble. It's so much fun to be in our own bubble. I love the graph community, but getting out there as well and bringing other people in is important.
And I think we're starting to see that change. I was, actually, thinking about this this morning and I wanted to share it with you. I think we're actually on, I would call it the third wave of graph awareness and adoption right now, or we're tipping into it. And I think that is going to see a proliferation of a variety of graph technologies out there in the community.
Matthieu Besozzi
That's interesting. So can you please describe this third wave of graphs? What was the first one, the second one, and what is the third one?
Amy Hodler
This was something I've been thinking about and I'm not the only one to ever say that we're in a new wave. But I was thinking about it and in particular this morning. You know, as you are just waking up, these ideas come to you, good or bad, but it occurred to me, we had the wave that was all about knowledge—how do you capture knowledge appropriately and how do you transfer it?
That I would say is semantic web. You know, we know our history, right? So that I think where you have knowledge managers, ontologists, people who are thinking about things like, do I appropriately describe things so that they can be shared? The next wave I would say is more around transactions, processing, and the IT organization. How do I take this capture of knowledge and put it within enterprises, within businesses.
And that clearly had a strong growth in property graphs: easy to use, more familiar to computer scientists, and focused maybe a lot around developers. You know, we're going to build some things with this. And I think we're tipping into this third wave. I'm not sure what to call it, but to me, the focus there is around graphs, the graph model itself: the model of having nodes and relationships, the model of topology being important. So not a particular technology, but the actual model of the graph itself being important. I think the hallmark of that wave is variety. There are a lot of graph technologies coming out, solutions, and their growth in different areas.
And that I think is the hallmark of this new—actually two hallmarks. One would be a variety of, let's call it the graph model age. And the other one is hybrid. So people are combining and using things together where it used to be much more siloed. And so that to me feels like this third wave. And I think there's some things we can see that to me says we're in a new wave or a new era, which is we're seeing the gap between a variety of things shrinking. Like the gap between RDF and property graph is starting to shrink, both combining it. So we have companies that are using both. Why not? They're good for different things.
The Graph Data Council has a Lex project, which is fascinating to me, which is taking some of the principles of RDF and bringing it to property graphs, strong types, cardinality, potential for inference. It's still in a working group stage, but it’s shrinking that gap and culturally, we now can talk to each other. Property graph people and RDF people are talking together. And then the other thing that to me says we're tipping into this new age is just the frothiness of the market, acquisitions, fundings, fizzles, forks of code, and there's just all this kind of frothiness as well. So I think it's an exciting time to be in graphs.
Matthieu Besozzi
Do you think that one day we will see an RDF guy speaking or having a drink with a property graph guy at the same table?
Amy Hodler
I think it's already happening. They just might not fully—I don't think the guard is fully down, but it is already happening. And I actually was at a conference a couple of weeks ago and I asked just a regular attendant what really struck them. And they said, my goodness, the RDF and the property graph people are actually talking to each other.
And nicely, and getting along and sharing learnings. And it just felt like a hallelujah moment for me. It's like, okay, we all love graphs. We all care about knowledge and sharing. And it's good to see that interaction start to happen.
Matthieu Besozzi
Yeah, I appreciate it. You know, at Linkurious, one of our missions is to make graph accessible to the business users, to people who are non-technical. I have the feeling that we are not 100% there yet. What do you think about that?
Amy Hodler
No, we're not, but we're moving in that direction. And thinking about the kind of this third wave analogy, I think there's a couple things that are pushing us into that. AI, of course, and I say of course, because that's a pushing function of everything I think in the last year it feels like. But I think the need to take a look at both how you interact with the machines with AI writing for AI and modeling for AI and modeling for humans business is really clear.
AI suffers without context, it suffers without business intelligence and what is business intelligence other than meaning, semantics, business rules. I will say the word rules because you can model rules in a graph, you can be deterministic. And so I think bringing those things together, we're not there yet, but AI is kind of pushing us to focus on that intersection between machines and humans or business. And the more we push in that, I think that kind of brings us into a third wave graph, because pushing into that text to cipher entity extraction resolution, how do you construct graphs more easily?
That helps us bring those two together, but we're not there yet. In fact, I do think one of the things that I'm looking forward to the most in the next 12 months, hopefully sooner, but in the next 12 months, I would say is ontology and schema, making it easier to get our data into a graph, but also model because I do think this new era is about the graph model. So then modeling becomes your primary importance. And if that's your primary importance, it has to be easy. It has to be easy to develop your schema, your ontology, your semantics, all of that, how you define it.
And I think there are some new tools coming out and some approaches that are going to make that much easier. And so that's my number one wish list for the next year: let's make the modeling and the graph as easy as possible. AI can help a little bit. It's still not there. Humans are still and always will be important. So how do we make that human interaction more important?
Matthieu Besozzi
Interesting, yeah. So going back to the Knowledge Graph conference in May, we had this conversation about the shift in the graph technology, in the graph industry. What would you say happened in the last 12 months in the industry?
Amy Hodler
I think again, AI has really pushed us and I think it's shown the need for context, more than just being bigger, just more data, more than bigger graphs even. And we've seen that with RAG, then it'd be using graph for graph RAG, but graph RAG really isn't graph RAG, it's hybrid RAG.
And so again, I think this trend of hybridization, combining approaches is really quite interesting. So we've seen that neuro-symbolic probabilistic, but deterministic, a lot of interest there, maybe not as much application as I'd like to see. And then I think also this shift for, I would say, let it. Are AI models having less knowledge themselves? So the models are being more representative and reaching out to data itself.
So reaching out to your knowledge, reaching out to your SME, reaching out to things that change more frequently. I think that's been a shift because I think previously, if we think about 12 months ago or 18 months ago, we were all trying to push knowledge into systems so that you could then compute it as a system.
And I think what we're seeing in small language models in, you know, looking at AI models as functions versus containing all the data, containing all the knowledge itself—it's kind of a nuanced shift, but it has a lot of implication because then we can hold knowledge separately from how it's computed is as well and go back and forth from that.
And I think in the last year the biggest change, though, is everything's gotten easier. Every time you turn around it's like, well that was so much easier. We were talking right before we started this. You know a particular tool we're using like my gosh it's gotten so much better in the last six months. Everything's getting easier and to your point we need to make it easier for the business to adopt. It is continuing to get easier.
Whether we're talking about how you use natural language with graphs or we're talking, you know, that has just been a game changer. I actually had somebody say that language models are—or not language models, excuse me—that query languages don't matter anymore. Now, I don't agree with that, but their point was if we are going to put natural language in front of everything, the interface with the human in the way they're used to, the visibility of the query language kind of goes under the hood.
And so I think we're moving a lot of complexity under the hood as well. And you're seeing that in a lot of IDEs (Integrated Development Environment), visualizations, now, NLP is as well. So anyhow, it was a little bit of a ramble, but those are some of the changes I’ve seen in the last year or so.
Matthieu Besozzi
Yeah, I agree. I remember five years ago, a lot of clients were asking for this kind of natural language querying tool and technically it was just impossible.
So we tried developing a work around, but nothing was enough. And the LLMs are actually very good at writing queries. So it's really been a game changer allowing people to actually speak, discuss in natural language with graph databases. We see that every day at Linkurious.
If tomorrow you have this magic power, what would be the tool or the capability that you would create for the graph industry?
Amy Hodler
I am going to go back to—well two things. One, I will go back to the ontology slash schema. If I had a magic tool, I would say: somehow to work with the SMEs at companies and be able to transfer their specialized knowledge about what their data actually means into a framework that graphs can use that is still difficult.
And if you have all your data clean, well-defined from a single source, maybe you have all this already. But in every client I've worked with, that is usually the very first stumbling block. I've worked with a client where capitalization—they actually use that to denote meaning in one of their data sets. That's a valid choice, I suppose. But, as data proliferates and more people use it, the definition and the enforcement of that meaning erodes.
There's a perishability to things when you have humans being the only ones that know. So that would be my, if I had a magic wand, that would be my magic wand: How do we go from subject matter experts who know their data really well to actually having definitions and semantics that we can put into a formalized ontology and or schema. I'm all about starting simple.
So if you start with a schema and then develop a more formal ontology, that's completely fine as well, but just to be on that road. And I would say the successful teams are the ones that can get into a room and argue about the meaning of something and come out at the end of the day on what the meaning is going to be.
They're going to move really fast and they do really well. If they can't do that, either politically or they don't have that knowledge, then they tend to struggle quite a bit. So that would probably be my magic tool. You know it's just getting people started easier.
Matthieu Besozzi
One last question. What advice, what recommendation would you make to somebody who wants to start in the graph today?
Amy Hodler
I would say to start where you're at. There are a lot of tools for different platforms. And there's a lot of—there are classes, there's training, there's a lot of content. You can join communities (shameless plug), like GraphGeeks, where you can ask people different opinions about different technologies. But I would say look at the environment you're in now.
If you're a heavy AWS user, look at what resources they have. Look at what Neptune has. Or look at something that can query across your data stack. So, Puppygraph is an example. If you are heavily into transactional, there's some really good resources out there on more transactional graphs.
Neo4j has a ton of really great resources as well. They probably have the most extensive for people that are more on that developer community side. If you're in academia, there are a lot of great academic resources as well that we probably don't think about as vendors. So I would say my recommendation is to look around you and to kind of reach out from there.
And to think about the data model itself before you think about the technology. So what's the problem I'm trying to solve? What's the business problem I'm trying to solve? And what does that imply about the kind of data model that I need? Maybe it's a graph, maybe it's not. And then from there, looking at your own stack and saying what technology fits the stack and marries up with the problems.
Matthieu Besozzi
Thank you, Amy.
Amy Hodler
Yes, thank you. Matthieu, there is one other thing I do want to tell people about, and it's something I've been hearing lately that I think is going to be important in this third wave of models being important, the data model itself. And that's something that I think there's going to be interesting growth in different areas in graphs. But I think it's important that people think about meaning and semantics.
Call it what you want. Your graph model as a moat both competitively and with AI. And that's something that I just really want to convey is don't give away your subject matter expertise. Don't give away the meaning that is so integral to your business and what's competitive with your business. Keep that as your moat.
Keep that as what makes you special and I would say embrace that into a semantic environment. I don't know if I like the term semantic layer, but anyhow, keep that as an area that is important in either a knowledge graph layer or something of that nature.
Matthieu Besozzi
Alright, well noted.
Thank you, Amy.
Amy Hodler
Alright, thank you, Matthieu.
A spotlight on graph technology directly in your inbox.
