Alongside the publication of this year’s Graph Technology Landscape, we interviewed seven leading voices from across the graph and data community. These practitioners, strategists, and industry leaders discussed where graph technology stands today, and what the next phase of adoption will require.
Among the experts we were fortunate to discuss with is Joe Hilger, COO and Co-founder of Enterprise Knowledge (EK), a consulting firm that helps organizations capture, manage, present, and leverage their information. Operating globally, EK provides practical solutions to ensure organizations’ knowledge, information, and data can be found, used, and reused.
In this chat with Matthieu Besozzi, Head of North America at Linkurious, Joe explains why the semantic layer—combining metadata, taxonomies, and knowledge graphs—has become the critical foundation for AI success at large organizations, and why so many enterprise AI projects fail without it. He also shares why 2026 may prove to be a pivotal year for graph technology, and why the companies that invest in getting control over their information now will be best positioned to make agentic AI actually work.
Read the full interview transcript below and watch the full video to listen to the whole discussion.
Matthieu Besozzi
To start this discussion, some people actually don't know you. So can you please introduce yourself and also introduce Enterprise Knowledge?
Joe Hilger
For those that don't know me, my name is Joe Hilger. I am one of the co-founders of Enterprise Knowledge. I founded the company about 12 years ago with Zach Wahl and since then have watched the company grow from two people dreaming of having an international consultancy to a firm that does nearly $20 million in business on nearly every continent in the world.
So that's me. I also co-wrote with Zach Wahl Making Knowledge Management Clickable, and have just wrapped up another book with Zach Wahl and Lulit Tesfaye on semantic layers, which will be coming out from Springer Publishing in the next two to three months.
That's a little bit of my history. And then, about Enterprise Knowledge: We are the largest dedicated knowledge management consultancy in the world. But what I really would stress is when we founded the company, knowledge management—which is for us getting the right people, the right information at the right time—was as much about technology as it was about people in process. And as we started to look for ways to make information more findable and understandable, by 2015 we were starting to use graphs and knowledge graphs. And of course, now we've moved pretty heavily into AI in conjunction with graphs and a lot of what we call the semantic layer to organize information.
Matthieu Besozzi
So what you're saying is that knowledge management and graph technology and obviously AI today are inseparable, right?
Joe Hilger
I think they're critical, especially at large organizations. And it's interesting, what started as a way to organize information and make it more findable has become one of the key tools in making internal AI initiatives work. So yes, they're pretty much hand in hand, and super important.
Matthieu Besozzi
If you want to remember one main thing about 2025, what would it be—for the technology and also for your company?
Joe Hilger
I think 2025 has been a learning year for organizations about how to work with AI. Everyone picked up ChatGPT and in the outside world just said, “Oh, wow, look at the amazing things this does.”
Organizations have started to get smarter about it. Everyone quotes the MIT study where 5% of these proof of concepts actually ever work and move into production or are considered successful.
So this to me was for large organizations the year of learning and I think people are getting a lot smarter as we've come to the end of the year. We’re seeing requests for ways to manage your content ahead of the AI. The AI readiness story is starting to become really important.
For us at EK, it was funny. Two years ago, we bet heavily on the semantic layer. We said this idea of a central layer that maps information across the enterprise using metadata, taxonomies, and knowledge graphs, and looking at both data—structured data and unstructured data—we called that the semantic layer, and we said it was going to be big.
As organizations have learned, the realization of the semantic layer is a critical component for any large organization to get a handle on their information. For us, it was a nice win and nice to see that happen.
Matthieu Besozzi
And so, do you believe that the methodology that was pushed and developed by Enterprise Knowledge is a key for the success of a GenAI project within a big organization?
Joe Hilger
I will go bold here. I don't just believe it. I know it. And I know it because half the projects we're doing the second half of this year are projects where someone tried to do it the other way and failed. And then turned around and said, we have to do this a different way. And then they called us, and guess what? It's working.
The understanding that you need a knowledge graph to map things across your organization, that you have to do cleanup of old and outdated content. Actually, one of the things we look for is near duplicate content. That information that is versions of a document where the old versions are inaccurate, right? Fixing that. And then getting metadata to just provide more context for the LLMs.
All of that work is critical to make these LLMs successful. And we're having great success stories now with clients because we're doing things the right way, especially when we pull things together with graphs.
Matthieu Besozzi
Okay, so 2025: a year of learning. What do you expect from 2026?
Joe Hilger
You know, it's interesting, we've been chatting about that internally among the leadership team and there's some worry about the economy, which I think we're all paying attention to. But companies have made big investments in AI and they're not yet getting the returns. And I think what we're going to see is organizations are going to be much smarter about how they approach this.
Last week, I was in Australia and I met with three of the largest banks there. All of them were talking about a semantic layer. Not all of them had heard of us, so it wasn't just our doing. They weren't from our little knowledge management space. They had just realized that to manage their data, to manage their content, they need a semantic layer… which includes as a key component of it a graph that maps information across the enterprise.
So seeing that companies are starting to get it, I think there's going to be real investment in getting better control over your information so that when they do the LLMs, particularly the agentic AI piece (because I think that's really the piece that will take off next year) that they're more successful.
I would say I think we're going to have a smarter group and we're going to have a focus on agents that create more meaningful solutions.
Matthieu Besozzi
Okay, and if I say we are nearing the golden age of graph technology, how would you react?
Joe Hilger
Wow. So look, they've hit the big time when people are using it not just for search and not just for kind of the old-school stuff like that. Knowledge graph's got popular again because of Google and Google's knowledge panels. But this is different. So yes, I think it's the golden age.
I think it's going to be interesting to watch, because I had a client challenge me the other day. It's a very large organization that had spoken to some other consulting firms as well. And they said, “We feel like we need a graph, but we talked to a few other consultancies who said, they did it for some places and it didn't work.” They tried to go too big, too broad. They didn't have a focus.
So I think graph is in the spotlight. In that sense, it's in the golden age. I think there's going to be a lot of learning of how to do semantic layer and truly enterprise-scale knowledge graphs the right way this year. If that’s the golden age people are talking about, then absolutely yes. If it's the golden age of people who are going to do it well, I think we're going to see a mixed bag. There will be learning there.
Matthieu Besozzi
I definitely agree with you. Another question is, what's in it here for graph visualization and also for graph investigation?
Joe Hilger
You know, that to me is the fun part that we tell our clients. They're diving into semantic layers because Gartner told them to. Honestly, I think that's often a driver. They're finding out that AI requires a graph.
We see all these data people that never use Knowledge Graphs. They say, “Now I need a graph.” And the first thing we tell them is, “You need it to do your GenAI. You do. But if that's all you do with it, you're wasting your time. And you need it to do more.”
That type of navigation through something that's so natural as a graph in the way people think and act is huge. I think a byproduct of everyone trying to solve their generative AI failures is going to be implementing a graph. And then they're going to say, well, how do I see what's in this? And then after that, they're going to say, what more can I do with it? And for me—you all are going to be in a wonderful place. I think you should have a great run next year of people saying, now I've got this graph. How do I see it? How do I take advantage of it? Other than just asking questions to a chatbot.
So, absolutely. I think you will be the surprise package that adds great value to what a lot of people have started to build for maybe not the initial reason of having a graph.
Matthieu Besozzi
I hope you're right! So you're opening offices all around the world now, so I've heard about Europe. Now I see you posting on LinkedIn about hiking trails in Australia. So what's next now?
Joe Hilger
I don't know that we've yet committed to opening an office in Australia, but as the semantic layer has caught on, as graphs have caught on, companies are reaching out to us saying we'd like to work with you, but we want you closer.
And so Zach and Lulit and I have been figuring out how to do that and really how to replicate the EK of the US elsewhere. The good news is we've made big investments in training. We have Camp EK, which is a six-month long process of learning EK and how we do things. We have cohorts, we have communities of practice that people engage in.
Because what's most important for us is to be closer to our clients, but we want them to get the same level of service and expertise that they've got from home base. As we do this, we’ve figured out the financial and other logistics, but now the most important thing is making sure we have the right people. So we have some tools in place, and we're going to see how that goes.
We will definitely open at least one new office this year and maybe a second as well. And the second might in fact be in Australia.
Matthieu Besozzi
That’s exciting!
Joe Hilger
Yeah, neat to see us again grow from two guys in a kitchen or dining room. We had a real firm and have moved to something that's actually now starting to sprout branch offices.
Matthieu Besozzi
All right. So thank you, Joe.
A spotlight on graph technology directly in your inbox.
