Noz Urbina - Panelist at Semantic Layer Symposium 2026

Yeah.

You know, talk about the mainstream.

You know, if you're a senior data leader right now and listening to all these conversations around semantic layers, probably for the hundredth time this year, I think what a lot of people are probably thinking, okay, what is really the actual, what is real, what it is, and what is the actual real problem it's solving? So kind of bringing back to that point, we have many leaders now that are coming to us and saying, you know, I need a semantic layer.

For those who are really contemplating this decision, I really empathize, right? There's a lot of noise on the market right now.

So before they buy anything or launch a major initiative, what would you recommend they ask?

Oof, you're stimulating the purist in me.

So I'm seeing a lot of people throw around the words semantic and knowledge graph very willy-nilly.

I had someone from, I'm not going to name the business, but a

household name.

globally known household name.

And I gave a presentation on the differences between an RDF knowledge graph, an LPG knowledge graph, a traditional database.

And sorry, you know, it's like when you're naming the seven dwarfs.

The embeddings, the statistical embedding systems kind of databases.

And he said I was the first one who didn't

say that I could just throw a particular type of database over my data, have it do its similarity analysis, and that was the semantic layer job done.

So I think that with excitement comes hype, with hype comes confusion, and then you have to actually understand what is this system capable of doing and what is the real work I need to do to get the most out of it.

So I think that the AI rush has come with a lot of snake oil salesmen.

A lot of magic wands are being waved around, so to speak.

And I think that really understanding the scalability, the portability, the uh the exit plan around technologies you know is this is this based on standards can i understand what's happening inside it can i export it to another system in five years the temptation is just to buy these magic wands and throw them in front of your people and they will you know magic will happen but i think that we're also looking down the barrel of the biggest potential vendor lock-in that we've ever seen.

Because if we're using tools like AI and semantic technologies without an interpretable, queryable, human-understandable backend, you can never leave.

You have a system that gives your team superpowers.

You don't know how it works.

You don't understand what it's actually doing with your data.

So I think things that support formal ontologies, RDF standards and interoperability, and that mix semantics and the chat GPT unstructured sense that people call that semantic understanding and inference.

And then the kind of work that we've been doing that supports that, which is symbolic AI and inference that is real reasoning over hard structured knowledge.