Ontologies Are Back: How AI Agents and Neo4j Are Reviving the Semantic Web

In 2026, AI agent engineers are returning en masse to ontologies, the structured knowledge graphs of the Semantic Web era, to keep probabilistic language models inside deterministic logical boundaries. The approach of combining neural networks with symbolic rules is called neurosymbolic AI.

Why Agents Need Ontologies

Ontologies give an agent hard boundaries within which it cannot drift away from the facts. They describe data as a graph: entity types, their properties, and the relationships between them in a specific domain. A large language model is strong at probabilistic reasoning, but that same probabilistic nature makes an agent unpredictable. An ontology acts as a logical constraint, keeping the model inside a defined set of rules.

There is no need to build such schemas from scratch, because the web has already accumulated mature standards.

• "Neurosymbolic AI" is a pairing of neural networks with symbolic, rule-based systems
• Ready-made web ontologies: Schema.org, FOAF, Dublin Core, RDFS, and OWL (Web Ontology Language)
• Emil Eifrem (CEO of Neo4j) identifies three types of ontologies for agents: business, technical, and execution traces
• OpenLink Software is building an "engineering stack for agents" on Semantic Web technologies
• The topic came up at AI Engineer World's Fair 2026; the talk by Frank Coyle of UC Berkeley ran for 20 minutes

Neo4j's Three Layers of Knowledge

Emil Eifrem, CEO of the graph database company Neo4j, breaks agent ontologies down into three levels. The first is the business ontology, which humans can read and understand. The second is the technical ontology, meaning enterprise metadata. The third is execution traces, the signals an agent leaves behind while it works.

Neo4j is building ontologies directly into its agent products. Its competitor OpenLink Software, led by Kingsley Idehen, is assembling a full stack for agent engineering on Semantic Web technologies.

Why the Semantic Web Failed the First Time

The first attempts to build a semantic web in the 1990s and 2000s failed mainly because ontologies were too expensive to maintain by hand. Standards such as RDFS and OWL had existed for decades, but people had to describe and update knowledge schemas manually, and they couldn't keep up.

AI developer Prasenjit Sarkar proposes a solution that wasn't available twenty years ago: let the agent maintain the ontologies itself, updating them as it works. What once killed the Semantic Web may now save it.

"The beauty of an ontology is that it defines the types of entities and relationships through which language gains computable context," said

Kingsley Idehen, founder of OpenLink Software.

What This Means

Ontologies are not a step back into the past but the missing control layer for agents. The more companies trust language models with real actions, the greater the demand for deterministic boundaries that stop a probabilistic model from making costly mistakes.

Frequently Asked Questions

What is an ontology in simple terms?

An ontology is a formal knowledge schema in the form of a graph. It lists the types of objects in a domain, their properties, and the links between them. For an AI agent, it is a map of permissible facts and relationships that the agent does not step outside of.

What is neurosymbolic AI?

Neurosymbolic AI is the combination of neural networks with symbolic, rule-based systems. The neural network handles probabilistic language processing, while the symbolic ontology provides deterministic logical constraints.

Why didn't the Semantic Web take off in the 2000s?

According to participants in the discussion, the main reason was the cost of maintaining ontologies by hand: the RDFS and OWL standards existed, but people had to update the knowledge schemas themselves. The 2026 idea is to entrust that maintenance to the agents themselves.