Building AI Agents with Symbolic World Models

Hack Session

About the session

In this live hack session, we will build an ontology-grounded AI agent and use the ontology as the symbolic backbone of its world model. Starting with a conventional prompt-driven agent, we will progressively enable it to ground ambiguous requests, traverse domain relationships, select compatible tools, simulate candidate state transitions, and validate proposed actions before execution.
Built entirely with open-source technologies such as RDF, OWL, SHACL, RDFLib, pySHACL, LangGraph, and a locally hosted language model, the session will give attendees a reusable architecture for moving beyond agents that merely generate plausible actions toward agents that can explain what they understand, why an action is valid, and what is expected to happen next.

Speaker

Download Brochure