Build Daily
Tinley Park · August 23, 2026Making Knowledge Explicit: Jesús Barrasa on Ontologies, Graphs, and AI
Jesús Barrasa discusses his breakout session on ontologies at the Graph Gathering in San Francisco, emphasizing how making knowledge explicit improves AI and LLM performance. He also promotes a new O'Reilly book he co-authored on GraphRAG, which covers knowledge graphs, retrieval strategies, and practical examples.
The outline
Tap a timestamp to jump there in the video.
Graph Gathering Overview
Barrasa introduces the San Francisco event's focus on agents and AI in enterprise production.
Ontology Breakout Session
He describes leading a session that aligned participants on the overloaded term 'ontology' and its value in making knowledge explicit.
Using Explicit Knowledge
The discussion covered how to apply explicit knowledge through graph construction, retrieval improvements, and LLM combinations.
GraphRAG Book Announcement
Barrasa shares his work with Stephen Chin and Michael Hunger on a definitive O'Reilly guide for GraphRAG.
Book Content Details
He outlines the book's practitioner-focused approach to knowledge graphs, retrieval strategies, evals, and end-to-end examples.
Graphs in AI Applications
The event highlighted how graphs are central to modern AI applications by efficiently grounding LLMs.
Broader Ontology Understanding
Barrasa reflects on learning that the community views ontologies more broadly than just formal graph definitions.
Knowledge Representation Value
He concludes that making hidden knowledge explicit is the core theme of ontology, regardless of its specific materialized form.
Formalizing domain concepts before building AI systems can reduce ambiguity in agent behavior. For solo builders shipping products, making assumptions explicit may be a cheaper path to reliable outputs than adding more model complexity.