Why Attend?

Large Language Models (LLMs) are increasingly used to clean, map, enrich, and migrate enterprise data. However, they don’t inherently understand domain-specific business rules. An invoice may reference a non-existent vendor, a sales order may be loaded before its customer exists, or a legal matter may violate relationship constraints.

These plausible-looking errors can break referential integrity, cause failed imports, and undermine data quality. Prompt engineering alone cannot guarantee correctness, and manual review does not scale.

This session explores Ontology-Constrained LLM Agents, combining formal models of domain entities, relationships, and constraints with automated validation to make AI-driven data curation more reliable. Learn how the Propose → Validate → Repair → Commit workflow supports self-correction, and see a live demonstration comparing unconstrained and ontology-constrained agents on the same ERP-style data curation task.

What Makes This Webinar Different? 

🔹 Live Demonstration: Compare unconstrained and ontology-constrained agents performing the same ERP-style data curation task.

🔹 Practical Agent Blueprint: Explore the Propose → Validate → Repair → Commit workflow, including human-in-the-loop escalation.

🔹 Real-World Applications: Examine data migration challenges involving referential integrity, relationship rules, and data quality.

🔹 Practical Technologies: Understand how RDF/OWL, SHACL, JSON Schema, and structured LLM output support domain-aware data validation.

🔹 Actionable Insights: Learn how to measure constraint-violation rates, automatic repair rates, and human-review rates.