Insights: Neurosymbolic AI Powered by LLMs and Knowledge Graphs

AI is going through a shift – from just generating content to actually making decisions, taking actions, and working alongside humans. This workshop is all about what’s making that shift possible: the combination of Large Language Models (LLMs) and Knowledge Graphs.

We’ll look at how these two technologies complement each other. LLMs are great at understanding and generating natural language, but they struggle with things like accuracy, explainability and domain-specific knowledge. Knowledge graphs, on the other hand, are built on structured, explicit information – perfect for grounding AI in facts, helping it reason, and making it more transparent.

Together, they open up new possibilities for Agentic AI – AI systems that can plan, explain their reasoning, work with various types of tools, and adapt over time. Think: smart assistants that don’t just answer questions, but actually help you get work done, make decisions and collaborate with others.

In this session, we’ll:

• Break down the basics of neuro-symbolic AI (aka combining neural and symbolic approaches)
• Talk about how LLMs and knowledge graphs can enhance each other
• Share what we’re building at metaphacts with metaphactory and metis – real-world examples of agentic AI in action
• Touch on important topics like explainability, ethical AI and why structured knowledge really matters in AI development

This workshop is for anyone curious about the next wave of AI tech – where it’s going, what’s possible, and how to build systems that are actually useful and trustworthy.