Turn Documents into Graph Context: Unstructured2Graph + Memgraph AI Toolkit

Most enterprise knowledge lives in PDFs, DOCX, and TXT and not in a graph. Unstructured2Graph changes that. In this live, technical session we’ll (re)introduce the Memgraph AI Toolkit, explain the real bottlenecks behind building Knowledge Graphs for GraphRAG from unstructured data and showcase it live. You’ll see how Unstructured2Graph ingests documents, extracts entities (via LightRAG integration), links them with page-level provenance, and writes a connected, queryable graph into Memgraph ready for retrieval.


What’s in the Webinar?

• Why extracting enteprise knowledge trapped in unstructure documents is important
• Live Demo: how Unstructured2Graph ingests documents
• How to query the resulting graph in Memgraph

What You’ll Learn

• Why GraphRAG needs connected structure (not just embeddings) for explainability and multi-hop answers
• How Unstructured2Graph turns PDFs/DOCX/TXT into nodes, relationships, and attributes with preserved provenance
• Where LightRAG fits for LLM-based entity extraction
• Practical guardrails: handling messy layouts, duplicate entities, ambiguous mentions, and update cycles