Why
Many GraphRAG discussions stay too vague: they say 'we need a graph' without clarifying which reality the graph represents. This lab makes document, domain, knowledge, conversational, and dynamic graphs distinguishable.
About
System GraphRAG Lab
System GraphRAG Lab is a public learning and architecture project for understanding GraphRAG through five graph types.
The current learning path uses an EU AI Act launch decision to show how user intent, graph context, and LLM synthesis interact.
Who for
For people in product, architecture, compliance, and governance who want to understand how AI answers become explainable and reviewable.
How
Through a simulated learning mode, detailed graph-type pages, a GraphRAG story, and background essays. Live integrations remain optional experiments.
About me
I am Michael Meierhoff, an independent software and systems architect from Hamburg. I have worked in IT for more than 20 years with a focus on system design, complexity reduction, and architecture governance.
My background ranges from software engineering to enterprise architecture in international organizations.
System GraphRAG Lab is built as a working reference. It makes transparent how AI can be embedded into real architectural decisions across business, technology, and organization.
Guiding principle: GraphRAG is not one graph. It is a modelling decision.
Project status
The project is currently being refocused around a simulated EU AI Act learning mode. The goal is to make intermediate GraphRAG steps observable before connecting additional live services.