Concepts

These pages describe the backend systems that turn source books into stored chunks, vectors, extracted graph candidates, manifested graph data, and retrieval-ready context.

World Ingestion Pipeline

• World Ingestion Pipeline
• World Storage
• Text Splitting
• Vector Storage And Chunk Embeddings
• Knowledge Graph Extraction Pipeline
• Graph Manifestation

Storage Layers

• Storage Layers
• Qdrant Vector Store
• Neo4j Graph Store

Shared Backend Systems

• Shared Backend Systems
• Provider Key Scheduler
• Model Registry

Retrieval

• Retrieval