Core Concepts
The technical foundations of how Moongraph processes and structures your data.
Core Concepts
Understanding these concepts helps you get more out of Moongraph and troubleshoot when things don't work as expected.
These pages explain why things work the way they do—not step-by-step instructions. For task-oriented guides, see How-To Guides.
Overview
Moongraph transforms documents into structured, queryable knowledge. Here's what happens:
- Documents are chunked — Split into smaller segments for search and retrieval
- Chunks are embedded — Converted to vectors that capture semantic meaning
- Entities are extracted — People, organizations, locations, and other "things" are identified
- Relationships are found — Connections between entities are extracted
- Duplicates are resolved — Entity resolution merges different mentions of the same thing
- A graph is built — Entities and relationships form a connected network
The Agent uses this structure to answer questions with citations, drawing from both document chunks and graph relationships.
Concept guides
Knowledge Graphs
When and why to build graphs
Entities & Resolution
How entity extraction and merging work
Document Processing
How documents are parsed, chunked, and embedded
Structured Extraction
How schemas define what data to extract
RAG & Retrieval
How the Agent finds and uses context
Credits & Usage
How the credit system meters AI features