Back to Learn
Intermediate

Document Store

A specialized storage layer in a RAG pipeline that persists the raw text, structured metadata, and document fragments, acting as the authoritative source for context injection. It maps unique IDs from a vector index back to the human-readable content required by an LLM to generate a response.

Definition

A specialized storage layer in a RAG pipeline that persists the raw text, structured metadata, and document fragments, acting as the authoritative source for context injection. It maps unique IDs from a vector index back to the human-readable content required by an LLM to generate a response.

Disambiguation

Not a generic NoSQL database; it is the 'source of truth' for retrieval-augmented generation, often decoupled from the vector index for scale.

Visual Metaphor

"A warehouse of full-text manuscripts where each page is assigned a barcode that corresponds to a digital coordinate in a separate GPS system."

Conceptual Overview

A specialized storage layer in a RAG pipeline that persists the raw text, structured metadata, and document fragments, acting as the authoritative source for context injection. It maps unique IDs from a vector index back to the human-readable content required by an LLM to generate a response.

Disambiguation

Not a generic NoSQL database; it is the 'source of truth' for retrieval-augmented generation, often decoupled from the vector index for scale.

Visual Analog

A warehouse of full-text manuscripts where each page is assigned a barcode that corresponds to a digital coordinate in a separate GPS system.

Related Articles