Definition
The baseline architectural pattern for Retrieval-Augmented Generation that follows a linear 'Index-Retrieve-Generate' workflow without iterative loops or query optimization. While simple to implement and offering low latency, it frequently suffers from poor retrieval precision, 'Lost in the Middle' context issues, and a lack of reasoning over complex, multi-part queries.
The standard 'hello world' of RAG, lacking advanced steps like re-ranking, query expansion, or agentic feedback loops.
"A basic keyword search in a physical library where you grab the first three books on the shelf and read them regardless of whether they actually answer your specific question."
- Vector Database(Component)
- Chunking(Prerequisite)
- Semantic Search(Component)
- Advanced RAG(Evolutionary Successor)
Conceptual Overview
The baseline architectural pattern for Retrieval-Augmented Generation that follows a linear 'Index-Retrieve-Generate' workflow without iterative loops or query optimization. While simple to implement and offering low latency, it frequently suffers from poor retrieval precision, 'Lost in the Middle' context issues, and a lack of reasoning over complex, multi-part queries.
Disambiguation
The standard 'hello world' of RAG, lacking advanced steps like re-ranking, query expansion, or agentic feedback loops.
Visual Analog
A basic keyword search in a physical library where you grab the first three books on the shelf and read them regardless of whether they actually answer your specific question.