⚠ Switch to EXCALIDRAW VIEW in the MORE OPTIONS menu of this document. ⚠ You can decompress Drawing data with the command palette: ‘Decompress current Excalidraw file’. For more info check in plugin settings under ‘Saving’

Excalidraw Data

Text Elements

RAG Query Pipeline

User Query

Embedding

⚠️ Same model as indexing!

Vector Search

Top-K similar vectors

Retrieved Chunks

  • metadata

Re-ranking

(optional but recommended)

LLM Generation

Context + Query → Answer

Key Considerations

1

Embedding consistency is critical

2

K value affects recall (typically K=5-20)

3

Re-ranking improves precision significantly

4

Context length limits how much to pass

5

Monitor latency at each stage