← Simulators
🔍 RAG Pipeline
Documents → embeddings → vector DB → query → top-k search → LLM.
RAG = Retrieval Augmented Generation. Flow:
1
Documents— Your corpus (PDFs, docs, etc.)
↓
2
Embed— Convert chunks → vectors
↓
3
Vector DB— Store embeddings (e.g. Pinecone, Weaviate)
↓
4
Query— User question → query embedding
↓
5
Search— Find top-k closest vectors (ANN)
↓
6
LLM— Send query + retrieved chunks → generate answer