RAG Architecture

An interactive visualization of the RAG pipeline. Click on each step to understand why it exists and the trade-offs involved.

Data Flow Overview

Documents

PDF, TXT

Chunking

1000 chars

Embeddings

1536-dim

FAISS

Vector Store

Query

User input

Retrieve

Top-K

Generate

GPT-4o-mini

See it in action

Try the live demo to see this architecture working in real-time.

Launch Live Demo