Stepthrough · RAG Query Lifecycle

Following a Query Through the RAG Pipeline

Watch a user query traverse the 6 stages of a RAG system — from the raw question through embedding, retrieval, generation, and back to the user with citations.

Step 0 / 6

Click "Start" to begin

👤 User asks a question API API Gateway + Lambda EMBED Titan query → vector SEARCH Vector DB k-NN top-3 GENERATE Bedrock FM Claude / Llama response flows back to user "how do I..." [0.21, -0.45, 0.67, ...] 3 chunks retrieved grounded answer
Ready to start
Click Start below, or press →, to begin the walkthrough. Each step will highlight the active component and explain what's happening at that stage of a RAG query.
Step 0 / 6

Why interactive for sequences

Static diagrams make you rebuild the sequence in your head every time. Stepthroughs let your eye track one thing at a time — the active arrow shows what's happening right now, floating tokens show what's being passed, done steps fade. You can tab forward and back to build intuition fast.

Related

Pattern 1: Basic RAG — the full architecture diagram
Stepthrough 2: Agent ReAct Loop · Stepthrough 3: Defense-in-Depth Trace