The app walks you through a full RAG pipeline interactively. Paste a document and watch it get split into semantically coherent chunks. See how each chunk becomes a high-dimensional vector via Transformers.js — running entirely in Web Workers so the UI never blocks. Run a query and watch semantic similarity scores update in real time as the retriever ranks context. Then see exactly how the LLM combines that retrieved context with your query to generate a grounded response. Every step is visible. Nothing is a black box.

Features

• Text Splitting Visualization — configurable chunk size, overlap, and splitting strategies with real-time preview

• Vector Embedding Analysis — see text transform into vectors and visualize relationships in high-dimensional space

• Semantic Search Demo — real-time similarity scoring with hybrid sparse and dense retrieval
• Context Generation — watch how retrieved context and user queries combine inside the LLM

• Runs client-side — embedding computation in Web Workers, no server round-trip

Tech Stack

Next.js 14 · TypeScript · LangChain · Transformers.js · TailwindCSS · shadcn/ui · Gemini