Built on LangChain.js · runs in your browser

LangChain RAG Playground

A retrieval-augmented generation pipeline wired with LangChain Expression Language (LCEL): split → embed → store → retrieve → prompt → model → parse. Embeddings run locally via Transformers.js; generation uses your own OpenAI key, which never leaves this page. Every intermediate step is shown.

1 · Document

chunk size overlap

2 · Ask

top-k

Answer

Run the chain to see a grounded answer stream in.

Retrieved context

Assembled prompt

Chunks

split → embed → store → retrieve → prompt → model → parse, all in src/rag.ts (~60 lines). The same chain in langchain for Python is a near line-for-line translation — see the README. Built by Osama Ansar.