I built this because I was tired of setting up the same RAG pipeline for every side project.
The Problem: Most RAG tutorials give you an answer but don't show where it came from. Hallucinations are a dealbreaker for real apps.
* The Solution:* I built a pipeline using Pinecone metadata to map vector chunks back to the original PDF page/text. The UI highlights the source snippet when the AI answers.
Stack:
Next.js 14 (App Router)
Pinecone (Serverless)
LangChain (Streaming)
Supabase (Auth)
Demo: https://www.fastrag.live
Happy to answer questions about the chunking strategy or the citation logic.
workwithtrp•1mo ago
I’m Atul, the maker of FastRAG.
We all have that one AI idea we want to build. But every time I started a new project, I found myself spending the first 20 hours doing the same boring setup: Configuring Pinecone indexes Wrestling with LangChain pipelines Figuring out how to parse PDFs
By the time the backend was ready, I was too tired to build the actual product.
I built FastRAG to skip the "boring" part. It is a production-ready boilerplate that gives you a fully functional RAG (Retrieval Augmented Generation) pipeline out of the box.
What’s inside: Citation Highlighting: The UI actually highlights the exact source text in the PDF (users trust this way more than a black-box answer). Markdown Support: Full rendering for code blocks, tables, and lists. The Stack: Next.js 16, Tailwind, Supabase, and Pinecone.
The Holiday Launch Deal I want you to actually ship something over the break, so I’ve set up a race:
First 69 Makers: Get 69% OFF (approx $9).
Code: FAST69
Too Slow? You can still get 40% OFF.
Code: HOLIDAY40
(Base price is $29, so grab the code while it lasts).
I’ll be hanging out in the comments all day. Let me know what you think about the citation feature!
Happy Building, Atul