QA RAG Platform

Next.js·OpenRouter·Pinecone·Vercel
Ask a question

Upload a document → it is chunked and embedded → the vectors go into Pinecone → an OpenRouter LLM answers from the retrieved chunks only, citing its sources. Retrieval-augmented generation, end to end.

1Upload

PDF, DOCX, XLSX/XLS, CSV, or text (TXT, MD, JSON, HTML, XML, YAML)

2Extract & chunk

Split into overlapping pieces

3Embed

OpenRouter turns each chunk into a vector

4Store

Vectors are upserted into Pinecone

5Retrieve

The closest chunks to your question

6Answer

An OpenRouter LLM answers from those chunks only

Documents…
Total chunks…

Upload a document

PDF, DOCX, XLSX/XLS, CSV, or text (TXT, MD, JSON, HTML, XML, YAML) — chunked, embedded, stored in the index

Click to browse or drag files here