● PRODUCTION RAGpgvector · OpenAI ada-2 / text-embedding-3 · Claude

AI Chatbot with Document RAG

Upload PDFs, extract text with pdf-parse, chunk into 500 tokens with 50 overlap, store 1536d embeddings in pgvector, and stream answers via Claude with live source citations.

END-TO-END RAG PIPELINE7 Steps Verified
1
📄
PDF UploadedUser uploads PDF
2
⚙️
Text Extractedpdf-parse
3
✂️
Chunked500 tokens · 50 overlap
4
🧬
EmbeddedOpenAI text-embedding-3
5
🗄️
StoredPostgreSQL pgvector
6
🔍
SearchedCosine similarity · Top 5
7
⚡
AnsweredClaude + context citations

Upload & Index Documents

📂

Click to upload or drag & drop PDF

Extracts text via pdf-parse, creates 500-token chunks, embeds with text-embedding-3, and stores in pgvector.

PDF documents up to 25MB

Indexed Documents (0)

Explore All Chunks →
Loading indexed documents...