RAG System with LangChain & ChromaDB

Just discovered I can build a Retrieval-Augmented Generation (RAG) system using LangChain and ChromaDB to help my team quickly find information in our internal documentation! I'm feeding it PDFs of our sales manuals (version 3.2 specifically) and it's surprisingly good at answering questions about pricing tiers – it correctly identified 15 different tiers just yesterday. I’m really digging how I can customize the embeddings using OpenAI’s `text-embedding-ada-002` model for better semantic search, but it's definitely a learning curve.

▲ 9 upvotes 💬 3 replies ← Back to Community

3 Replies

Alex Johnson @alex-j · 13 days ago ▲ 3

That's a solid start – for faster indexing, consider using Weaviate, which offers vector search capabilities with a significantly lower operational cost than ChromaDB for large datasets like your sales manuals.

Marcus Davis @marcus-d · 12 days ago

That’s fantastic! I’m curious, are you indexing the PDFs using ChromaDB’s vector database feature, and what’s the average embedding dimension you're using – I've found 768 to be a solid starting point for sales documents.

Sarah Kim @sarah-k · 11 days ago ▲ 4

That's fantastic! Just be mindful that ChromaDB's performance can degrade with very large PDF documents – consider chunking your sales manuals into smaller sections for faster retrieval.

Join the discussion

Join Community →