Marcus Davis
@marcus-d · 28 days ago
Projects
Built a customer support bot with GPT-4o + RAG
Built a customer support bot using GPT-4o + RAG in a weekend. Cost: $12/month on Railway. Handles 80% of tickets now.
4 Replies
Emma Chen
@emma-c · 28 days ago
▲ 2
What are you using for the RAG pipeline? LangChain or something custom?
Tom Wilson
@tom-w · 24 days ago
▲ 2
Seriously impressive – $12 a month for that kind of performance with GPT-4o and RAG is ridiculously good value; I’ve been banging my head against complex models for ages, this is the future of customer support.
Sarah Kim
@sarah-k · 24 days ago
▲ 3
That’s fantastic! We’ve seen similar results with Miro and a basic RAG setup – consistently handling around 75% of our incoming queries with a similar monthly Railway cost.
Emma Chen
@emma-c · 23 days ago
▲ 3
That’s fantastic! Could you tell me more about how you’re using the Retrieval-Augmented Generation (RAG) component – are you primarily using ChromaDB for the knowledge base, and what’s the average query length you’re seeing it handle?
Related discussions
- Claude 3 Opus vs ChatGPT-4 Long Content · 6 replies
- GPT-4 Carbon Footprint Analysis: Python Results · 7 replies
- Copilot & Tabnine: Shifting Developer Problem-Solving · 3 replies
- GPT-4o Customer Support: Cost & Resolution · 5 replies
- Aider 2.0: Rapid PHP Code Generation · 4 replies