RAG Brag with My AskAI founders, Mike Heap and Alex Rainey

What this post added

This post details My AskAI's technical implementation of a RAG system, specifically highlighting their use of Pinecone for vector storage and retrieval. It also discusses their chunking strategy (400-token chunks with 20-token overlaps) and data preprocessing considerations for LLMs. The post mentions the use of LangChain for text splitting and PortKey as an LLM gateway for managing model requests, including fallback models and response caching.