Multi modal RAG with Gemini API in Vertex AI vs Building a RAG Chat Bot the Hard Way with Vertex AI

Google Cloud Build with AI Vertex AI - Custom ML & MLOps
, , ,
1

I’m currently evaluating two approaches for building a Retrieval-Augmented Generation (RAG) chatbot and would appreciate your insights on the best use cases for each option:

  1. Multimodal Retrieval-Augmented Generation (RAG) using the Gemini API in Vertex AI
  2. Building a RAG Chatbot the Hard Way with Google Vertex AI, BigQuery, and LangChain

I’ve reviewed some labs on both approaches, but I’m still a bit confused about their respective use cases.

Could anyone clarify in which scenarios each option would be most suitable? For example:

  • When should I choose the Gemini API in Vertex AI for a multimodal RAG system?
  • When is it better to go with Building a RAG Chatbot the Hard Way with Vertex AI, BigQuery, and LangChain?

Additionally, which approach is generally better for building a chatbot system, and why?

Looking forward to your insights!