RAG Webapp

art271

Hello everyone,

In my Dataiku Flow, I have a RAG setup that includes embeddings and prompts. I’d like to replicate this process—achieving the same results as in Prompt Studio—in a Dash web app. The goal is to reuse the knowledge base built in the Flow and leverage the augmented LLM created by the embedding recipe.

Does anyone know the best approach to achieve this? Should I deploy an API, use LangChain, or directly leverage Dataiku’s API?

I’m working with Dataiku 12.6. Thanks in advance for your help!

Alexandru

Hi,
You can leverage headless API via Dash Webapp :

https://developer.dataiku.com/latest/tutorials/webapps/common/api/index.html

Use the KB and LLM mesh APIs

https://developer.dataiku.com/latest/concepts-and-examples/llm-mesh.html#using-knowledge-banks-as-langchain-retrievers

THanks