Prompt Engineering Options (for create-ui chat)

1

New user here - the create-ui chat works fine, insofar as it can accurately resolve queries. But the out style needs refining - and it seems that prompt engineering is only provided for Scale users.

I’ll submit feedback on how this impacts startup adoption, but in the meantime, do most folks simply integrate the back-end retrieval functionality into a custom app that then handles the actual response generation via OpenAI (or other) API calls?

2

Hi @IMTanuki - you are right - the best way is to use custom prompts which is a Scale feature.
To refine the style of the output query, you can certainly make a separate call to an LLM and ask it for refinement.