The latest release of https://github.com/amaiya/onprem (a simple Python package to help use/apply local LLMs based on llama.cpp) includes a customizable, streamlit-based Web app for both general prompting and retrieval augmented generation (i.e., to talk your documents) and can be setup fairly easily:
# run at Python prompt
from onprem import LLM
llm = LLM()
llm.ingest('/your/document/folder')
# run at command line
onprem --port 8000
