RAG & Knowledge Basesfree

Evaluate RAG Response Accuracy with OpenAI: Document Groundedness Metric

by JimleukUpdated Aug 2026

RequiresOpenAI

1/5

FLOWS

STEPS · 5

Run manually by an operator

On manual run, builds an in-memory vector store, runs a RAG agent, and scores answer groundedness against documents with OpenAI.

Google Docs — Google Gemini

On manual run, indexes documents into Pinecone, then a Gemini agent retrieves context and writes outputs to Google Docs, Sheets, and Drive.

On manual trigger, answers queries via a Gemini and OpenAI agent over a Pinecone vector store built from Google Docs, Sheets, and Drive.

On manual trigger, loads files from FTP, embeds them with OpenAI, and stores the vectors in Qdrant in batches.