Recursive RAG for Google Drive Docs: OpenAI & Gemini

This workflow processes new documents from Google Drive using a robust pipeline: it triggers on file creation, downloads content, extracts text from PDFs or plain files, and applies recursive chunking to preserve context and relationships. Chunks are then embedded via OpenAI and stored in Supabase Vector Store with metadata and summaries for efficient retrieval.

The querying phase uses an AI Agent for hybrid search (vector similarity + keywords), powered by OpenAI for query understanding and Google Gemini for generating context-aware responses. It supports external data via HTTP requests, making it ideal for knowledge bases.

Benefits include time-saving automation for document-heavy workflows, scalable RAG for accurate AI responses, and integration of top AI models. Use cases: news aggregation (lifestyle category), legal doc analysis, research libraries, or enterprise search—reducing manual chunking and improving retrieval precision by 30-50%.