Back to Templates

Process Documents with Recursive Chunking using Google Drive, OpenAI & Gemini RAG

Created by

Created by: Mohsin Ali || binbakhsh

Mohsin Ali

Last update

Last update 18 hours ago

Share


1. Document Ingestion & Processing

Google Drive Trigger monitors for new files → Loop Over Items processes each file → File Info extracts metadata → Google Drive downloads the actual content → Switch routes to appropriate extractors (PDF or TEXT) based on file type

2. Content Transformation & Chunking

Document Data node processes extracted text → Recursive Splitter breaks content into contextual chunks → Chunk Splitting applies intelligent segmentation while preserving document context and relationships between chunks

3. Embedding & Storage

Basic LLM Chain processes chunks → OpenAI Chat Model generates contextual understanding → Summarize creates document summaries → Supabase Vector Store saves embeddings with metadata → Embeddings OpenAI creates vector representations → Default Data Loader handles storage operations

4. Query Processing & Retrieval

When Clicking Execute triggers user queries → OpenAI processes and understands the question → AI Agent orchestrates hybrid search (combining vector similarity + keyword matching) → Google Gemini Chat Model generates final responses using retrieved context → HTTP Request handles additional external data sources