What it does
This n8n workflow reads a document from Google Drive, extracts its text and splits it into chunks at section markers. A model generates context for each chunk, which is added to the chunk before embeddings are created and stored in a Pinecone vector store. It is based on an Anthropic article about contextual retrieval.
Use cases
- 01Index policy documents stored in Google Drive
- 02Prepare long manuals as chunks with context
- 03Load a Pinecone store for answering questions
Connects
Google Drive · Code · AI Agent · Recursive Character Text Splitter · Pinecone Vector Store · Default Data Loader · Embeddings Google Gemini