Automated Knowledge Database with Google Drive and Supabase

This n8n workflow automates the process of extracting, processing, and managing data from a Google Drive file to create a dynamic, retrievable knowledge base using Supabase. It is ideal for building an intelligent document repository that can be queried via natural language, making it highly useful for knowledge management, chatbots, or digital assistants.

The workflow begins by downloading a file (such as an EPUB or text document) from Google Drive. It then loads the data using a specified data loader node designed for EPUB files. Several sticky notes provide crucial setup and operational details, including best practices for embedding models, Prisma setup, and database preparation.

Next, the content is split into manageable chunks using a recursive text splitter, enabling effective embedding and retrieval. The workflow then generates vector embeddings from the text using OpenAI’s embedding models, and these vectors are inserted into or updated within a Supabase vector table called ‘Kadampa’. A specific SQL function in Supabase allows similarity-based search queries.

The retrieval process involves querying the vector store for relevant documents based on user input, which is triggered via a webhook that initializes a chat interface. The language model (OpenAI Chat) interacts with the user, retrieves relevant documents from the vector store, and generates contextualized responses. This forms a real-time question-answering system that can be used for educational content, spiritual teachings, or any textual knowledge base.

The workflow also discusses how to delete or update records within the vector database, ensuring data integrity and up-to-date content. Overall, this automation facilitates a comprehensive, AI-powered knowledge system integrated within WordPress or other platforms for seamless information retrieval.

Practical use cases include creating a virtual assistant for a spiritual organization, automating content updates for educational materials, or developing an intelligent FAQ system connected to a content repository.