RAG on live data

How it works

1. 1.Load documents from Notion or Supabase and split them into manageable text chunks using token-based splitting
2. 2.Generate vector embeddings for all content segments via OpenAI models and persist them in the Supabase vector store
3. 3.Accept user queries through a chat interface and retrieve relevant documents, then synthesize answers using LLM chains

Use cases

• Build a customer support chatbot that answers questions by searching a constantly updated knowledge base
• Create an internal search engine for organizations to query policies, documentation, and structured data across multiple sources

Requirements

• Valid OpenAI API key with access to embedding and chat models
• Supabase account with a configured vector database and proper read/write permissions

Service Value

Ideal as a smart automation service combining integrations and AI to produce ready-to-use results.