Retrieve answers from live data using AI embeddings

How it works

1. 1.Trigger automation when a chat message arrives or on a scheduled interval
2. 2.Fetch live documents from Notion and Supabase and process them with text splitter
3. 3.Divide results into batches and create vector embeddings via OpenAI
4. 4.Store embeddings in a Vector Store backed by Supabase
5. 5.Search and retrieve the most relevant documents based on user query
6. 6.Send context and question to LLM chain and deliver final answer

Use cases

• Answer employee questions instantly based on live company knowledge base
• Retrieve contract terms and policies from constantly-updated document repositories
• Summarize and analyze Notion reports and notes in natural language
• Build an intelligent assistant that learns from new data automatically

Requirements

• OpenAI API key for embeddings and chain completion
• Supabase connection with a table to store vector embeddings
• Notion workspace containing documents to be searched and indexed

Service Value

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