AI-Powered Document Evaluation with n8n

This n8n workflow enables comprehensive evaluation of AI responses based on retrieved documents, specifically focusing on how well the answer is grounded in source material. Triggered manually, the workflow fetches a PDF document (e.g., Bitcoin Whitepaper), processes it through natural language embeddings, and stores it in a vector store for efficient retrieval. When a user submits a question, the system retrieves relevant documents, generates an AI response grounded in this information, and assesses the alignment between the response and the source documents using an embedded evaluation model. The outcome includes a score indicating the response’s groundedness, which is recorded in Google Sheets. This workflow is particularly useful for ensuring high-quality, factually accurate AI outputs in knowledge-based applications, FAQs, or technical support scenarios.