This n8n workflow automates the process of validating AI-generated outputs for PDF documents stored on Google Drive, useful for quality control in data extraction or content verification pipelines.
The workflow starts with a manual trigger, allowing users to initiate the process on demand. It then loads test cases from a Google Sheet containing URLs to PDFs, identifiers, and input prompts. The workflow loops through each test case, downloading the corresponding PDF from Google Drive and extracting its text content.
For each document, it uses an OpenRouter GPT-4.1 model to analyze and judge the accuracy of AI outputs related to the source material. The judgment—whether the output passes or fails—is based on criteria specified in the prompt, with explanations provided for each decision.
The results, including the AI platform used, relevant source references, inputs, outputs, and decision reasoning, are recorded back into a Google Sheet to facilitate tracking and analysis. The workflow includes pauses to avoid exceeding API rate limits and employs an intelligent conditional structure to handle non-PDF files.
This setup is ideal for teams conducting large-scale AI data validation, automated content review, or any process requiring structured AI performance evaluation on document datasets.
Reviews
There are no reviews yet.