This n8n workflow streamlines the process of extracting structured data from PDF documents using multiple AI models for comparison. The workflow begins with a manual trigger, allowing the user to initiate the process on demand. It then downloads a specific PDF file from Google Drive and converts it into a base64-encoded string, which is necessary for the AI APIs. The encoded PDF is simultaneously sent to two different AI services: Claude 3.5 Sonnet and Gemini 2.0 Flash, each capable of processing PDFs and extracting information based on a defined prompt.

The workflow uses several sticky notes to provide guidance on configuration, such as setting prompts and output formats. The core nodes are the Google Drive download, the file encoding, and the API calls to both AI models with customizable prompts. The responses from these AI services can then be compared to evaluate their performance, latency, and cost effectiveness.

This workflow is ideal for scenarios where organizations need to automate data extraction from invoices, contracts, or other PDFs, leveraging multiple AI models to determine the best solution or to cross-verify extracted data. It can be adapted for various use cases by changing the prompt and target files, making it a flexible tool for document processing automation.