Extract and process information directly from PDF using Claude and Gemini

Overview

• This workflow helps you compare Claude 3.5 Sonnet and Gemini 2.0 Flash when extracting data from a PDF
• This workflow extracts and processes the data within a PDF in one single step, instead of calling an OCR and then an LLM”

How it works

• The initial 2 steps download the PDF and convert it to base64.
• This base64 string is then sent to both Claude 3.5 Sonnet and Gemini 2.0 Flash to extract information.
• This workflow is made to let you compare results, latency, and cost (in their dedicated dashboard).

How to use it

• Set up your Google Drive if not already done
• Select a document on your Google Drive
• Modify the prompt in "Define Prompt" to extract the information you need and transform it as wanted.
• Get a Claude API key and/or Gemini API key
• Note that you can deactivate one of the 2 API calls if you don't want to try both
• Test the Workflow