Python OCR Script - Claude/GPT-4 API

Required: A local Python script to process roughly 4,000 scanned legal PDFs and extract specific entities into a CSV. The source files contain blurry faxes and handwritten notes. The script should use PyPDF2 and Tesseract for OCR, then pass the text to the Anthropic or OpenAI API for extraction (Date, Amount, Sender, Recipient). Must include a basic interface like Streamlit so non-developers can run it locally. Needs solid error handling to skip and log illegible pages instead of crashing.