Recognising documents without AI
No AI models: text extracted from the PDF and deterministic rules, which always give the same result and can be checked one by one.
I validated them on 21 real PDFs from the client: 10 out of 10 recognised and no false positives, even on scans that went through OCR.