Stript identifies and pseudonymizes sensitive information in documents using models that run locally. It combines contextual detection with pattern matching and lets users review replacements before exporting documents. An encrypted mapping supports consistent replacements and later restoration of the original information.