Protect personal data, credentials and confidential text before AI prompts, LLM workflows and document sharing. Reversible text masking with local English and Italian detection.
I built this because I wanted a simple way to use AI tools without sending sensitive data in plain form.
The tool detects PII in text and PDFs, lets you review or manually select what should be protected, replaces it with reversible placeholders, and gives you a private mapping so the original data can be restored later.
The focus is on keeping the workflow practical: anonymize first, use the cleaned version with an LLM, then restore the result locally when needed.
I’m especially looking for feedback on the PDF workflow, detection accuracy, and whether the privacy model is clear from the first visit.
Note: Currently active under development