A new desktop tool named PII GUI lets users scrub personal data from PDFs, Markdown, or plain text without sending any content to a server. Built on Tauri 2 with a React 19/TypeScript front end and a Rust back end, the app uses regex rules or quantized ONNX models for on‑device inference. The only network call occurs during an optional one‑time model download.

PII GUI supports two local workflows: a text‑to‑PII route that detects names, emails, phone numbers, URLs, dates, IDs and secrets before exporting a fully redacted PDF, and a PII‑to‑text flow that restores reviewed placeholders for reversible review. PDF input is parsed with pdf.js, preserving per‑character positions, while exported PDFs use pdf‑lib to burn opaque rectangles so hidden text cannot be recovered.

The application stores task history, custom rules, and results in a local SQLite database, allowing users to resume work across sessions. Optional backends include OpenAI Privacy Filter and Bard’s AI EU PII, both downloadable from Hugging Face. Released under the GNU Affero General Public License v3.0, PII GUI invites contributors to improve detection accuracy and broaden multilingual support.