Hey Product Hunt! I built DocSanitizer out of personal frustration with everyday AI workflows. We all paste logs, contracts, PDFs, and internal notes into ChatGPT, Claude, and Gemini. But doing so creates two massive problems: Accidental PII leaks: Sending client emails, credit cards, internal IPs, or medical identifiers to cloud LLMs violates compliance and common sense. Token fluff: Raw documents are packed with repeated headers, broken line wraps, and whitespace that burn through context windows and degrade model accuracy. Most anonymizers rely on cloud APIs-which defeats the entire purpose of privacy. DocSanitizer takes a zero-cloud approach: High-performance Rust core: Everything runs locally on your device. Transient RAM execution: Documents live in temporary memory during processing and are wiped immediately. Zero telemetry, zero analytics, zero network requests. Domain Pattern Library: Out-of-the-box support for financial data, medical codes (MRN, ICD-10, NDC), and court filings (FRCP 5.2 masking), plus custom regex. Token Analytics: Accurate on-device BPE token counts for OpenAI and Claude so you know exactly what you’re sending. It's live on desktop and web, with iOS and Android builds planned. I’d love your feedback on our redaction patterns and token optimization rules-what features or domain libraries would make this indispensable for your workflow? Thanks for checking it out!

DocSanitizer

Censor private PII & shrink docs to fit any AI prompt

Censor private PII & shrink docs to fit any AI prompt