RedactDoc

AI chat platforms like Claude, ChatGPT, Gemini, and Mistral have become essential productivity tools, but users routinely upload documents containing sensitive personal information without realizing it. Names, Social Security numbers, credit card numbers, medical records, and addresses get sent to cloud servers with a single drag-and-drop.

There was no frictionless way to protect this data. Existing solutions required manual redaction, separate desktop apps, or uploading documents to yet another cloud service, defeating the purpose entirely. We needed to build something that sits invisibly between the user and the AI platform, catches PII in real-time, and never lets sensitive data leave the device.

• Zero-friction protection: The extension works silently in the background. Users upload files exactly as they normally would, and RedactDoc catches sensitive data before it leaves the browser
• Smart detection across 17 categories: Names, credit cards, Social Security numbers, medical records, and more, each validated to minimize false positives. Everything runs locally on the device, with no data ever sent to external servers
• Works with the files people actually use: Supports plain text, CSV, PDF, and Word documents. The redacted file preserves the original format, ready to download in one click
• Clear, non-intrusive feedback: After redaction, a small badge appears showing exactly what was found (category breakdown, item count) with a one-click download for the clean file
• Built-in product analytics: Usage tracking designed for product decisions: daily active users, retention, redactions per day, and platform breakdown

• Multi-language detection: Detects sensitive data in English (US), French (France), and Spanish (Spain), with country-specific categories like French national IDs or Spanish DNI numbers
• Guided onboarding: A six-step setup flow that asks users about their context (personal, business, healthcare) and pre-configures the right protection settings, with friendly illustrations and smooth transitions
• Built-in feedback loop: Users can report issues or suggest features directly from the redaction badge or the settings page, making it easy to collect insights from day one
• Privacy by default: If anything goes wrong during a scan, the upload is blocked entirely. The extension never lets unredacted content through to the platform, no matter what