You choose a PDF
The document enters through the device's standard file picker or share mechanism.
SECURITY ARCHITECTURE
Your documents don't need a detour.
Didentify Docs is built around a local processing boundary: documents are opened, read, reviewed, redacted, and exported on your device.
How does Didentify Docs protect documents?
Didentify Docs avoids a document-processing backend. It uses on-device OCR, stores work locally, asks the user to review detected content, and rebuilds redacted output with selected information removed.
A short path from input to output.
The product's core security property is architectural: Didentify does not need to receive the document to provide its main features.
Your device recognizes text
OCR is performed locally using platform tooling: Apple Vision on Apple platforms and a bundled ML Kit model on Android.
You review proposed redactions
Detection is assistance, not an irreversible automatic decision. You decide what is removed.
The app rebuilds the export
Selected content is removed in a clean PDF output, with an optional certificate that records SHA-256 verification details.
Platform references: Apple states that Vision text recognition processing happens on-device in its Vision documentation. Google documents the bundled Android ML Kit option as statically linked and immediately available in its text recognition guide.
What the architecture excludes.
Reducing the number of systems that can receive sensitive information reduces the trust surface around the workflow.
No uploads to Didentify
The core workflow does not send document pages or recognized text to a Didentify server.
No account required
There is no Didentify sign-in needed to open and process a document.
No analytics SDK
The app is designed without a product analytics or third-party crash-reporting service.
The app contacts RevenueCat to validate optional Didentify Pro purchases using an anonymous ID and can download a PDF from an HTTPS link you choose to share into the app. RevenueCat never receives documents, page images, or recognized text. Details are in the privacy policy.
Security boundaries, stated plainly.
No app can eliminate every risk. These boundaries help you decide how Didentify Docs fits your workflow.
OCR can make mistakes
Scans, handwriting, unusual layouts, and low-quality images can produce missed or incorrect text. Always review before export.
Your device remains important
Local processing does not replace device encryption, operating-system updates, access controls, or safe backups.
Destinations have their own rules
Once you intentionally share an exported file, the receiving app or service controls what happens next.
SECURITY FAQ
Clear answers.
Does Didentify Docs send documents to a server?+
No. Didentify Docs does not use a document-processing backend. Reading, OCR, redaction, and export happen on the device.
Is OCR on-device?+
Yes. Apple platforms use Apple's Vision framework. Android uses Google ML Kit with the text-recognition model bundled in the app.
Can automated detection miss sensitive information?+
Yes. OCR and pattern detection can miss or misread content. Didentify Docs keeps the user in the review loop before export.
Does a SHA-256 certificate prove every redaction is correct?+
No. A hash helps identify a specific exported file and detect later file changes. It does not replace the user's review of what was selected for removal.