Evaluate browser PDF and OCR redaction

[aanishs]

Clinician value

Explore whether clinicians can redact simple PDFs in the browser without uploading documents.

Scope

• Treat this as a second-phase evaluation after the text redactor is useful.
• Prototype PDF.js text extraction and Tesseract.js OCR for image-based pages.
• Document limitations: OCR misses text, redaction must be burned into output, and legal de-identification is not guaranteed.
• Use synthetic documents only.

Acceptance checks

• Evaluation notes state whether browser-only PDF redaction is reliable enough for a public tool.
• No sample file contains real client information.
• Prototype, if built, never uploads file contents.
• The user-facing caveat is explicit before download/export.