MemoRay Founder Access for Windows 0.4.0

PDF reading order and local refresh

• Born-digital PDFs with supported multi-column layouts preserve reading order more accurately in extracted text, search results, and source citations.
• Eligible existing PDFs can be refreshed locally with the corrected extraction. Their previous searchable representation remains available until a validated replacement is activated. Protected PDFs wait for unlock.

Ask and activation

• Founder Ask questions that begin with what or which and contain a hyphenated factual identifier are no longer misclassified as extensionless filename lookups.
• Founder Access activation ends in one clear, keyboard-accessible completion step that opens Import. The former Pin MemoRay action has been removed.

Documents and import cleanup

• Documents keeps the document title as the primary line and shows the source filename beneath it with the label File name:. Truncated titles and filenames can be revealed using keyboard focus or pointer interaction.
• Failed-import entries can be cleared individually or together without deleting source files or successfully imported documents.

Protected documents and password operations

• Locked Protected Documents continue to mask titles, filenames, suggested titles, previews, and open actions until the protected-document session is unlocked.
• Password operations and search run separately from the work that draws the app window. Password changes show progress and prevent conflicting document operations.
• Settings refreshes the Protected-document count after a saved access-level change. Settings and Transparency distinguish an unset password from a locked session.

Passport text reading

Passport text reading favors validated holder and labeled validity fields and can retry the data panel on later pages. Extracted details can still be incomplete and should be checked against the source.

Capability boundaries

MemoRay continues to process supported documents locally and keeps source verification visible.

This release keeps the existing local text-reading capabilities. It does not add general document reprocessing, changes to Protected Documents storage, payments, multi-device support, telemetry, or Mac and Linux support. Difficult scans, IDs/cards, handwriting, and unsupported page layouts can still require manual review.

Known limitations

• Needs-attention review requires returning to Import and selecting the next document. Continuous review of successive documents is not included in 0.4.0.
• Difficult ID/card scans can have incomplete names, dates, titles, or document types. Report authors and holders of rotated Social Security cards may be missing. Compare the preview with saved details and correct them manually.
• Search may take longer than desired.
• Setting, changing, or clearing a password can take several minutes while saved copies are updated, including when no documents are marked Protected. Submit once and keep MemoRay open until completion.
• Unsupported or ambiguous questions can return unrelated excerpts labeled Strong match. A match label does not establish that a document answers the question. Open the source, verify its relevance, and refine the query.
• When opening a locked Protected document from Documents, unlock in Settings first, then return to Documents and open it again.

Founder Access OCR boundary

• Founder Access uses the production Tesseract/PDFium OCR and text-extraction path.
• The experimental Worldclass OCR option is not included in Founder Access.
• Text-based and scanned/image-based PDFs are supported; scanned PDFs and images use on-device text reading.
• Readable, unencrypted DOCX files are supported. Encrypted or malformed DOCX files are rejected.

Known limitations

• MemoRay includes local text reading for scanned PDFs and images. Standard documents usually work best. Some low-quality scans, rotated cards, IDs, passports, membership cards, business cards, and handwriting-heavy images may be classified less precisely or may miss structured details. Review the source document when accuracy matters.
• Difficult scans, low-quality images, IDs/cards, complex layouts, and OCR-derived details may require review.
• OCR-heavy files may process more slowly, especially on weaker consumer machines.
• Multi-column manuals and PDFs can have imperfect reading order or interleaved text.
• Search is limited to text MemoRay successfully extracts and indexes.
• Successful import does not guarantee that every word or structured detail is searchable.
• Compare generated details and answer evidence with the source preview before using them for important decisions.