Making cards from notes, PDFs and photos

Memora generates cards from material you provide rather than from open-domain knowledge. That is a deliberate choice: transforming your text is exactly what a small on-device model does reliably, which makes it both useful and structurally hard to hallucinate.

The three inputs

• Typed or pasted notes. The most reliable input, because you control what goes in.
• PDFs with an embedded text layer, read with PDFKit.
• Photos and scanned pages, read with on-device Vision OCR. A photograph of your own handwriting usually works well.

Every proposed card is a draft. Nothing enters a deck until you approve it.