Paperless mail with a burden of proof

Request: „Off to paperless.“

1. Scan the letter; the document archive with OCR (paperless-ngx) reads the content.
2. Assign title, correspondent, document type, storage path and date.
3. Pull key facts and deadlines into the related note — contract, device, person.
4. An audit proves deterministically that every document is linked somewhere.

A scanner that scanned into folders, and a stack of 551 sheets that never shrank.

Feeding the scanner, deciding what stays on paper, and every deadline in the calendar.

The archive search silently capped at 15 hits although 22 documents existed — since the fix the tool reports every truncation out loud.

“Ask the archive first, then the person” reverses the burden of proof: a note saying “open” does not count while the archive disagrees.

Every letter is scanned, a self-hosted document archive with OCR reads it, and the AI assigns title, correspondent, type and storage path. Key facts and deadlines land in the matching note. The reverse also holds: “the note says it's open” does not count — ask the archive first, then the person. An audit proves that each of the 1,423 documents is linked somewhere. The 551-sheet paper backlog shrank to zero within a week.