Adding documents
Upload documents, check supported formats, and review processing results in the Inbox.
Drop a file on Add documents and it appears in the Inbox at once, with a spinner. The original bytes are stored first, encrypted under a key of their own, before anything else happens. Everything after that is a background stage that can fail and be retried. If text extraction fails, the stored original remains available, but its contents may not be searchable. Retry processing from the Inbox.
What Harbor does with each kind of file#
| You drop | Harbor does |
|---|---|
| A PDF that already has a text layer, such as a statement or a downloaded policy | Reads the text directly and skips OCR entirely. Processing time depends on the document. |
| A PDF that is only images, such as a scan | Runs OCR page by page. Processing time depends on the hardware, scan quality, and page count. The row shows Reading page 3 of 12. |
| A phone photo, HEIC or JPEG or PNG | Wraps it in a PDF and reads it like a scan. Crooked and skewed pages are straightened first. |
| A ZIP of a decade of scans | Each file inside, one by one. |
| A Word or Excel file | Stored as it is, not searchable. Office formats are not read in this version. |
Duplicates#
Harbor recognises a file it has already seen and asks whether to add it as a new version of the existing document or skip it. Files are never silently merged.
Long documents#
Long documents take more time to process. If Harbor asks for confirmation before processing, review the page count before continuing.
From a phone#
Scan with the phone's own scanner and email the result to the vault. Emailed photos arrive as HEIC or JPEG and are handled like any other photo. See Email.