Last updated: 2026-04-25
Bulk uploading documents
The fastest way to get a backlog of leases, invoices, BER certs and receipts into Anchorlet is the bulk uploader.
How to upload
- Open Documents in the top nav.
- Click Bulk upload (top right).
- Drag a folder of files into the dropzone, or click to pick. Up to 50 files per batch.
- The uploader runs each file through:
- Dedupe — if the exact same file (matching SHA-256) is already in your workspace, the row is marked Duplicate and skipped.
- Storage upload — file lands in Supabase Storage, scoped to your workspace path.
- Text extract + classify — for PDFs with a text layer, Anchorlet pulls the text and runs the Haiku classifier. For images (JPEG/PNG/WebP), Claude reads the image directly.
- Property match — looks at the filename + extracted text and tries to link the document to a property in your workspace.
- Files appear in the table with their classification result + matched property.
What you can upload
| Format | Path | Notes |
|---|---|---|
| text-layer or scanned (vision) | Text-layer PDFs go through the cheap classifier; scanned PDFs fall back to vision. | |
| JPEG / PNG / WebP / GIF | image classifier | Phone photos of receipts, signs, etc. work. |
| XLSX / XLS / CSV | spreadsheet path | Rent statements, expense ledgers. |
HEIC / HEIF aren't supported directly — convert first, or upload via the property page (which handles conversion downstream).
Per-property uploads
If you only have a single document for a specific property, use the Documents → Upload button on the property's detail page instead. Same pipeline, but the property is pre-linked so no fuzzy matching is needed.
Where things land
- Successfully classified files go into the property's documents tab.
- Files where the classifier was unsure end up in Documents → Needs review with a reason chip (e.g. "type uncertain", "no property match").
- Files where the classifier hard-failed (corrupt PDF, no text layer, network error) stay in the queue with a "Try again" button.
See What the classifier does for the details on which document types Anchorlet recognises.