Uploading Files
Get PDFs, spreadsheets, contracts, and scans into the Documents hub, and understand what happens to them afterwards.
Overviewβ
There are three ways to upload:
- Drag & drop β drop files anywhere on the Documents page. A hint appears as soon as you start dragging.
- Upload button β click Upload in the page header.
- Upload here β right-click any folder in the tree and choose Upload here to preselect the destination.
You can upload up to 10 files at once, and each file may be up to 50 MB by default. Administrators can lower or raise the size limit for a deployment β see Settings.
The Upload Dialogβ
Before the upload starts you can set everything the files should inherit:
| Field | Description |
|---|---|
| Destination folder | Where the files land in the tree β defaults to the folder you were browsing |
| Categories | Categories applied to every file in this batch |
| Tags | Tags applied to every file in this batch |
| Add to AI knowledge | Make these documents available to AI-powered answers |
| Classify with AI | Let AI suggest categories, tags and a summary for this batch |
| Visibility | Organization or Private |
Each file gets its own progress row. If one fails, the rest continue β use Retry on the failed row, or Clear finished to tidy up the list.
Classify with AI starts from your organization's default and applies to this batch only β turning it off here does not change the setting for anyone else. An administrator sets the default in Settings β Documents. The toggle is hidden when no AI provider is configured, since there would be nothing to classify with.
Supported File Typesβ
Uploads are checked by inspecting the file's actual content, not just its extension. A file whose content does not match its extension is rejected with "This file's content doesn't match its extension".
| Format | Extensions |
|---|---|
.pdf | |
| Word | .docx |
| Excel | .xlsx |
| PowerPoint | .pptx |
| OpenDocument Text | .odt |
| OpenDocument Spreadsheet | .ods |
| CSV | .csv |
| Plain text | .txt |
| Markdown | .md |
| HTML | .html, .htm |
| Images | .png, .jpg, .jpeg, .webp, .gif |
SVG files and other XML-based markup are never accepted, in any form. They can carry active content that would run in a colleague's browser, so the hub rejects them outright rather than trying to clean them.
What Happens During Processingβ
Every uploaded file moves through a short lifecycle. The current stage is shown as a badge in the table and in the detail panel.
| Status | Meaning |
|---|---|
| Processing | The file is stored and its text is being read out |
| Ready | Processing finished β the file is searchable and can be previewed |
| Failed | The text could not be read; the file itself is safe and still downloadable |
Reading the text out is what makes content search and AI answers possible. It works for:
PDF Β· Word (.docx) Β· Excel (.xlsx) Β· CSV Β· plain text Β· Markdown Β· HTML
Other accepted formats β PowerPoint, OpenDocument files, and images β upload and store correctly, but their text cannot be read yet. They finish as Failed and go to the review queue so somebody notices. The file itself is intact: it can be downloaded, previewed, linked, categorized, and tagged like any other.
A Failed status only ever means "the text could not be read". It never means the file was lost or damaged.
When Extraction Failsβ
Open the document and use the detail panel:
- Retry β re-runs processing. Useful after a transient problem.
- Edit extracted text β type or paste the text yourself (see below).
- Do nothing β the file stays usable as a plain attachment.
A failed file is flagged Needs review with the reason Extraction failed, so it turns up in the review queue instead of quietly disappearing.
Correcting Extracted Textβ
Scans, unusual layouts, and multi-column PDFs can produce messy text. Because search and AI answers read that text, it is worth fixing.
- Open the document's detail panel
- Click Edit extracted text
- Correct the text β the dialog reminds you that "search and AI answers use this text"
- Click Save & re-index
Once you have edited the text by hand it is protected: re-running processing will not silently overwrite your corrections.
Storageβ
Uploads count against the organization's storage quota if one is set. Archived files still count β their bytes still exist. Administrators can see current usage in Settings β Documents.
Related Pagesβ
- Documents Overview β what the hub is
- Organizing Documents β folders, categories, search
- Reviews & Approvals β the review queue
- AI Knowledge β adding documents to AI answers
- Settings β size limits and storage quota