Appearance

Upload ​

Drag-and-drop is the simplest way to get a document into Recognito. Use it for one-off uploads, testing, or when an email forwarder isn't the right fit. Two places in the product accept drops: the Documents view and Studio.

This page covers what the manual upload paths support, the batch limits, and the edge cases.

Where to upload from ​

From the Documents view ​

Click Upload new files in the top-right of the toolbar. A modal opens with two options:

  • Drag-and-drop or click-to-browse — for files on your local machine.
  • Email instructions — the project's mailbox address shown for reference.

Drop files into the picker or click to open a file browser. Selected files upload immediately and show up in the table as Processing.

From Studio's left rail ​

Open any document in Studio and you'll see the document preview pane on the left. The pane accepts file drops — drop a PDF or image onto it and a new document is created.

This is less common than the Documents view drop. Most reviewers stay in Documents view for upload, then go to Studio for review.

File format support ​

The same formats as the email path:

  • PDF
  • PNG, JPEG, GIF, BMP, TIFF, WEBP, SVG (image formats)

Other formats aren't accepted. The drag-and-drop modal rejects unsupported files at the picker level — they don't make it to the server.

Batch size limit ​

The total size limit for a single drag-and-drop batch is 70 MB. This applies to the sum of all files in the drop, not per individual file.

If you need to upload more than 70 MB at once, split the upload into multiple batches. For routine high-volume uploads, the API path scales better than drag-and-drop.

Split large batches

A 100-page scanned PDF can exceed 70 MB. If a file is too large, compress it externally before uploading, or split the upload into smaller batches.

Multi-file uploads ​

You can drop multiple files in one go. Each file becomes its own document — the classifier doesn't run on manual uploads (only on email uploads).

If one file in the batch is invalid (wrong format, too large), the upload UI flags it at the picker level for unsupported formats.

What happens after upload ​

Once a file makes it to the server:

  1. Hash check. If Reject duplicate files is on (Settings → General), Recognito computes the file hash. A match against an existing document rejects the file silently.
  2. Page count check. If the file's page count exceeds the Auto-scan page threshold, the document's Document Status is set to Needs action rather than Processing. The user must open the document and pick pages.
  3. Extraction. OCR plus the project's extraction model run on the document (or selected pages, in the Needs action case).
  4. Initial Mapping. The project's Initial_Workflow runs after extraction completes.
  5. Document arrives in the queue. Status is Done, Approval Status is Pending.

The whole flow takes seconds to a couple of minutes depending on document complexity and queue depth.

Hash-duplicate rejection from upload ​

When Reject duplicate files rejects a drag-and-drop upload, the UI surfaces an error message. The file isn't created in the project; you can't see it anywhere afterwards.

If you didn't realize duplicate rejection was on and a file disappeared, check Settings → General → Reject duplicate files. The behavior is "silent" in the sense that no document is created, but the upload UI typically shows the rejection at upload time.

When manual upload makes the most sense ​

Drag-and-drop is a good fit for:

  • One-off documents that didn't come via the usual channel.
  • Testing. Trying out a new field configuration or Mapping rule on a sample document.
  • Backfills. Loading historical documents that suppliers didn't send through email.
  • Reviewer-initiated cases. A reviewer notices a document was missed; they grab the file from their inbox and drop it in.

For volume cases (suppliers sending 50+ invoices per day), email forwarding is easier. For programmatic cases (an ERP pushing documents on its own schedule), the API is the right path.

What's next ​