Bulk and remote upload

There are two ways to bring documents in at volume: the portal, for people, and the remote upload API, for systems.

Uploading from the portal

The document processor screen accepts PDF, Word (DOCX, DOC) and image files (JPG, JPEG, PNG). Each upload creates a document and a processing job, and the portal shows progress while the job runs. For a large backlog, add documents in batches you can check, rather than all at once, and review the first results before you commit the rest.

Uploading from another system

The remote upload API lets a line-of-business application, scanner or integration send documents without a person signed in.

1. Check that the remote API is enabled with its status endpoint, which also reports supported formats and authentication methods.
2. Authenticate. Service principal tokens and API keys are the supported methods.
3. Send a multipart request with the file and a JSON metadata string.
4. Store the document ID and job ID from the response.
5. Check status per document, or register a webhook for completion.
6. Fetch the result, and optionally download the summary, when processing completes.

Metadata can carry a predefined classification (which skips AI classification), custom fields, index values for filtering later, and a callback for completion. Uploads with an unsupported extension are rejected.

A note for shared-host customers: API keys are not currently available to tenant accounts. Ask about the right pattern for your deployment before you design around keys.

Schema ownership for remote uploads

Remote uploads are attributed to a system user rather than a person, and that system user owns no schemas. To still match documents to your schemas, Droqify resolves the schema owner from the document itself. The owner is stamped at upload from the owner of the API key. For older documents it is looked up from the key record behind the upload.

What this means in practice:

• Create your API key from the account that owns the schemas you want applied.
• If documents arrive flagged for no match, check key ownership first.
• You can assign a schema to a flagged remote document manually and reprocess.

Planning a bulk load

• Sample first. Run 20 to 50 representative documents, review the results and correct the schema before the full load.
• Pace the load. Queue depth and rate limits apply. Back off on rate-limit responses.
• Decide on review. Choose which steps should pause for a person before you load, so the queue is staffed.
• Track usage. Each document counts for at least 2 pages against your plan allowance, and included pages do not roll over. Plans show your usage in the portal.
• Keep identifiers. Store the document ID returned for each file so you can reconcile later.
• Reprocessing. Failed or changed documents can be reprocessed without extra page charges.