Eligible transfers can be categorized from filename, MIME type, and size through a quota-controlled background workflow.
One clearly bounded workflow turns transfer metadata into a stored category result.
The configured language-model provider receives the filename, MIME type, and file size and returns categorization metadata. File contents are not sent by this workflow.
Access is checked against the user's subscription tier before a job is queued. Monthly usage limits are enforced before each operation.
Successful categorization and its completion timestamp are stored in the transfer's AI metadata for the application to display.
Categorization runs as an Oban background job with bounded retries. Success and failure outcomes are recorded for quota accounting and operations.
The provider must be configured, and the user's plan and quota must allow the operation.
Upload a transfer normally. The application keeps the filename, MIME type, and file size needed for categorization.
An eligible request queues a background job. The configured provider receives only the transfer metadata used by this workflow.
The result and completion time are stored on the transfer, usage is logged, and the application broadcasts completion.
The implemented categorization workflow sends transfer metadata—not file contents—to the configured provider. Provider handling is governed by that provider's configuration and terms.
Categorization sends filename, MIME type, and file size; it does not upload file contents.
The deployment controls which language-model provider and model handle categorization.
Plan access and monthly quota are checked before a categorization job is queued.
Provider, model, token usage, latency, status, and errors are recorded for each operation.
Trusted by enterprise teams
Categorization is intentionally narrow, observable, retryable, and quota controlled.
The transfer filename gives the provider useful context without sending file contents.
The recorded MIME type helps distinguish documents, media, archives, and other categories.
File size is included as bounded metadata to help classify the transfer.
Transient provider failures can be retried by the background job without blocking the upload request.
Only successful operations count toward the user's monthly AI quota.
A PubSub event announces successful categorization so connected application views can refresh.
Join thousands of teams using Transfilio to transfer files faster, smarter, and more securely with the power of AI.