A console to see every document: what a customs agency gets beyond OCR

When a customs agency moves from reviewing documents by hand to processing them with artificial intelligence, the first question is not technical: how many today, how long did they take and which ones got stuck? RECO's operations console answers that without opening a single document: it works only with business metadata, never with the content.

In one week of September 2026 the console logged 11,546 documents and 21,591 pages. 74% arrived scanned, not digital, and the automatic processing per document still sat at a median of 18 seconds.

The training view splits each document's journey into stages: automatic processing, waiting, human review and total. It shows at a glance which part belongs to the machine and which to people, by agency and by period. That is the conversation an operations director wants to have with data: where to cut, whom to support.

Every administrative action, human or machine, lands in an immutable record: logins, credential exchanges, assigned plans. It filters by actor and by event and exports by event, by day or by user. In a regulated sector, being able to show that record is worth as much as speed.

The same console can process a batch of invoices or bills of lading through the same public endpoint the agencies use, with each one's credential. The result comes back grouped by invoice and the documents are not stored. That is how RECO's team tests what their clients will see, before they see it.

The figures in this post were read from the live console, week of September 14 to 20, 2026.