dScribe · AI Extraction Engine

Documents in.
Declarations out.

Scribe reads your invoices and packing lists, extracts every header field and line item with AI, validates them against customs rules, and turns it all into a ready-to-file declaration — in seconds, not hours.

See all products

99.8%

Field extraction accuracy

< 60s

From upload to proposal

0

Manual re-keying

Meet Scribe

Customs work starts with a pile of documents — invoices, packing lists, order confirmations — and ends with a structured declaration. The gap between the two is hours of manual data entry, and a single typo can hold a shipment at the border.

Scribe closes that gap. Upload your documents and Scribe reads them, extracts every field and line item with a confidence score, reconciles your invoice and packing-list data into one source of truth, and aggregates it into customs-ready declaration lines. You review what it flags, then publish.

From document to declaration

Scribe runs a seven-stage pipeline on every dossier — fully automated, fully auditable.

1

Upload documents

Drop in invoices and packing lists as PDFs — scanned or digital. Scribe stores them securely and starts reading immediately.

2

Extract header fields

AI pulls parties, dates, procedure codes, EORI numbers and more — each with its own confidence score.

3

Extract invoice lines

Every line item is read from the invoice: product code, description, origin, quantity, unit price and weights.

4

Extract packing-list lines

Packing-list line items are extracted in parallel so they can be cross-checked against the invoice.

5

Merge into detailed lines

Invoice and packing-list data are reconciled into one detailed line set — no duplicates, no gaps.

6

Aggregate declaration lines

Detailed lines are grouped into customs declaration lines by HS code and country of origin.

7

Validate & publish

Built-in rules check totals, HS codes, EORI and dates — then Scribe exports a clean Excel declaration.

Built to be trusted

Accuracy and control at every step — not a black box.

Confidence scoring

Every extracted field carries a confidence score. High-confidence values auto-accept; anything uncertain is flagged for a quick human check.

One-click review

Confirm or correct any field by hand with the magic-marker tool, and Scribe carries it through the whole dossier.

Rule-based validation

A built-in rule engine checks invoice totals, HS code format, EORI numbers and date consistency before anything is published.

Smart master data

Addresses and HS codes are fuzzy-matched against your master data, so parties and codes resolve automatically.

Live extraction stream

Watch fields populate in real time as Scribe reads each document — no waiting on an opaque batch job.

One-click Excel export

Publish to Excel with the sections you choose: header fields, invoice lines, packing list and declaration lines.

Common questions

Commercial invoices and packing lists today, as PDFs — including scanned documents, thanks to built-in OCR. More document types are being added.

Scribe reaches around 99.8% field accuracy and scores every field, so low-confidence values are surfaced for a quick review rather than slipping through unnoticed.

Scribe produces the declaration proposal and Excel. It feeds directly into Droblet’s filing products — ICS2, ENS, ELO and CDS — to submit to the authorities.

Yes. Scribe is multi-tenant with a separate database schema and document storage per customer, plus a complete audit trail on every dossier.

Turn your documents into declarations

See Scribe read a real invoice and build a declaration end to end.

Try free tools