All case studiesCase study

AI-assisted customs declarations for Lett 2000

DeclAI is an AI platform we built for the customs brokerage Lett 2000. It turns invoices, packing lists and transport documents into a validated customs declaration, ready to file in EDS, the State Revenue Service's electronic declaration system.

DeclAI declaration table listing each item's tariff code, description, weight, value and country of origin
About the client

Lett 2000 is a customs brokerage specializing in international trade compliance and cross-border logistics. It handles shipments across multiple jurisdictions.

The challenge

Two parts of the declaration work depended on people: processing the documents and classifying the goods.

  • Manual document processing caused delays and errors
  • Tariff classification required expert knowledge for every shipment
  • Manual processing became a bottleneck as trade volumes grew

What we built

DeclAI, also known as Customs Monster, is an automated platform that covers the full declaration workflow end to end.

01

Document recognition

The shipment documents are recognized automatically, and the data taken from them is validated in real time.

02

HS code classification

Machine learning assigns an HS code to each item of goods in the shipment.

03

Sanctions and compliance checks

Sanctions screening is built in, and compliance risks are flagged proactively.

04

Declaration assembly

The recognized, classified and screened data is assembled into the finished declaration.

The results

In testing, DeclAI replaced the manual steps with one automated flow that ends in a checked declaration.

  • Brokers review and confirm the extracted data instead of retyping it from invoices and packing lists
  • HS classification starts from a suggested code, so expert time goes into checking rather than searching
  • Sanctions and compliance problems come up while the declaration is being prepared, before it is submitted

Planning something similar?

Tell us about the process or system you have in mind, and we will suggest where to start and what a first version would include.

Contact us