One of kt.team clients is a major logistics company that processes thousands of shipments a day. This is a complex business process that involves working with multipage packages of paper documents, their digitization and sending to several information systems used at the company.

Our task was to develop a solution that would automate the processing of paper documentation and would be integrated into all the client's business processes in order to reduce manual effort in documentation management.

We decided to develop a system for automated verification of documentation packages in Python with an opportunity to implement machine learning and computer vision technologies.

Our client even had a separate department (about 50 employees) that was processing document packages manually.

The main task was to automate processes and simplify the human efforts of paper and digitized documents processing by more than a half.

As part of the project, we developed an intelligent iCdocs system for the automated verification and sorting of document packages developed in Python.

The operator uploads the documentation package into the scanning device. iCdocs launches the scan cycle.

Converts scanned images into text, defines the type of a document and the number of pages.

The operator has the opportunity to assess the correctness of data recognition in the document and mark the fields that are not correctly processed.

Recognizes the legal entity's name, order or shipment numbers. Receives information about counterparties, transactions, and deliveries from accounting programs. Checks the presence of page numbering and the total number of pages in the document, combines all the pages of the document. Combines documents with a single attribute (invoice number, trip number, counterparty) into one package.

If necessary, iCdocs is integrable with any document management system used in the company. It is also capable of sending documentation packages to both Russian and international accounting programs and repositories.

iCdocs backup copies of documents that it has processed previously, providing an opportunity to track the history of changes and see who made them.

The automated document package verification system can process several document packages simultaneously. At the start of development, we managed to achieve the 80% recognition rate.

We keep working on the project and we plan to improve the recognition algorithms to be as close to 100% result as possible.

Both electronic and paper document management are applied when working with different contractors. We need to bring everything into a single system to facilitate the management process.

Each order comes with a package of certain documents. It is necessary to promptly assess their correctness, sort them out, and hand off for further work.

You have signed several contracts with partners for different legal entities, and you need to check the completeness of the packages for each contract separately and make sure each document is filled in correctly (the legal entity, seal, signature correspond to the contract requirements).

Several contracts have been concluded with each of the partners, and the terms of cooperation are different (different deliverers, different methods or approaches to payment, etc.).

You work with hundreds or even thousands of contractors. For each contractor, you need to check a large package of documents.