Challenge
One of kt.team's clients, a large logistics company, processes thousands of shipments every day. This is a complex business process that involves multi-page paper document packages, digitizing them, and sending them to several information systems used by the company.
It was decided to build the automated document package verification system in Python, with support for machine learning and computer vision technologies.
Solution
We needed to develop a solution that would automate paper document processing, integrate into all of the client's business processes, and eliminate manual operations from the workflow.
Client challenge
Our clients had an entire department of around 50 people that processed document packages manually.
The main goal is to automate processes and reduce staff workload on paper and digitized document processing by more than half.
As part of the project, iCdocs was developed - an intelligent system for automated verification and sorting of document packages, built with Python.
What iCdocs checks
Digitizing paper documents
The operator loads the document package into the scanner. iCdocs starts the scanning cycle.
Converting images to text
Converts scanned images to text, identifies the document type, and determines the number of pages.
Document verification
The operator can assess whether the data on the documents has been recognized correctly and mark fields that do not match reality.
Document package assembly
Reads the legal entity name and recognizes the order and shipment number. Retrieves information about counterparties, deals, and deliveries from accounting software. Checks whether page numbers are present and how many pages the document contains, then combines all pages into a single document. Groups documents with a common attribute (invoice number, trip number, counterparty) into one package.
We'll curate materials for your task
We'll reply within 30 minutes and send relevant cases, diagrams, or analyses tailored to your context.
Send to external applications
If needed, iCdocs can be integrated into any company document management system and used to send document packages to accounting applications and repositories, both CIS and international.
Document storage
iCdocs stores backup copies of the documents processed in it, with the ability to view the change history and who made each change.
Development results
The automated document package verification system can process several document packages at the same time. At the start of development, we were able to reach about 80% recognition accuracy.
Work on the project is still ongoing, and we plan to improve the recognition algorithms, getting as close to 100% as possible.
Many paper documents
With different counterparties, both EDI and paper document workflows are used. Everything needs to be brought into a single system to make control easier.
Many orders every day
Each order comes with a document package. Incoming packages must be checked quickly for completeness, verified, sorted, and passed on for further processing.
Multiple legal entities
Several contracts have been signed with partners for different legal entities. The completeness of each package must be checked for every contract, as well as the accuracy of each document (the legal entity, seal, and signature match what is specified in the contract).
Multiple contracts
Several contracts have been signed with each partner, with varying terms of cooperation (different carriers, different payment methods or approaches, etc.).
Many suppliers
You work with hundreds or thousands of counterparties. For each one, a large document package must be checked.
iCdocs is built with Python
A key advantage of projects developed in Python is the ability to implement machine learning and computer vision.
Python is a versatile, object-oriented programming language.
More about Python
