Extraction of invoice data via the INVOICE module

Client

Capital group managing real estate in Poland and Western European countries

Expectations

Relieving the day-to-day manual work of central secretarial staff in the mass processing data from incoming invoices and accounting notes. Using advanced tools to optimise this process and automatically (without employee participation) transfer critical accounting data to the EDM class system.

Project

The project included a stage of in-depth analysis of the documents being processed (including their types and the structure of the data they contain, the image quality of the scanned paper invoices, the channels of entry into the process, and the target needs for changing the current interface of the EDM system).

  • checked

    testing and implementation of optimal algorithms for autonomous extraction of selected accounting data from invoice images,

  • checked

    changing the current EDM system interface to enable automatic filling of accounting data from the extraction process

The real challenge was to use the most efficient algorithms from the field of machine learning so that they could adapt to new invoice templates and patterns autonomously. It can extract the desired accounting data (such as document issue dates, sales dates, payment dates, contractor names, net and gross amounts, VAT rates, currency type, bank account number, and more).

Results

The effectiveness of the implementation of FINTURE's proprietary solution was confirmed by two months of algorithm testing in the target environment (in expert supervision mode by designated client employees). The expected effectiveness of the INVOICE Reader Module exceeded the customer's baseline requirements, achieving a stable accuracy of over 85% for all processed documents and their extracted accounting fields.

Info

  • Sector

    Real Estate

  • Service

    Machine Learning

  • Technologies

    NLP Libraries,
    Neural networks

    • Key figures

    +2000​

    accounting documents processed per month

    +30 000​

    number of accounting fields automatically read out per month

    2

    the number of reduced staff positions

    Competencies used

    Hyperautomation
    Digital Transformation

    Contact

    • Skierniewicka 10A, 01-230 Warsaw
    • +48 (22) 122 04 04

    © Copyright 2025. All Rights Reserved by Finture 

    Badge "Accredited Agency Designrush.com 2023"
    Codes, Guides, Policies