A dynamically growing international automotive supplier company employing more than 1,000 people.
Upon invoice receipt, administrators manually extract invoice data and upload it to the enterprise resource planning system. Afterwards, the posting of these invoices occupies the full working time of several accounting staff members, as thousands of invoices must be assigned account codes based on defined rules.
To replace the manual work hours, we proposed developing a software robot that can perform the workflows in the existing ERP system without significant integration costs. During development, an AI-based OCR system was created that can automatically read invoice data in a self-learning manner and pass it to the robot in a structured format. The robot then cross-references the data with the national tax authority's system, verifies it, and posts the invoices based on data from a purpose-built accounting database. The robot supports the accounting of monthly subscription invoices and is also capable of matching based on procurement records.
With the help of the robot, the accounting team was relieved of 70-80% of their tasks, so they only need to focus on the more specialized cases. The solution significantly accelerates the accounting process, as invoices are posted within minutes of receipt in the system.
For the software robot development, we used Power Automate Desktop, Python, and a MySQL database.
