Invoice Indexing and Business Expense Management using Generative AI

Devadarshini, M and Karthikeyan, A S and Manimegalai, R (2025) Invoice Indexing and Business Expense Management using Generative AI. 2025 International Conference on Next Generation Computing Systems (ICNGCS). pp. 1-7.

[thumbnail of Invoice Indexing and Business Expense Management using Generative AI.pdf] Text
Invoice Indexing and Business Expense Management using Generative AI.pdf - Published Version

Download (289kB)

Abstract

Enterprise Resource Planning (ERP) tools are essential for both large and small, but growing organizations. ERP is basically business management information systems that support the day-to-day operations of an organization from end- to-end. Spend management or expense management is part of an ERP toolset, and can help companies reduce costs, streamline processes by providing all the necessary tools to control expenses, including travel and reimbursements. In addition to supporting the entire expense process, from receipt capture to card controls, these solutions also integrate with other systems, such as ERP accounting software. The in class integrations allow the data to flow directly from the expense management software into the ERP without data entry or multiple imports and exports. In this work, an application is developed to classify the invoices submitted by employees in an organization. The proposed system in this work uses a Small Language Model (SLM), making it ele-gant and easy the for employees to submit their expense reports. With the elimination of shifting between applications, this system therefore makes the process quicker and saves more time and effort in processing. A bot is developed as part of this work, to assist the user when navigating through submitting details such as the type-of-expense and the receipt images. In this regard, the SLM automatically categorizes expenses by separating travel, accommodation, and meal types. This intelligent categorization minimizes errors on the part of the system user and saves the burden of manual classification by the users themselves. Moreover, it is equipped with Optical Character Recognition (OCR), where the uploaded receipts are scanned to extract text from there. Notifications are sent, to speed up the approval, allowing managers to review and approve submissions quickly without unnecessary delays. In nutshell, the proposed system streamlines expense reporting by automating categorization and speeding up the approval workflow. This work minimizes manual data entry, eliminates errors, and increases the productivity of staff and management, allowing them to have time and space for high value work.

Item Type: Article
Subjects: Computer Science and Engineering > Artificial Intelligence
Computer Science and Engineering > Computer Networks
Divisions: Computer Science and Engineering
Depositing User: Dr Krishnamurthy V
Date Deposited: 17 Dec 2025 08:32
Last Modified: 17 Dec 2025 08:32
URI: https://ir.psgitech.ac.in/id/eprint/1571

Actions (login required)

View Item
View Item