AUTOMATION OF THREE-WAY MATCHING FOR BILL PAYMENT USING LARGE LANGUAGE MODELS

Manasa Varsini S , Emimanimcy M

Accounts Payable (AP) Three-Way Matching Procurement Workflow Bill Payment Verification Large Language Models (LLMs) Optical Character Recognition (OCR) Financial Document Automation Invoice Processing Purchase Orders (POs) Goods Receipt Notes (GRNs) Document Data Extraction Unstructured Data Processing Discrepancy Detection Payment Authorization Tolerance Thresholds Audit-Ready Reporting Real-Time Dashboards Procure-to-Pay (P2P) Enterprise Automation Financial Compliance

The accounts payable process in most manufacturing and service organizations still relies heavily on manual verification of financial documents, which is both time-consuming and prone to errors. Three-way matching, which involves cross-verifying purchase orders (POs), goods receipt notes (GRNs), and invoices before authorizing payment, is a critical control mechanism in procurement workflows. However, the diversity in document formats and the presence of unstructured data make automation of this process a persistent challenge. This paper proposes a system that leverages Large Language Models (LLMs) to automate three-way matching for bill payment verification. The proposed approach uses LLMs in combination with Optical Character Recognition (OCR) to extract relevant fields from documents in varying formats, including scanned PDFs and images. Once extracted, the data from POs, GRNs, and invoices is systematically compared to identify discrepancies in quantities, pricing, item descriptions, and totals. Bills are then classified as payable or non-payable based on predefined tolerance thresholds. The system also generates real-time dashboards and audit-ready reports for finance teams. By reducing manual intervention, this approach aims to minimize payment errors, speed up invoice processing cycles, and strengthen compliance across the procure-to-pay pipeline. The results demonstrate that LLM-based extraction and matching significantly outperforms traditional rule-based systems in handling diverse and unstructured document formats, making it a scalable solution for enterprise-level accounts payable operations.

Volume 14 Issue 2, February-2026

Pages : f826-f831