University of Missouri Cuts Invoice Processing Time 80% with OCI GenAI-Powered Automation

Problem:

Facing the challenge of managing thousands of invoices across a decentralized multi-campus system, University of Missouri moved beyond traditional OCR into intelligent document understanding.

Solution:

By leveraging OCI’s Large Language Models (LLMs) and specialized GenAI Vector Search, the system now performs high-precision data extraction and autonomous 2-way and 3-way matching against purchase orders and receipts.

This allows the University to process complex, unstructured invoice data with minimal human intervention, effectively turning a legacy cost center into a streamlined, intelligence-driven operation.

Impact:

The impact of this digital transformation has been both immediate and scalable.

Since the deployment, the University has achieved an 80% reduction in invoice processing cycle times, shifting the focus of the Shared Procurement and Payment Services (SPP) team from repetitive data entry to strategic exception management.

This increased velocity has enabled the University of Missouri to consistently capture early-payment discounts and improve vendor relationships through real-time visibility into the payment lifecycle.

Implementation Details:

The architecture at the University of Missouri leverages a sophisticated multi-service pipeline within OCI to achieve near-perfect extraction accuracy. Invoices are ingested into OCI Object Storage, triggering an automated workflow where OCI Document Understanding performs initial layout analysis and text extraction.

This is then enhanced by OCI Generative AI (powered by Cohere/Llama models), which uses curated prompts to interpret complex line items and unstructured data that traditional OCRs typically miss.

This hybrid approach allows the system to differentiate between various billing types and "read" the intent of the document, ensuring that only high-confidence data is pushed to the Oracle Autonomous Database for final validation.

By utilizing Oracle APEX as the low-code interface for exception management, the solution provides a seamless bridge between AI insights and human oversight. The integration with the University’s ERP is handled via OCI API Gateway, ensuring secure, real-time data flow for 2-way and 3-way matching.

This architecture creates a self-learning financial engine that utilizes AI Vector Search to cross-reference historical billing patterns, effectively preventing duplicate payments and fraud before they occur.

Conclusion:

By choosing a solution natively integrated with their existing Oracle ecosystem, the University of Missouri has established a robust, AI-enabled foundation for financial operations that supports their mission of research and academic excellence while setting a new benchmark for administrative innovation in higher education.