AI Financial Reimbursement System for Universities Based on “Rule Engine + Foundation Model” A Closed-Loop Design from Burden Reduction to Self-Service

Research areas:

• Volume 14,Issue 3, Nov. 2025

Year:

2025

Type of Publication:

Article

Keywords:

AI Financial Reimbursement System, Rule Engine, Foundation Model, University Finance

Authors:

Qunying He; Leman Huang; Xi Zhang

Journal:

IJAIM

Volume:

14

Number:

3

Pages:

11-23

Month:

November

ISSN:

2320-5121

Abstract:

To address the long‑standing problems of “high repetition, high return rates, and high costs” in traditional university expense reimbursement, this paper proposes an AI‑driven reimbursement system for higher education with the dual goals of “reducing the financial workload” and “faculty self‑service.” The system adopts a “rules engine + large model + encrypted knowledge base + human fallback” architecture. Methodologically, we implement a “4+2” overall design-four technical layers (front‑end conversational interaction, back‑end asynchronous pre‑audit, encrypted knowledge base, and human fallback) that serve two user groups (finance staff and faculty). The rules engine handles rigid compliance checks, the large model tackles flexible needs such as complex risk identification and semantic understanding, and the human fallback serves as the final gatekeeper for special scenarios, complicated vouchers, and system testing. Implementation and measurement show that the system can automate over 85% of repetitive auditing tasks on the finance side, cutting the average processing time per claim from 1.5 days to 15 minutes, with 80% of low‑risk vouchers paid straight‑through. On the faculty side, users complete self‑service reimbursement via a conversational interface, which markedly reduces return rates and the cost of policy comprehension. In addition, we establish a closed‑loop mechanism of “policy learning - model updating - results feedback - iterative optimization,” enabling continuous improvement. This exploration offers a practical technical pathway and implementation plan for deploying and iterating AI reimbursement systems in universities.

Full text: IJAIM_689_FINAL.pdf