RPA or Generative AI? A Comparative Review of Accounting Automation in Accounts Receivable and Expense Reimbursement

Mainstream accounting automation technologies are mainly divided into robotic process automation (RPA) and generative artificial intelligence (GenAI). RPA is a relatively mature technology that has been widely used, especially for standardized and repetitive tasks, whereas GenAI is an emerging technology that has demonstrated strong potential and remarkable capabilities. This paper employs a structured literature review approach, primarily covering relevant literature from 2020 to 2025, and compares the application performance of RPA and GenAI in accounts receivable and expense reimbursement processes. Regarding accounts receivable, efficiency, accuracy, and reliability are used as key evaluation criteria. For expense reimbursement, the comparison emphasizes flexibility, error-handling capabilities, and user interaction. The findings indicate that RPA demonstrates superior consistency and stability in standardized, repetitive task processes, remaining dominant in tasks such as invoice matching and cash receipts. GenAI significantly outperforms RPA in ambiguous scenarios that involve unstructured receipts, policy interpretation, and anomaly detection. The study concludes that a hybrid adoption strategy based on task characteristics and data structure is more effective than relying on either technology alone. This research provides empirical guidance for accounting practitioners in technology selection and highlights the importance of developing RPA-GenAI hybrid architecture in future accounting functions.