ACCOUNTGUARD AI as Continuous Accounting Intelligence for Early Revenue Fraud Detection in Cashless Companies

Purpose: This study develops ACCOUNTGUARD AI (Accounting Continuous Intelligence for Revenue Fraud Detection), a continuous accounting intelligence model based on artificial intelligence designed to detect fraud early in the revenue cycle of digital companies operating in a cashless society. 
Research Methodology: The study uses a conceptual design approach based on a literature review of the revenue cycle, fraud theory, continuous accounting and continuous auditing, artificial intelligence for fraud detection, and the digital payment ecosystem. The review is used to map fraud red flags at each revenue cycle stage and to design a model architecture and a conceptual dashboard prototype. 
Results: The model consists of four layers, namely a Data Integration Layer, a Feature Engineering Layer, a Hybrid Fraud Detection Engine that combines rule based detection, unsupervised anomaly detection, supervised classification, and graph analytics into a risk score from 0 to 100, and a Human in the Loop Reporting and Escalation Layer. The model monitors every stage from sales order to refund and covers the entire transaction population in real time.
Conclusions: ACCOUNTGUARD AI has the potential to accelerate and improve the accuracy of early fraud detection while keeping auditors and management as the final decision makers.
Limitations: The model is conceptual and has not been tested with real or simulated transaction data, so its accuracy, cost efficiency, and false positive rate remain unmeasured.
Contributions: The study integrates artificial intelligence with revenue cycle controls and revenue recognition principles and offers guidance for companies, auditors, regulators, and system developers.