Murugan Venkatachalam
Abstract Internal audit functions in U.S. manufacturing and retail face a growing disconnect between increasingly sophisticated fraud schemes and legacy detection methods that rely on static rules and manual sampling. The Association of Certified Fraud Examiners estimates that organizations lose approximately five percent of annual revenue to fraud, with manufacturing and retail sectors particularly vulnerable due to complex supply chains, high transaction volumes, and decentralized operations. Modern fraud schemes have evolved well beyond simple expense manipulation; they now involve multi-party collusion, cyber-enabled invoice fraud, supply chain manipulation through fictitious vendors, and coordinated point-of-sale skimming networks. Traditional audit approaches typically cover only three to five percent of transactions through periodic sampling, leaving the vast majority of activities unexamined and creating significant windows of exposure. Learningter demonstrates how applied AI (machine learning anomaly detection, natural language processing, and agentic AI) transforms fraud detection from reactive forensics into proactive, continuous assurance. Drawing on five anonymized case studies from active industry engagements, the presentation illustrates measurable outcomes: false-positive rates reduced by up to 70 percent, detection time compressed from months to minutes, and coverage expanded from sample-based testing to full-population analysis. Each case maps legacy controls against AI-augmented alternatives, providing a clear migration pathway. In particular, Agentic AI enables autonomous and continuous monitoring through self-correcting feedback loops that recalibrate detection models in real time without requiring manual intervention, adapting dynamically to emerging fraud patterns and shifting transaction behaviors. The presentation addresses practical adoption challenges (data quality, algorithmic bias, SOX/ICFR compliance, and change management) and offers a structured readiness framework for consulting engagements or dissertation research. Grounded in Boyer's Scholarship of Application, this work connects data science and auditing to real-world problems, demonstrating how cross-disciplinary collaboration produces actionable improvements in governance and risk management. The research is directly relevant to doctoral candidates seeking applied dissertation topics with measurable industry impact and to faculty developing curricula that bridge theoretical foundations with practitioner-oriented pedagogy.