This paper presents a comprehensive survey of auditing and governance challenges in large language model (LLM)-integrated decision systems within customer relationship management (CRM) platforms. As CRMs evolve into AI-driven decision engines, LLMs increasingly influence workflows such as lead qualification, customer engagement, and service automation. This study examines limitations in traditional audit mechanisms, highlighting gaps in transparency, traceability, and accountability of AI-assisted decisions. It reviews existing frameworks, monitoring techniques, and governance approaches, and identifies key risks related to bias, hallucinations, and decision opacity. The paper outlines future directions for building auditable, risk-aware, and trustworthy AI systems in enterprise CRM environments.
Sri Krishna Sai Sameera Polapragada (Fri,) studied this question.