Published Paper Details:

Data quality poses a significant challenge for organizations deploying Customer Relationship Management (CRM) systems like Salesforce. Despite progress in data cleansing techniques, existing methodologies are prone to be dependent heavily on manual intervention or rule-based systems lacking the flexibility to learn from changing data patterns. Past research has focused mainly on static algorithms and traditional machine learning models for cleansing, which lack the ability to generalize to the dynamic and complex datasets prevalent in Salesforce environments. This limitation highlights the need for intelligent, autonomous agents with continuous learning and real-time adaptation. Autonomous AI agents with reinforcement learning and adaptive feedback capabilities hold immense potential to address these limitations by cleansing data autonomously with little or no human intervention. However, utilization of such agents specifically for Salesforce datasets has remained under-explored. This paper attempts to bridge this gap by developing a new framework where AI agents systematically find, fix, and remove data anomalies leveraging contextual knowledge of Salesforce data structures and business rules. The framework employs unsupervised learning to detect inconsistencies and reinforcement learning to improve cleansing strategies based on outcomes, promoting scalability and robustness in complex CRM datasets. Experimental testing with real-world Salesforce datasets shows spectacular improvements in accuracy, efficiency, and minimized manual effort compared to existing methodologies. This research makes contributions towards the development of autonomous data management for CRM systems towards more intelligent, self-sustaining data ecosystems that fuel decision-making and operational efficiency.