EVALUATING AN AI-ENABLED INTERACTIVE ANALYTICS AND CONVERSATIONAL DECISION-SUPPORT SYSTEM FOR LIVESTOCK PHARMACEUTICAL SUPPLY CHAIN MANAGEMENT IN PAKISTAN

Abstract

Livestock pharmaceutical supply chains depend on reliable information about inventory, purchasing, sales, supplier activity, and product demand. However, when operational records are maintained in separate spreadsheets, managers may face difficulties consolidating information, identifying stock imbalances, evaluating product profitability, and anticipating replenishment needs. This study presents the development of a self-service interactive analytics dashboard with a conversational assistant for livestock pharmaceutical supply chain management in Pakistan. The system was developed using spreadsheet-based records obtained from veterinary pharmaceutical wholesalers with consent. The project documentation describes three Excel files covering approximately 12 months of activity and more than 45 suppliers. Data were processed using Python-based tools, while the application combined a Flask backend with a React.js interface. Reported functions included inventory monitoring, low-stock identification, product-level sales and profit summaries, supplier analysis, and demand forecasting using an autoregressive integrated moving average (ARIMA) approach. A conversational interface was also incorporated to facilitate access to analytical information. The thesis reports illustrative outputs from the dashboard, including product stock quantities, sales volumes, and supplier-level purchasing summaries. However, the available documentation does not establish forecast accuracy, controlled usability outcomes, or measurable improvements in business performance. One exceptionally large reported profit value also requires verification before publication. The study contributes an applied system-development example for spreadsheet-dependent livestock pharmaceutical businesses and proposes an evaluation framework covering data integrity, forecast performance, usability, and operational impact. The findings should be interpreted as reported system functionality rather than proof of improved supply-chain outcomes.