<![CDATA[Student admission management in higher education increasingly depends on data-driven. Systems that can support predictive decision making beyond descriptive dashboard reporting. Existing Business Intelligence dashboards commonly display admission indicators such as leads, applicants, registrations, targets, program distribution, and historical trends, but they often do not include predictive processing, risk classification, warning generation, or recommendation outputs. This study aims to design an Artificial Intelligence (AI) Business Intelligence architecture for predictive student admission decision support. Using a design oriented model development approach, the proposed architecture integrates admission data collection, preprocessing, Prophet based forecasting, risk scoring, Representational State Transfer (REST) API communication, and dashboard wireframe design. The system architecture consists of Vue.js as the presentation layer, Laravel as the application and API layer, MySQL as the data layer, and Python as the predictive analytics layer. Admission records are transformed into forecasting-ready datasets., processed through Prophet time series prediction, classified into program level risk categories, and converted into warning and recommendation outputs. The proposed model contributes a technical data pipeline and modular predictive workflow that can extend existing descriptive admission dashboards without replacing the entire information system. The architecture aligns with SDG 4 on Quality Education by supporting informed admission planning and with SDG 9 on Industry, Innovation and Infrastructure by promoting modular digital innovation in higher education. Future research should implement the architecture using longitudinal admission datasets, compare Prophet with Autoregressive Integrated Moving Average (ARIMA), Random Forest, and XGBoost, evaluate forecasting accuracy, test API performance, and assess dashboard usability in operational admission management.]]>
