AI & Analytics: From Prediction to Decision

Overview Build an AI-supported business or system that can withstand the consequences of its decisions. The AI & Analytics track at CaseUnion 4.0 explores how technology can address difficult operational and access problems. The submitted case pack contains four challenges. The Fraud War Room Redesign a bank or payment service provider’s UPI fraud detection and investigation workflow over a 90-day planning horizon. Connect detection to response, prioritise investigation and explain the tradeoffs between missed fraud and incorrectly flagged customers. The Marketplace That Fixed the Wrong End Design a claims-first AI health insurtech. Choose and defend a business model, explain the claims workflow and model its unit economics. Address human review, false rejections, appeals and hospital gaming. The Mandi Decision Design an AI advisor that helps smallholder farmers decide whether to sell, store, transport or contract-sell their harvest. Define the crops, states, delivery channels and revenue model. Model economics at a scale of 100,000 farmers and address the downside when recommendations fail. The Justice Gap Build a focused AI legal services company around one user group, one problem and one business model. Define the role of AI and licensed professionals, select two deployment states and establish a path to break-even within 24 months. Submission Expectations Follow the requirements of the assigned case. Explain the product or system, data requirements, operational workflow, economics and risk controls. Rules Clearly distinguish case inputs, external evidence and assumptions. Justify the relevant tradeoffs and specify where human oversight is required. Process, Format and Eligibility The final case allocation or selection process, rounds, team size, eligibility, deadlines and submission format will be announced in the competition instructions.