The article introduces the AI delegation matrix, a framework to help designers determine when to delegate tasks to AI, when to assist human users, and when to keep decisions human-led. It highlights the shift in design philosophy prompted by AI capabilities, emphasizing the need to rethink who controls the decision-making process in user interfaces. The author argues that merely adding AI functions to existing tools isn't effective; instead, designers must clarify the role of AI in each specific context. The matrix categorizes tasks into three control modes: Human-Led, Assist, and Delegate. In the Human-Led mode, the human maintains full control, using AI to provide insights and structure but making the final decision. This approach is best for high-stakes scenarios that require ethical judgment. The Assist mode allows AI to handle heavy lifting—like generating drafts or analyzing data—with the human reviewing and approving actions before they are finalized. Lastly, the Delegate mode allows AI to execute tasks independently within set parameters, shifting humans to an oversight role. To implement this framework, the article presents a scoring model that assesses the feasibility and value of automating tasks. It considers factors like frequency of occurrence, data readiness, and AI proficiency. By plotting scores for Automation ROI against Automation Suitability, designers can effectively categorize workflows and decide the most appropriate control mode. Each category—Delegate, Assist, or Human-Led—directs how to structure the user interface and the interactions between humans and machines.