Meridian wants a system that autonomously determines the best sequence of gate reassignments, crew swaps, and delay absorptions to minimize total passenger disruption during a weather event, adjusting its plan as new constraints appear. Which pattern best fits, and why?

CPMAI Task 2 lists configuring goal-driven systems for complex problem-solving as an explicit enabler, and this scenario — an evolving, constraint-heavy sequence of decisions (gate reassignment, crew swaps, delay absorption) aimed at a stated objective (minimizing disruption) — is the defining shape of the goal-driven pattern, distinct from a single classification or a one-shot response. The 'conversational pattern' distractor again confuses a downstream human communication step (reading the plan to passengers) with the underlying AI task, which is optimization/planning under constraints, not dialogue. The 'hyper-personalization' distractor mistakes a side effect (different passengers end up with different outcomes because their situations differ) for the defining mechanism; hyper-personalization specifically means tailoring output to individual preferences or profiles, not simply that different inputs produce different outputs, which is true of almost any system. The 'recognition, because it must recognize gates in a database' distractor trivializes a lookup step into the core pattern, when the actual hard problem is the multi-step decision optimization under shifting constraints — recognition/anomaly detection is a different pattern entirely. Correctly identifying this as goal-driven tells the PM to staff for optimization/planning expertise, not conversational-AI or fraud-detection expertise.