Public systems increasingly face crises such as natural disasters, pandemics, infrastructure failures, cyberattacks, and socio-political disruptions. Traditional crisis-response systems—including emergency services, municipal authorities, and disaster management bodies—often struggle with information overload, slow decision cycles, and coordination failures. Artificial Intelligence (AI) offers transformative capabilities for improving situational awareness, forecasting risks, optimizing resource allocation, and supporting real-time decision-making. This paper proposes an AI-Enhanced Crisis-Response Decision Architecture (AICRDA) for public systems that integrates predictive analytics, real-time sensing, multi-agent coordination, geospatial intelligence, and human-in-the-loop controls. Conceptual figures, system diagrams, and structured models illustrate how AI can support crisis-preparedness, response, recovery, and resilience. The study contributes a comprehensive architecture for public sector crisis governance and proposes future research directions.