This chapter translates the theoretical foundations of Agentic Artificial Intelligence (AI) into concrete architectural design principles. It presents a modular view of agentic systems, outlining the functional roles of perception, reasoning, learning, communication, and control within individual agents and across agent collectives. The chapter explores orchestration strategies that enable coordination among heterogeneous agents while maintaining decentralised autonomy. Design considerations such as scalability, robustness, and adaptability are examined, with particular attention to how language-based reasoning and learning components are embedded within agent architectures. It provides a systematic framework for designing agentic systems that support long-term operation, collaborative problem-solving, and dynamic reconfiguration, along with detailed descriptions of the different frameworks.