How to Design an AI Agent: Architectures, Protocols, and Technical Evaluation of Agentic AI Systems for Law & Finance

Agents are not magic; they are architecture. In high-stakes domains like law and finance, the difference between a reliable tool and a runaway process lies in design. This chapter focuses on the architectural principles required to allow agents to function as cognitive work systems analogous to, and in concert with, professional teams. We organize our analysis around ten fundamental questions that shape an agent's operational reality. These range from input mechanisms (triggers, intent, perception, and memory) to execution strategies (planning, delegation, and action tools). Crucially, we examine the safety layers required for professional deployment: termination conditions, human escalation protocols, and governance. Behind each of these ten questions lies a design decision with real tradeoffs. These choices determine what a system can do, how reliably it performs, and how it fails. Ultimately, this chapter argues that robust design requires architectural literacy—a necessary bridge between technical implementation and professional obligation.