Instruction: Describe the components and control points for a low-risk internal agent that can escalate before acting externally.
Context: Assesses whether the candidate can design a practical architecture and explain the main tradeoffs. Describe the components and control points for a low-risk internal agent that can escalate before acting externally.
Official answer available
Preview the opening of the answer, then unlock the full walkthrough.
I would split the system into three layers: knowledge retrieval, response synthesis, and action gating. The retrieval layer handles document access, permissions, and freshness. The response layer answers grounded questions and proposes next actions. The action layer requires explicit human approval before anything leaves...
easy
easy
easy
easy
easy
easy