A legal firm wants a drafting agent that may propose contract clause language but must never send email to counterparties or accept settlement terms. Which specification elements belong in the agent’s persona and autonomy design?
Select an answer to reveal the explanation.
Short Explanation
Correct answer: A. Persona and autonomy are the job description and the leash. Drafting-only, propose-not-send, and human approval before external or binding acts keep the agent useful without letting it become an unsupervised partner who emails the other side.
Full Explanation
Logical architecture requires specifying agent personas, scopes, boundaries, autonomy levels, and behavioral guidelines, plus tool scopes and permission boundaries. Option A correctly limits persona and tools to drafting support and enforces low autonomy with HITL before external/binding actions.
Option B violates the explicit “never send email / never accept settlement” boundaries.
Option C is incorrect: session-by-session verbal reminders are not durable architecture.
Option D is incorrect: temperature does not enforce tool authorization; unregistered or unauthorized tools must be absent or blocked by policy.
Exam tip: Autonomy levels should match action irreversibility—draft/propose high freedom, external side effects low freedom + HITL.