01 / Identify and discloseSet expectations immediately
Use the firm’s name, explain that the caller is speaking with an AI assistant when applicable, and state what the assistant can do next: collect details, help schedule, or route the request. Clear disclosure is better than a caller discovering automation halfway through a sensitive conversation.
02 / Capture the essentialsAsk only for information the next person needs
Most workflows need a name, a reliable callback method, a concise matter description, relevant timing, and a preferred next step. Each practice should approve its own questions and avoid asking for unnecessary sensitive information before the firm has determined how to proceed.
03 / Follow the firm’s boundariesRoute, do not adjudicate
Define what an assistant may explain, what it must never promise, which calls require urgent human escalation, and where it should stop. An intake workflow can recognize configured categories and collect details; it should not give legal advice, assess a claim’s merits, or make a representation decision.
04 / Create a concrete next stepBook, transfer, or hand off
The best next step is specific: a confirmed consultation within the firm’s availability, a warm transfer when someone is on call, or a summarized callback task assigned to the right person. If availability or routing cannot be confirmed, tell the caller exactly what follow-up to expect rather than creating a vague promise.
05 / Review the outcomesTune the workflow with real calls
Review call summaries and recordings where appropriate, booking accuracy, escalations, response time, and the percentage of qualified conversations that become consultations. Use those findings to improve questions, availability, routing, and handoff language.