A user messages an HR AI assistant: 'I need help with benefits.' The NLP model returns three intents with similar confidence scores: health_insurance_benefits at 42%, retirement_plan_benefits at 38%, and time_off_benefits at 20%. What is the BEST disambiguation design strategy?

When intent confidence is split across multiple plausible options, best practice is to surface a clarifying question presenting the options explicitly. This preserves automation while ensuring the correct path is taken. Silent defaulting to the top intent risks providing an entirely wrong answer. Immediate live-agent escalation for routine ambiguity is premature; most disambiguation resolves with a single clarifying turn. The correct answer, "Present the user with a concise clarifying question that lists the specific benefit categories so the user can identify their exact need", directly satisfies the scenario because it aligns with watsonx Assistant's design principles and the specific capability being tested. The incorrect options ("Randomly select one of the three matched intents and respond as if it is correct, then offer to correct if the user complains", "Silently default to the highest-confidence intent (health_insurance_benefits) and respond without informing the user of the assumption", "Immediately escalate to a live HR agent since the confidence scores are too low for any automated handling") may appear relevant but each misses a key requirement or introduces a step that is either unnecessary or belongs to a different workflow. Mastering the distinction between these approaches is essential for effective watsonx Assistant implementations and is a core focus of the Domain 2: Conversational AI Design section of the certification exam.