Domain 2: Conversational AI Design
IBM Certified Specialist - watsonx Assistant · 14 questions
- A retail bank wants to deploy an AI assistant to handle customer inquiries. The project team has been assembled and is ready to begin the design process. What is the MOST important first step in defining conversational requirements?
- A healthcare company is planning to deploy watsonx Assistant to handle patient appointment scheduling. During requirements gathering, which artifact would BEST capture the full range of user behaviors, including error scenarios and alternative paths?
- In IBM watsonx Assistant, a designer is building a travel booking assistant. The Book a Flight capability must collect origin city, destination city, travel date, and passenger count before executing a search. How should this capability be structured?
- A customer service assistant for an insurance company handles both new claim submissions and existing claim status checks. Users often begin with ambiguous messages like 'I need help with a claim.' What is the MOST appropriate design approach for routing these users correctly?
- A designer building an e-commerce assistant has identified four candidate intents: trackorder, orderstatus, whereismypackage, and deliveryupdate. Analysis shows all four represent users wanting to know the current status or location of their shipment. What is the BEST practice for the intent taxonomy?
- An IT helpdesk assistant handles requests such as 'reset my password,' 'change my password,' and 'I forgot my password.' The designer needs to capture the specific action the user requires. Which entity type BEST handles these surface-form variations?
- A financial services company has deployed watsonx Assistant for customer support. The design team must specify which conditions should automatically trigger a live agent handoff. Which scenario MOST warrants escalation to a human agent?
- When IBM watsonx Assistant escalates a conversation to a human agent, the designer must define what context is transferred. Which approach BEST ensures a seamless handoff experience for both the customer and the receiving agent?
- A bank is designing an AI assistant named Aria for their mobile banking app. The bank's established brand values are trustworthy, approachable, and efficient. Which persona design guideline BEST aligns Aria with all three brand values?
- An e-learning platform has deployed a watsonx Assistant to help students with course navigation and technical issues. A student sends the message: 'I have been trying to submit my assignment for over an hour and nothing is working. I am completely frustrated!' What tone should the assistant's response adopt?
- A company is designing an AI assistant in IBM watsonx Assistant to guide new customers through account onboarding. The flow follows a consistent sequence: collect name, email address, preferred contact method, and a security question answer. The process is the same for all users. Which design approach is MOST appropriate?
- A user interacting with an insurance company's AI assistant asks: 'What is the best stock to buy right now?' This topic is entirely outside the assistant's defined scope. What is the BEST fallback response design for this scenario?
- A user messages an HR AI assistant: 'I need help with benefits.' The NLP model returns three intents with similar confidence scores: healthinsurancebenefits at 42%, retirementplanbenefits at 38%, and timeoffbenefits at 20%. What is the BEST disambiguation design strategy?
- A design team is preparing documentation for a watsonx Assistant deployment at a B2B software company. The assistant will handle customer technical support inquiries. The team needs an artifact that clearly defines what the assistant will handle and what it will not. Which artifact is MOST effective?