Free AB-731 sample questions with answers
Here are five free AB-731 sample questions covering all three skill areas, with the answer and the reasoning behind it. Try each one before opening the answer. These are original scenario questions written against the official skills outline, not questions from the real exam.
Identify the business value of generative AI solutions
Question 1. A company pilots a generative AI assistant that summarises long contracts. The monthly bill is far above the forecast, even though the number of users is as planned. Which factor most directly explains the usage cost?
- A. The number of users who were given licences
- B. The colour theme and layout of the application
- C. The number of tokens processed in prompts and responses
- D. The department that sponsors the pilot
Show answer
Answer: C
Generative AI services are typically billed by the amount of text processed, measured in tokens, both in the prompt and in the response. Long contracts mean many input tokens per request, so volume drives cost. The number of licensed users was as planned, so it does not explain the overrun. The colour theme of the app and the department that owns the pilot have no effect on usage cost.
Want more questions like this? Full AB-731 practice tests →
Question 2. A chatbot answers employee questions fluently but sometimes states leave-policy details that do not exist in any company document. Leadership wants answers based on the current policy documents. What approach should they require?
- A. Ask the assistant to give longer, more detailed answers
- B. Ground the assistant on the policy documents by using retrieval-augmented generation
- C. Switch the assistant off for all employee questions
- D. Replace it with a larger pretrained model and change nothing else
Show answer
Answer: B
Grounding the model on company documents with retrieval-augmented generation (RAG) lets it retrieve the relevant policy text and answer from it, which reduces fabricated answers and keeps content current. Asking for longer answers does not make them more accurate. Switching off the assistant gives up the value instead of fixing the cause. A larger pretrained model alone still has no knowledge of private, changing policies.
Want more questions like this? Full AB-731 practice tests →
Identify benefits, capabilities, and opportunities for Microsoft’s AI apps and services
Question 3. A finance director wants staff to ask questions that draw on their own emails, meetings and files, with each person seeing only what they already have access to. No development effort is available. Which solution fits best?
- A. Microsoft 365 Copilot
- B. A custom model built from scratch in Microsoft Foundry
- C. A public consumer chatbot used from a personal account
- D. A spreadsheet macro that searches mailboxes
Show answer
Answer: A
Microsoft 365 Copilot grounds its answers in work data through Microsoft Graph and respects each user’s existing permissions, and it needs no code. A custom model built in Foundry would require development effort that is not available. A public consumer chatbot has no access to the tenant data. A spreadsheet macro is not a generative AI solution and cannot reason over mail and meetings.
Want more questions like this? Full AB-731 practice tests →
Question 4. A retailer wants a customer-facing app that answers questions from its product catalogue, using a model the company selects, with scalable and secure hosting. A developer team is available. Which combination is the best fit?
- A. Microsoft 365 Copilot Chat for employees
- B. Researcher in Microsoft 365 Copilot
- C. Analyst in Microsoft 365 Copilot
- D. Microsoft Foundry with Azure AI Search
Show answer
Answer: D
Microsoft Foundry provides model choice and a scalable, secure platform, and Azure AI Search can index the catalogue so answers are grounded in it. Copilot Chat for employees is not a customer-facing app. Researcher is a Copilot agent for deep research on work and web data, not a hosting platform. Analyst analyses data for users in Copilot and does not serve a public application.
Want more questions like this? Full AB-731 practice tests →
Identify an implementation and adoption strategy for Microsoft’s AI apps and services
Question 5. A Copilot pilot succeeded in one team. Six months later, most other departments barely use it. Leadership asks for the step that best addresses this. What should they do?
- A. Purchase more licences for every employee immediately
- B. Establish an AI champions programme across departments
- C. Ban other AI tools and stop all communication about Copilot
- D. Wait a year and review again without any action
Show answer
Answer: B
An AI champions programme places trained, enthusiastic peers inside each department to share practical use cases and support colleagues, which directly tackles low adoption. Buying more licences does not create usage where people do not see the value. Banning the tool does not drive adoption. Waiting without action leaves the barriers in place.
Want more questions like this? Full AB-731 practice tests →
What to do next
If you missed question 1 or 2, read the business value guide. Questions 3 and 4 map to the Microsoft AI apps guide, and question 5 to the adoption guide. When you are ready for more, try the 20-question practice test.