Free Microsoft AB-730 Practice Questions: Generative AI Fundamentals

Topic snapshot

Sample questions

These are original IT Mastery practice questions aligned to this topic area. They are not official Microsoft questions, copied live-exam content, or exam dumps. Use them for self-assessment, scope review, and deciding what to drill next.

Question 1

Topic: Understand Generative AI Fundamentals

A sales operations analyst is using Microsoft 365 Copilot in Teams to prepare a quarterly account summary. The analyst asks Copilot to include renewal-risk notes from a private leadership channel and compensation details stored in an HR file, but the analyst does not have access to those sources. The summary must be accurate and safe to share with the account team. What is the best business action?

Options:

• A. Ask Copilot to ignore permissions for this one summary

• B. Have a colleague paste the restricted content into the chat

• C. Use only accessible sources and request approved access or an owner-provided summary

• D. Upload the restricted file to another AI tool for summarization

Best answer: C

Explanation: Microsoft 365 Copilot is designed to use organizational data within the user’s existing permission boundaries. If a user cannot access a work file, message, or private channel, Copilot should not reveal that content through a prompt response. For a business summary, the analyst should use sources they are allowed to access, ask the content owner for an approved summary, or request access through normal business processes. This protects sensitive information while still allowing Copilot to help draft and organize permitted content. Trying to bypass permissions or move restricted data into another tool creates privacy and data-protection risk.

• Ignoring permissions fails because Copilot should not override organizational access controls for convenience.
• Pasting restricted content fails because it can expose sensitive information outside the approved audience.
• Using another AI tool fails because it may move protected work data outside approved Microsoft 365 protections.

Question 2

Topic: Understand Generative AI Fundamentals

A customer success team answers the same types of renewal questions every week. They want Microsoft 365 Copilot to use the team’s approved renewal playbook, follow a consistent question-and-answer flow, and make the experience reusable by all team members. What should they do?

Options:

• A. Save one broad renewal prompt for personal use

• B. Start a new Copilot chat for each renewal

• C. Create a team agent with the playbook as knowledge

• D. Ask Copilot to summarize past renewal emails

Best answer: C

Explanation: An agent experience is useful when the task repeats, needs specialized knowledge, or should guide users through a consistent process. In this scenario, the team wants Copilot to rely on an approved renewal playbook, follow a repeatable Q&A flow, and be reusable by multiple team members. Creating an agent with the playbook as knowledge and appropriate instructions supports those needs better than a one-time chat or personal prompt. A normal chat can answer individual questions, but it does not provide the same reusable, guided team experience.

• New chat each time omits the requirement for a reusable team workflow.
• Personal saved prompt may help one user repeat a request, but it does not provide shared specialized knowledge and guided behavior.
• Email summarization addresses past communication review, not a repeatable renewal support experience.

Question 3

Topic: Understand Generative AI Fundamentals

A product marketing manager asks Microsoft 365 Copilot in Word to draft a one-page announcement for an upcoming feature release. Copilot returns a generic draft that does not match the audience or approved messaging.

Draft prompt:

Write an announcement about the new feature release.

Which additional context would best improve the prompt?

Options:

• A. Include every product document the team has created.

• B. Reference the approved launch brief, target audience, tone, and one-page format.

• C. Ask Copilot to make the announcement more exciting.

• D. Paste customer names and private account issues into the prompt.

Best answer: B

Explanation: Copilot responses are strongly affected by the context included in the prompt, such as relevant work files, audience, purpose, constraints, and desired output format. In this scenario, the generic result shows that Copilot lacks approved messaging and audience guidance. Referencing the approved launch brief and specifying the target audience, tone, and length helps Copilot ground the draft in the right business information without adding unnecessary or sensitive content.

The key takeaway is to add focused, relevant context rather than vague style requests or broad document dumps.

• Vague improvement fails because “more exciting” does not supply the missing messaging, audience, or format context.
• Overbroad references fail because adding every product document can introduce irrelevant information and reduce focus.
• Sensitive details fail because private customer information should not be added unless it is necessary, permitted, and appropriate for the task.

Question 4

Topic: Understand Generative AI Fundamentals

A customer success manager needs a one-time executive summary of recurring issues from a recent Teams meeting recap and a related Word document. The summary is due today, and the manager must stay within normal business-user tools without building an app, agent, or AI model. Which action best satisfies the requirement?

Options:

• A. Create a new agent before summarizing the documents

• B. Use Microsoft 365 Copilot to summarize the referenced meeting recap and Word document

• C. Request a custom AI model trained on all support tickets

• D. Export the files to a public chatbot for faster summarization

Best answer: B

Explanation: Microsoft 365 Copilot is designed for business users who need help with productivity tasks such as summarizing meetings, drafting content, and working across Microsoft 365 files. In this scenario, the need is immediate and one-time, and the required information already exists in a Teams recap and a Word document. The best action is to use Copilot with those relevant work resources as context, then review the output for accuracy. Building a model, creating an agent, or moving content outside approved tools adds unnecessary complexity or risk for a task Copilot can handle directly.

• Custom model targets a development or data-science need, not a same-day business-user summary.
• New agent is unnecessary when the task is one-time and can be handled by Copilot using existing content.
• Public chatbot creates a data-protection concern and does not respect normal Microsoft 365 work boundaries.

Question 5

Topic: Understand Generative AI Fundamentals

A product manager must recommend whether to expand a pilot program next quarter. The evidence is spread across a Teams meeting recap, an Excel feedback workbook, and a Word risk register. Leadership expects a balanced recommendation with source-backed reasoning, and some customer comments are sensitive. What is the best business action to improve decision quality with Microsoft 365 Copilot?

Options:

• A. Create a cited decision brief with options, assumptions, risks, and next steps.

• B. Build a custom agent before preparing the recommendation.

• C. Paste all customer comments into a shared chat for Copilot to summarize.

• D. Ask Copilot to choose the expansion decision and send it to leadership.

Best answer: A

Explanation: Copilot improves decision quality when it helps organize information into a structure that people can review: evidence, options, assumptions, risks, and next steps. In this scenario, the product manager has multiple source materials and a quality expectation for balanced, source-backed reasoning. A strong business-user action is to ask Copilot to draft a decision brief that cites or references the relevant files, then review the output for accuracy and sensitivity before sharing. Copilot should support the decision, not replace the accountable business judgment.

The key takeaway is to use Copilot as an evidence organizer and drafting partner, with human review for accuracy and sensitive content.

• Letting Copilot decide fails because the business owner still needs to verify evidence and make the accountable recommendation.
• Sharing sensitive comments broadly fails because sensitive customer information should not be exposed unnecessarily for summarization.
• Building an agent first is unnecessarily complex for a one-time decision brief using existing Microsoft 365 content.

Question 6

Topic: Understand Generative AI Fundamentals

A customer success manager needs to brief leadership before a renewal meeting tomorrow. They have a Teams meeting transcript, recent customer emails in Outlook, and a support-summary document in Word. The brief must be concise, source-supported, and suitable for sharing internally without exposing unnecessary sensitive details. What is the best business action?

Options:

• A. Use Copilot to create a cited management summary from the relevant sources, then review it for accuracy and sensitive details.

• B. Build a custom agent before preparing the renewal brief.

• C. Paste the full email thread and transcript into a shared chat for faster team review.

• D. Ask Copilot to write a renewal recommendation from memory without referencing the customer sources.

Best answer: A

Explanation: The best action is to use Microsoft 365 Copilot to turn relevant work content into a concise, source-supported management summary, then have the business user review the output. This directly supports the business outcome: leadership gets faster decision support before the renewal meeting. Because the brief may influence a customer decision, the manager should check citations or source references, confirm important facts, and remove unnecessary sensitive details before sharing. Copilot can accelerate drafting and synthesis, but the user remains responsible for accuracy, judgment, and appropriate handling of customer information.

• No source grounding fails because a renewal recommendation based only on memory is less reliable for leadership decision-making.
• Over-sharing content fails because sharing full transcripts and email threads may expose unnecessary sensitive details.
• Unneeded complexity fails because building an agent is excessive for a one-time leadership brief from existing sources.

Question 7

Topic: Understand Generative AI Fundamentals

A product manager has a Word campaign brief stored in Microsoft 365 and needs a client-ready slide deck. They tried this in Microsoft 365 Copilot chat: Make slides for the launch. Which prompt improvement best uses app-specific Copilot capabilities instead of relying on general chat behavior?

Options:

• A. Use Copilot in PowerPoint to create a 6-slide deck from the Word brief, with speaker notes.

• B. Paste the full brief into chat and ask for a persuasive deck.

• C. Use Copilot in Outlook to draft an email asking someone to build slides.

• D. Ask general Copilot chat to write everything it knows about the launch.

Best answer: A

Explanation: Microsoft 365 Copilot capabilities vary by app. General Copilot chat is useful for brainstorming, summarizing, and answering questions, but app-specific Copilot experiences can work directly in the current app context. For a slide deck, PowerPoint Copilot is the better fit because it can help create and structure a presentation from a referenced Word document. A stronger prompt also states the source, desired output, length, and presentation details such as speaker notes. The key takeaway is to choose the Copilot experience that matches the work product, then provide clear context and constraints.

• General chat only is too broad because it does not use the PowerPoint presentation context to build the deck.
• Pasting the full brief may add unnecessary sensitive content and still misses the app-specific creation capability.
• Outlook delegation creates an email request, but it does not use Copilot to produce the PowerPoint deliverable.

Question 8

Topic: Understand Generative AI Fundamentals

A customer success manager needs a one-page summary of recurring themes from recent support-call notes stored in a Microsoft 365 file they are allowed to access. The summary will be shared with the leadership team, so it must avoid exposing unnecessary customer details and should be checked against the source notes for accuracy. What is the best business-user action?

Options:

• A. Paste the full notes into a public AI chat and request an executive summary

• B. Ask IT to train a custom AI model on all customer records for better summaries

• C. Use Microsoft 365 Copilot to draft a cited summary from the file, remove unnecessary details, and review it against the notes

• D. Request tenant-wide permission changes so Copilot can summarize every support file

Best answer: C

Explanation: The core boundary is choosing Microsoft 365 Copilot for no-code productivity work when the task is drafting, summarizing, or analyzing content the user is already permitted to access. In this scenario, the manager can use Copilot to create a leadership-ready summary from the support-call notes, then reduce unnecessary customer details and verify the output against the source. That fits normal business-user Copilot scope. Training a model, changing tenant-wide permissions, or using an external tool with sensitive content moves outside routine Microsoft 365 Copilot work or creates unnecessary risk. The key takeaway is to use Copilot for productivity tasks, not to turn a routine summary request into engineering or administration work.

• Custom model training is unnecessarily complex for a routine summary and belongs to AI engineering, not business-user productivity.
• Public AI chat risks exposing customer information outside approved Microsoft 365 protection boundaries.
• Tenant-wide permission changes are an administrative action and would overreach the user’s normal access needs.

Question 9

Topic: Understand Generative AI Fundamentals

A product manager wants to use Microsoft 365 Copilot to help decide whether to continue a pilot program. The accountable business owner must make the final decision.

Draft prompt: “Review the pilot notes and tell us whether to continue the pilot.”

Which prompt improvement best supports informed decision-making without replacing the business owner?

Options:

• A. Ask Copilot to ignore uncertain data and provide a confident recommendation.

• B. Ask Copilot to make the final go/no-go decision automatically.

• C. Ask Copilot for evidence, risks, options, citations, and a decision-ready summary.

• D. Ask Copilot to summarize only the most positive pilot feedback.

Best answer: C

Explanation: For decision-quality work, Microsoft 365 Copilot should help the business owner understand the available information, not replace the owner’s judgment. A stronger prompt asks Copilot to use relevant sources, summarize evidence, identify risks and assumptions, compare options, include citations or source references, and produce an output the owner can review. This supports informed decision-making while preserving human accountability. Prompts that ask Copilot to decide automatically, hide uncertainty, or focus only on favorable evidence increase the risk of over-reliance and poor business judgment.

• Automatic decision-making fails because Copilot should support, not own, accountable business decisions.
• Ignoring uncertainty is unsafe because uncertain or incomplete data should be highlighted for review.
• Positive-only feedback creates bias and omits risks needed for a balanced decision.

Question 10

Topic: Understand Generative AI Fundamentals

A sales operations manager must decide whether to renew a high-value customer discount. They want to use Microsoft 365 Copilot to prepare for the decision, but company policy requires a human manager to make and approve pricing decisions. Which action best satisfies the requirement?

Options:

• A. Ask Copilot to summarize evidence, risks, and options for manager review

• B. Ask Copilot to select the best discount and notify the customer

• C. Ask Copilot to ignore older account notes and use only recent chats

• D. Ask Copilot to approve the renewal if it finds enough supporting emails

Best answer: A

Explanation: Microsoft 365 Copilot is appropriate for preparing work content, such as summarizing emails, extracting relevant facts, comparing options, and drafting a recommendation. For a business decision with financial impact, especially one covered by company policy, Copilot should support the human decision-maker rather than replace them. The safest workflow is to ask for evidence, assumptions, risks, and possible next steps, then have the manager verify sources and approve the final decision. Copilot output can contain gaps or fabrications, so unreviewed approval or customer communication is not appropriate.

• Asking Copilot to approve based on emails turns decision support into an unreviewed business decision.
• Having Copilot select the discount and notify the customer skips human review and approval.
• Ignoring older notes may omit important account context instead of improving decision quality.

Continue in the web app

Use IT Mastery for interactive Microsoft AB-730 practice with mixed sets, timed mocks, topic drills, explanations, and progress tracking.

Try Microsoft AB-730 on Web

Related focused pages

• Free Microsoft AB-730 Full-Length Practice Exam
• Prompts and Conversations
• Business Content with AI