Microsoft • AB-730
Validates the ability to use generative AI-powered productivity tools, including Microsoft 365 Copilot, Researcher, and Analyst, to improve daily work and drive business outcomes. Designed for business professionals who leverage AI tools without building apps or writing code.
Questions
699
Duration
45 minutes
Passing Score
700/1000
Difficulty
FoundationalLast Updated
Mar 2026
The Microsoft Certified: AI Business Professional certification validates a candidate's ability to leverage generative AI-powered productivity tools — specifically Microsoft 365 Copilot, Copilot Researcher, and Copilot Analyst — to enhance daily work and drive measurable business outcomes. The credential is classified at the Beginner level and is designed for professionals who apply AI within existing business workflows rather than building AI applications or writing code. It demonstrates competency across three core skill areas: understanding how generative AI works within the Microsoft 365 ecosystem, creating and managing effective prompts and AI-driven conversations, and using AI to draft and analyze business content.
Earned by passing a single proctored exam (AB-730), this certification confirms that holders can identify responsible AI practices, protect sensitive organizational data, configure and share Microsoft 365 Copilot agents, and move insights fluidly between Microsoft 365 apps such as Word, Outlook, Teams, PowerPoint, and Excel. The certification was made generally available in early 2026 and sits within Microsoft's broader AB-series credential track, which spans AI fundamentals through enterprise-level AI deployment skills.
This certification is intended for business professionals across all industries — including marketing, operations, project management, human resources, finance, and customer service — who use or plan to use Microsoft 365 Copilot as part of their daily responsibilities. The exam is explicitly designed for the 'Business User' and 'Administrator' roles as defined by Microsoft, and no programming background or technical AI expertise is required.
Ideal candidates are comfortable navigating core Microsoft 365 applications such as Outlook, Word, Teams, PowerPoint, and Excel, and are familiar with common business tasks like drafting emails, preparing presentations, and managing documents. Business analysts, operations managers, project leads, consultants, and executive assistants looking to demonstrate measurable AI fluency and increase productivity through tools like Copilot Researcher and Copilot Analyst are the primary audience.
Microsoft does not list any mandatory prerequisites for the AB-730 exam. There are no required prior certifications, no minimum years of experience, and no technical or coding background needed. The credential is explicitly positioned as a no-code, no-development certification.
That said, Microsoft recommends that candidates have a working familiarity with core Microsoft 365 applications — Outlook, Word, Excel, PowerPoint, and Teams — and with common business processes such as drafting communications, creating presentations, generating images, and managing documents. Practical, hands-on experience using Microsoft 365 Copilot in a real or trial environment prior to sitting the exam is strongly recommended, as the exam includes scenario-based questions and may include interactive components.
Exam AB-730 is delivered as a proctored assessment through Pearson VUE and is available in English. Candidates have 45 minutes to complete the exam, which covers approximately 40–60 questions spanning multiple-choice, multi-select, and scenario-based formats; the exam may also include interactive components. A scaled score of 700 out of 1000 is required to pass.
The exam is priced at $99 USD (prices vary by country or region). If a candidate does not pass, a retake is permitted 24 hours after the first attempt, with subsequent retakes subject to Microsoft's standard retake policy. The exam environment can be previewed in advance using Microsoft's official exam sandbox at aka.ms/examdemo. Accommodations such as extended time or assistive devices are available upon request.
The AB-730 certification positions holders as AI-ready professionals capable of translating business goals into practical Copilot-driven outcomes — a skillset in high demand as enterprise adoption of Microsoft 365 Copilot accelerates. The credential is relevant across virtually every business function, making it broadly applicable regardless of industry vertical. It complements rather than competes with technical AI certifications, and pairs naturally with role-specific Microsoft certifications in project management, business analysis, or administration.
5 sample questions with correct answers and explanations. Start a practice session to test yourself across all 699 questions.
1. Contoso's IT department wants to understand the Copilot processing pipeline. After the user submits a prompt and the orchestrator performs grounding via Microsoft Graph and the Semantic Index, the grounded prompt is sent to the LLM. What happens immediately after the LLM generates its response? (Select one!)
Explanation
After the LLM generates its response, the next step in the Copilot processing pipeline is post-processing. This includes additional grounding calls, Responsible AI safety checks, content filtering, app-specific logic, and compliance verification. For app-specific actions like creating PowerPoint slides or Excel formulas, an Office Domain Specific Language (ODSL) code generator converts LLM output into executable Office calls. Delivering the response directly to the user skips critical safety and validation steps. Logging occurs after the response is delivered, not immediately after LLM processing. The Semantic Index is not updated based on individual LLM responses.
2. Tailspin Toys needs to understand the difference between Copilot Memory and Custom Instructions to train its employees. Which two statements accurately distinguish these features? (Select two!)
Multiple correct answersExplanation
Copilot Memory and Custom Instructions serve different purposes. Memory is dynamic and evolves over time — it includes both explicitly saved memories (when users tell Copilot to remember something) and inferred memories picked up from chat history. Memory stores facts and preferences like "I prefer Python for data science" or "I'm working on Project Alpha." Custom Instructions are static behavioral directives that users explicitly define to control how Copilot responds, such as tone, formatting, detail level, and style preferences. Custom Instructions persist until the user manually changes them. Importantly, deleting a chat does NOT delete saved memories — users must separately manage memories through Settings > Personalization > Saved memories.
3. Litware's project coordinator manages multiple ongoing projects and uses Copilot Chat extensively. She wants to organize her conversations more effectively. Which three actions can she perform to manage her Copilot Chat conversations? (Select three!)
Multiple correct answersExplanation
Microsoft 365 Copilot Chat supports three key conversation management actions: renaming a chat to change its title for better organization, deleting a chat to remove it from the user's view, and adding a conversation to a Copilot Notebook for focused, curated work with scoped context. Pinning conversations, exporting as PDF, and merging conversations are not supported conversation management features in Copilot Chat.
4. Northwind Traders' IT administrator is explaining the Copilot processing pipeline to business stakeholders. She needs to describe how the Copilot Orchestrator functions within the architecture. Which statement best describes the role of the Copilot Orchestrator? (Select one!)
Explanation
The Copilot Orchestrator is not an AI itself but a coordination layer that manages multiple components including AI models, data sources, plugins and extensions, and Responsible AI safety checks. It implements Retrieval-Augmented Generation by retrieving enterprise content from Microsoft Graph and the Semantic Index, coordinates content filtering and rate limiting, and logs interactions for auditing. It is not a specialized AI model that generates responses — that role belongs to the LLM via Azure OpenAI Service. It is not a database engine; Microsoft Graph and the Semantic Index handle data retrieval. Encryption is handled by Azure services using TLS and AES-256, not the orchestrator.
5. Adatum's project manager wants to share a Copilot agent she built with her team of 50 people, plus a partner team of 30 people at a different company that collaborates on their joint venture. What sharing limitation will she encounter? (Select one!)
Explanation
Cross-tenant sharing of Copilot agents is not supported. Agents can only be shared with people within the same Microsoft 365 tenant. While the agent supports sharing with up to 98 specific users, security groups, or Microsoft 365 groups, all recipients must be within the same organization. The project manager can share with her 50 internal team members but the 30 partner company employees in a different tenant cannot access the agent. To extend agent capabilities to external users, the agent would need to be copied to Copilot Studio for full publishing capabilities. The Agent Store is also limited to the organization's tenant.
One-time access to this exam