Teach on Udemy

Turn what you know into an opportunity and reach millions around the world.

Learn More

Your cart is empty.

Keep shopping

AB-730: Microsoft Certified AI Business Professional

Pass AB-730: Copilot grounding, permissions, prompt craft and agents, taught through one company's rollout

Created bySkillmint .

Last updated 9/2026

English

What you'll learn

Explain how Microsoft 365 Copilot actually works: the model, your Microsoft Graph and the app you are in, and how each one shapes the answer you get
Say exactly what leaves your tenant and what never does, and why your prompts are not used to train the model
Predict when Copilot will return nothing, using permissions, sensitivity labels, encryption and the EXTRACT right
Spot fabrications, prompt injection and over-reliance, and pick the verification step that fits the task
Write prompts that survive contact with real work: goal, context, source and format, then iterate instead of restarting
Save, schedule and share prompts, and choose correctly between a chat, a Notebook, a Copilot Page and Loop
Decide between the Agent Store, a template and a custom agent, and configure knowledge, instructions and sharing without leaking data
Walk into AB-730 knowing its question style, the four decisions it keeps testing, and the registration detail most candidates get wrong

Course content

12 sections • 55 lectures

Course introduction3:06
What the course is for, the company you spend it inside, and what the exam asks

Priya's first prompt, and the question it raises2:30
Understand how Copilot works to keep your organization's information private and secure
Three parts: the model, your Graph, your apps2:48
Understand how the context, like your work files, web data, or the app you're using, can affect Copilot responses
The boundary: what leaves your tenant, what never does2:44
Understand how Copilot works to keep your organization's information private and secure
Copilot only ever sees what you can already open2:35
Understand how Copilot works to keep your organization's information private and secure
Your prompts do not train the model2:48
Understand how Copilot works to keep your organization's information private and secure
Module summary2:19
Everything the module landed: one scene per lesson, then the definitions and boundaries collected
Module quiz 1 - How Copilot actually works

Same question, two answers: why the app matters2:27
Understand how the context, like your work files, web data, or the app you're using, can affect Copilot responses
Work, web, and the file you have open2:27
Understand how the context, like your work files, web data, or the app you're using, can affect Copilot responses
What Copilot can do in each app2:56
Understand the differences in features and capabilities of the Copilot experience in various Microsoft 365 apps
Chat or agent? The distinction that keeps appearing2:30
Understand the difference between a chat experience and an agent experience
When it's worth building your own2:28
Understand the use case for creating your own agent
Module summary2:19
Everything the module landed: one scene per lesson, then the definitions and boundaries collected
Module quiz 2 - Context, apps, and the chat-vs-agent decision

The number that wasn't there2:04
Identify common risks, including fabrications, prompt injection, and over-reliance
Why fluent and wrong travel together1:57
Identify common risks, including fabrications, prompt injection, and over-reliance
Prompt injection, in an email you didn't write2:12
Identify common risks, including fabrications, prompt injection, and over-reliance
Over-reliance: the risk that has no error message2:03
Identify common risks, including fabrications, prompt injection, and over-reliance
Checking citations, and choosing when a human must read it2:33
Select verification steps appropriate to the task, including citation checks and human review
Module summary2:26
Everything the module landed: one scene per lesson, then the definitions and boundaries collected
Module quiz 3 - Trust, fabrication, and the human in the loop

“Copilot won't show me my own file”1:52
Understand how data protection restricts prompt results
Sensitivity labels, encryption, and the EXTRACT right2:12
Understand how data protection restricts prompt results
Labels travel: what your Copilot output inherits2:02
Recognize and mitigate risks to sensitive data
Sensitive data you shouldn't paste in the first place2:11
Recognize and mitigate risks to sensitive data
Module summary2:26
Everything the module landed: one scene per lesson, then the definitions and boundaries collected
Module quiz 4 - Data protection you can actually see

Goal, context, source, format — the four things a prompt needs2:17
Understand how to create an effective prompt
Pointing at the right source1:53
Select appropriate resources to reference in a prompt
Referencing files, meetings, people, and threads1:50
Select appropriate resources to reference in a prompt
Iterating instead of restarting1:32
Understand how to create an effective prompt
When a prompt is the wrong tool1:39
Understand how to create an effective prompt
Module summary2:15
Everything the module landed: one scene per lesson, then the definitions and boundaries collected
Module quiz 5 - Prompt craft that survives contact with work

Saving a prompt you'll want again1:44
Save a prompt
Scheduling: the report that arrives before you do1:54
Schedule a prompt
Sharing a prompt with the people who need it1:54
Share a prompt
Save, schedule, share — choosing between them1:40
Save a prompt / Schedule a prompt / Share a prompt
Module summary1:55
Everything the module landed: one scene per lesson, then the definitions and boundaries collected
Module quiz 6 - Prompts as reusable assets

Finding the conversation you had three weeks ago1:34
Understand how to find previous conversations
Renaming, so future-you can find it1:27
Rename a chat
Deleting a chat, and what deletion does not do1:49
Delete a chat
Notebooks: a project that outlives a thread1:56
Add a conversation to a notebook
Notebook, Page, or Loop?1:46
Add a conversation to a notebook / Describe how to use Copilot Pages for collaboration
Module summary2:28
Everything the module landed: one scene per lesson, then the definitions and boundaries collected
Module quiz 7 - Conversations, and where work goes to live

Requirements

No prior AI experience, no coding and no Microsoft 365 administration knowledge needed. The course starts from zero.
Comfort with everyday office work such as email, documents and meetings is all the background required.
Access to Microsoft 365 Copilot is useful for practice but is not needed to follow the course or pass the exam.

Description

Disclaimer: This course contains the use of artificial intelligence.

Eleven months into a Copilot rollout, Harborline has 900 people and one question that keeps coming back: why did Copilot return nothing?

That question is the spine of this course, and it is also the highest-value question on exam AB-730. You spend the whole course inside Harborline, a fictional company with entirely real problems, meeting the colleagues whose mistakes teach the syllabus faster than any definition could.

Mechanism before technique. Most Copilot training opens with prompt tricks. This one opens with how the thing actually works: the model, your Microsoft Graph, the app you happen to be in, and the permissions and sensitivity labels that decide what Copilot is allowed to see. Once you understand why an answer came back empty, prompt craft stops being guesswork.

What is inside

65 videos across 12 sections, two hours and nineteen minutes in total, with nothing padded
Every module closes with a summary video that collects the definitions and the boundaries in one place
Situational quiz questions where every wrong option is a plausible sibling feature, refuted by name rather than waved away
A final assessment weighted to the published exam blueprint
A dedicated exam readiness module on how AB-730 builds its questions and what to expect on the day

The exam

AB-730 earns the Microsoft Certified: AI Business Professional credential. It runs 45 minutes, needs 700 of 1000 to pass, is delivered through Pearson VUE and may include interactive components. It has no prerequisites and requires no code, and it renews free every year on Microsoft Learn. One practical warning this course repeats: register with a personal Microsoft account, because a credential tied to a work account is lost if you leave that organisation.

The three domains, and where the course covers them

Generative AI fundamentals, 25 to 30 percent. Modules 1 to 4: how Copilot is grounded, how context and the app change the answer, fabrications and prompt injection and over-reliance, and the data protection you can actually see.
Prompts and conversations, 35 to 40 percent. Modules 5 to 8: prompt craft that survives contact with work, saving and scheduling and sharing prompts, chats and notebooks and Pages, and agents from the Agent Store through to building your own.
Draft and analyze, 25 to 30 percent. Modules 9 and 10: drafting and summarising, moving work between Microsoft 365 apps, meetings, Pages, memory and custom instructions.

No prior AI experience is needed, and there is no code anywhere in the course. If you use Copilot at work and want to stop guessing, or you need the credential on your profile, start with module 1.

Who this course is for:

Business professionals preparing for Microsoft exam AB-730, AI Business Professional
Anyone whose organisation is rolling out Microsoft 365 Copilot and who has to use it safely and well
Managers, analysts, marketers, HR and operations staff who want a Microsoft credential without a technical background
Teams who keep asking why Copilot returned nothing, and want the mechanism explained rather than a list of prompt tricks
IT and enablement leads who need a shared vocabulary for grounding, permissions and sensitivity labels

AB-730: Microsoft Certified AI Business Professional

What you'll learn

Explore related topics

Course content

Orientation1 lecture • 3min

How Copilot actually works6 lectures • 16min

Context, apps, and the chat-vs-agent decision6 lectures • 15min

Trust, fabrication, and the human in the loop6 lectures • 13min

Data protection you can actually see5 lectures • 11min

Prompt craft that survives contact with work6 lectures • 11min

Prompts as reusable assets5 lectures • 9min

Conversations, and where work goes to live6 lectures • 11min

Agents4 lectures • 8min

Drafting, summarising, moving work between apps0

Requirements

Description

Who this course is for: