CertSafari

    Free Microsoft Intelligent Applications Builder (AB-410) Sample Questions

    35 free sample questions from our bank of 357+, covering every exam domain, with answers and detailed explanations. Updated August 2026.

    Practice all 357 questions in quiz mode

    Domain 1: Create a foundation for intelligent applications

    Subdomain 1.1: Design Microsoft Power Platform solutions by using AI-enabled tools

    1.A solution architect wants an agent to be able to interact with legacy desktop applications that expose no API, by clicking buttons and reading screen text the way a human would. Which Copilot Studio capability should be evaluated for this requirement?

    1. A.Computer-using agent capabilities that operate a UI directly
    2. B.A standard connector-based agent action calling a REST endpoint
    3. C.A Dataverse formula column referencing the legacy system's fields
    4. D.A business process flow stage that pauses for manual entry
    Show answer & explanation

    Correct answer: A — Computer-using agent capabilities that operate a UI directly

    • A. Computer-using agent capability lets an agent operate an application's user interface directly, such as clicking and reading screen elements, which is exactly what a no-API legacy desktop app requires.
    • B. A connector-based action calls an API endpoint; it cannot interact with an application that exposes no API at all.
    • C. A formula column computes a value within Dataverse and has no ability to operate an external desktop application's interface.
    • D. A business process flow stage pause still requires a human to perform the legacy-system steps manually rather than having the agent perform them.

    Subdomain 1.1: Design Microsoft Power Platform solutions by using AI-enabled tools

    2.Which statement about multi-agent orchestration in Copilot Studio is correct?

    1. A.It allows one agent to delegate specific subtasks to other specialized agents within a solution
    2. B.It requires every agent in the tenant to share a single conversation transcript with no isolation
    3. C.It replaces the need for connectors when an agent must call an external system
    4. D.It is only available for agents built entirely from a blank canvas with no templates
    Show answer & explanation

    Correct answer: A — It allows one agent to delegate specific subtasks to other specialized agents within a solution

    • A. Multi-agent orchestration lets a primary agent hand off specialized subtasks to other purpose-built agents, composing capability rather than building one monolithic agent.
    • B. Orchestration coordinates task delegation between agents; it does not force every agent in the tenant to merge into one shared, unisolated transcript.
    • C. Agents still rely on connectors to call external systems and take action; orchestration coordinates between agents rather than substituting for connector-based integration.
    • D. Orchestration is not restricted to agents built from blank; agents created from prebuilt templates can also participate in orchestrated multi-agent solutions.

    Subdomain 1.1: Design Microsoft Power Platform solutions by using AI-enabled tools

    3.A pharmaceutical company wants a Power Platform environment where developers can safely reset, delete, and copy data while testing new solution changes without risking production data. Which environment type should you recommend?

    1. A.A sandbox environment
    2. B.The tenant's default environment
    3. C.A production environment with restricted security roles
    4. D.A developer environment shared by the whole team
    Show answer & explanation

    Correct answer: A — A sandbox environment

    • A. Sandbox environments are nonproduction environments isolated from production and specifically include capabilities such as reset, delete, and copy that would be harmful in production.
    • B. The default environment is shared by every user in the tenant and is not intended for isolated testing that includes destructive operations like reset or delete.
    • C. Even with restricted roles, a production environment is where live apps and data operate, so it is not the place to safely perform reset, delete, or copy operations.
    • D. A developer environment is a single-user environment meant for individual build-and-test work, not a shared space for a team to test solution changes together.

    Subdomain 1.1: Design Microsoft Power Platform solutions by using AI-enabled tools

    4.A team needs to apply a small, potentially reversible hotfix on top of an already-deployed managed solution in production, without altering the base layer. Which approach should they use?

    1. A.Import a solution patch layered on top of the parent managed solution
    2. B.Perform a full solution upgrade that replaces the base layer
    3. C.Export the solution as unmanaged and re-import it into production
    4. D.Delete the managed solution and reinstall a new version from scratch
    Show answer & explanation

    Correct answer: A — Import a solution patch layered on top of the parent managed solution

    • A. A patch contains only the changes for a parent managed solution and layers on top of it without altering the base layer, and because it can potentially be uninstalled later, it fits a small, reversible hotfix.
    • B. An upgrade installs a new solution layer immediately above the base layer, deletes existing patches, and can remove components no longer included, which is more disruptive than a small reversible hotfix requires.
    • C. Re-importing a solution as unmanaged into production breaks the managed-solution servicing model and exposes components to being individually edited or deleted, which is not appropriate for production.
    • D. Deleting and reinstalling the entire managed solution is far more disruptive than necessary for a small hotfix and risks losing configuration made on top of the solution.

    Subdomain 1.1: Design Microsoft Power Platform solutions by using AI-enabled tools

    5.True or False: Solutions can contain business data such as customer records in addition to app and configuration metadata.

    1. A.True
    2. B.False
    Show answer & explanation

    Correct answer: B — False

    • A. This is false; solutions carry metadata and certain configuration tables, but they never carry business data such as customer records.
    • B. Solutions transport components like apps, flows, tables, and configuration data between environments, but business data itself is excluded and must be migrated separately if needed.

    Subdomain 1.2: Build data models

    6.A finance team needs a number column for calculating precise interest amounts where exact equality comparisons matter, while a manufacturing team needs a column for sensor tolerance values queried mainly with greater-than/less-than comparisons. Which two column type choices correctly match each need?(Select 2)

    1. A.Decimal Number for the finance team, because decimal values are stored exactly as specified for accurate equality comparisons
    2. B.Floating Point Number for the manufacturing team, because it is well suited to values typically queried with comparison operators rather than exact equality
    3. C.Whole Number for the finance team, because whole numbers always provide the most accurate financial calculations
    4. D.Floating Point Number for the finance team, because floating point storage guarantees exact decimal precision
    5. E.Big (BigInt) for the manufacturing team, because BigInt is optimized for fractional tolerance values
    Show answer & explanation

    Correct answers: A, B — Decimal Number for the finance team, because decimal values are stored exactly as specified for accurate equality comparisons; Floating Point Number for the manufacturing team, because it is well suited to values typically queried with comparison operators rather than exact equality

    • A. Decimal Number values are stored in the database exactly as specified, which is why they suit financial interest calculations that depend on exact equality.
    • B. Floating Point Number stores a close approximation and is well suited to values typically compared using greater-than or less-than operators, matching the tolerance-checking use case.
    • C. Whole Number cannot represent fractional interest amounts at all, so it does not provide better financial accuracy than decimal storage.
    • D. Floating point storage is an approximation rather than an exact value, which is the opposite of what precise financial equality comparisons need.
    • E. Big/BigInt is meant for very large integers such as timestamps or IDs, not fractional tolerance measurements.

    Subdomain 1.2: Build data models

    7.An "Account" table has a 1:N parental relationship to a custom "Site Visit" table where each site visit belongs to exactly one account and doesn't make sense without it. The business wants every site visit deleted automatically whenever its parent account is deleted. Which Delete behavior should be configured on the relationship?

    1. A.Cascade All, because it performs the delete action on all related Site Visit rows
    2. B.Remove Link, because it only clears the lookup value on related rows without deleting them
    3. C.Restrict, because it prevents the account from being deleted while site visits exist
    4. D.Cascade User Owned, because it deletes only the site visits owned by the same user as the account
    Show answer & explanation

    Correct answer: A — Cascade All, because it performs the delete action on all related Site Visit rows

    • A. Cascade All on Delete removes every related Site Visit row whenever its parent Account row is deleted, matching the requirement that visits never outlive the account.
    • B. Remove Link would leave the Site Visit rows in place but clear their reference to the deleted account, not delete them as required.
    • C. Restrict would block deleting the account entirely while site visits exist, which is the opposite of the desired automatic cleanup.
    • D. Cascade User Owned would only delete site visits owned by the same user as the account, leaving others behind, which doesn't guarantee all visits are removed.

    Subdomain 1.2: Build data models

    8.While configuring a prompt column that references a "Customer Feedback" text column as input, a maker also tries to add a "Priority Score" formula column and an existing prompt column as additional inputs. What happens?

    1. A.The formula column and the other prompt column are ignored as input values, because prompt columns can't use formula, file, image, or other prompt columns as inputs
    2. B.Both are accepted as valid inputs, since any column type can feed a prompt column
    3. C.Only the formula column is rejected, since prompt columns are freely allowed to reference each other
    4. D.The prompt column definition fails to save entirely whenever an unsupported column type is referenced
    Show answer & explanation

    Correct answer: A — The formula column and the other prompt column are ignored as input values, because prompt columns can't use formula, file, image, or other prompt columns as inputs

    • A. Prompt columns explicitly ignore formula columns, file columns, image columns, and other prompt columns when selected as inputs, so their values won't feed the prompt.
    • B. Not every column type is a valid prompt input; formula, file, image, and other prompt columns are specifically excluded.
    • C. Both the formula column and the other prompt column are excluded as inputs, not just one of them.
    • D. Selecting an unsupported input type doesn't block saving the prompt column definition; the platform simply ignores that input's value rather than failing the save.

    Subdomain 1.2: Build data models

    9.A model-driven app team wants users to see an at-a-glance Copilot-generated overview of a "Case" record's key details in a collapsible bar at the top of the main form, without adding a new column to the table. Which capability should they configure?

    1. A.A row summary on the Case table's main form, using a custom prompt to specify which columns Copilot should summarize
    2. B.A prompt column on the Case table, since prompt columns render as a collapsible bar automatically
    3. C.A public view filtered to show only key columns, since views can appear as a form-level summary bar
    4. D.A business rule that concatenates key columns into a single read-only field on the form
    Show answer & explanation

    Correct answer: A — A row summary on the Case table's main form, using a custom prompt to specify which columns Copilot should summarize

    • A. A row summary uses a custom prompt to tell Copilot which columns to summarize, and it renders as a collapsible bar at the top of the main form without requiring a new stored column.
    • B. Prompt columns store their AI-generated result as a regular column value; they don't render as a collapsible form-header summary bar.
    • C. Views control how lists of records are displayed elsewhere in the app; they don't produce a summary bar on an individual record's form.
    • D. A business rule concatenating fields creates a static, non-AI-generated read-only field, not a Copilot-generated collapsible summary.

    Subdomain 1.2: Build data models

    10.A team is deciding how to expose a filtered list of "Overdue Invoice" records consistently to both canvas apps and model-driven apps. Which two statements about Dataverse views are accurate for this scenario?(Select 2)

    1. A.Views are named queries executed on the server, so the same view logic can be reused across canvas and model-driven apps
    2. B.Views must be recreated separately inside every canvas app and model-driven app that needs the same filtered list
    3. C.A view can define which columns display, their column widths, and the applied filters
    4. D.Views can only be created from the classic solution explorer, never from the maker portal's data workspace
    5. E.Public views are limited to a single filter condition and cannot combine multiple criteria
    Show answer & explanation

    Correct answers: A, C — Views are named queries executed on the server, so the same view logic can be reused across canvas and model-driven apps; A view can define which columns display, their column widths, and the applied filters

    • A. Views are named queries that Dataverse executes on the server, which is exactly why the same view definition can be reused across canvas apps and model-driven apps.
    • B. Because views are server-side and reusable, they don't need to be rebuilt separately inside every app that wants the same filtered list.
    • C. A view can indeed define the displayed columns, their widths, and the filters applied, which is the core of what a view configures.
    • D. Views can be created from the maker portal's data workspace as well as from solution explorer; the data workspace is not excluded.
    • E. Public views can combine multiple filter conditions; they aren't limited to a single criterion.

    Domain 2: Create intelligent applications

    Subdomain 2.1: Create model-driven apps

    11.Which privilege is typically required for a user to share a model-driven app with other users?

    1. A.System Administrator or System Customizer security role
    2. B.Basic User security role
    3. C.Environment Maker security role only
    4. D.Any predefined security role with read access to one table
    Show answer & explanation

    Correct answer: A — System Administrator or System Customizer security role

    • A. The app sharer must hold a security role with privileges equal to or greater than what they are assigning, which usually means the System Administrator or System Customizer role.
    • B. The Basic User role grants limited privileges to work with owned records and does not carry the elevated privileges needed to share an app with others.
    • C. Environment Maker allows building apps without a license but does not by itself grant the elevated data privileges needed to share an app with other users.
    • D. A role limited to read access on one table lacks the broad privileges required to assign that same access to other users through sharing.

    Subdomain 2.1: Create model-driven apps

    12.When editing a view directly through a table's Views area, which of the following actions are available in the view designer? (Select 3)(Select 3)

    1. A.Change column filtering by selecting the column header and choosing Filter by
    2. B.Change the sort order only through a separate View Properties dialog, not from the column header
    3. C.Reorder columns by dragging a column to a new position
    4. D.Configure column width by dragging the column to the desired size
    5. E.Convert the view directly into a system chart
    6. F.Publish the view as a new business process flow stage
    Show answer & explanation

    Correct answers: A, C, D — Change column filtering by selecting the column header and choosing Filter by; Reorder columns by dragging a column to a new position; Configure column width by dragging the column to the desired size

    • A. Selecting a column header and choosing Filter by is a documented way to change which records are included based on that column's values.
    • B. Sorting is actually done from the column header itself, selecting options like Sort A-Z or Sort descending, not through a separate dialog, so this statement describes the wrong mechanism.
    • C. Reordering columns by dragging them to a new position is a supported action in the view designer.
    • D. Dragging a column to resize it is a supported way to configure column width in the view designer.
    • E. There is no direct action that converts a view into a system chart; charts are created separately from the view designer.
    • F. Business process flow stages are a distinct feature configured on the process flow itself, not something a view is published as.

    Subdomain 2.1: Create model-driven apps

    13.Which of the following are valid components that a maker can add to a model-driven app using the app designer? (Select 3)(Select 3)

    1. A.Tables
    2. B.Dashboards
    3. C.Site map
    4. D.A canvas app screen embedded directly as a table row form
    5. E.A Power BI premium capacity assignment
    6. F.A SQL Server linked server connection
    Show answer & explanation

    Correct answers: A, B, C — Tables; Dashboards; Site map

    • A. Tables are added to a model-driven app through the app designer as the core data components the app exposes.
    • B. Dashboards can be added as components in the app designer to give app users visual summaries of data.
    • C. The site map is a component of the app designer used to configure the app's navigation structure.
    • D. There isn't an app designer capability that embeds a canvas app screen directly in place of a table's row form, so this describes something the tool doesn't do.
    • E. Power BI premium capacity is an administrative licensing assignment, not a component added within the app designer.
    • F. A SQL Server linked server connection is a database-level configuration, not something configured through the model-driven app designer.

    Subdomain 2.1: Create model-driven apps

    14.Which of the following can be used to control who can access a shared model-driven app? (Select 3)(Select 3)

    1. A.Assigning individual users
    2. B.Assigning teams
    3. C.Assigning security roles so all members of that role gain access
    4. D.Assigning a canvas app screen as a gatekeeper
    5. E.Assigning a Power Automate approval flow as the only access gate
    6. F.Assigning a Dataverse table relationship
    Show answer & explanation

    Correct answers: A, B, C — Assigning individual users; Assigning teams; Assigning security roles so all members of that role gain access

    • A. The Share pane allows selecting individual users from the People list to grant them access to the app.
    • B. Teams can be assigned in the Share pane, granting access to every member of the team.
    • C. Selecting a security role in the Share pane grants access to all users who hold that role, which is the primary way model-driven apps scale sharing.
    • D. There is no mechanism to gate model-driven app access behind a canvas app screen; the two app types use separate access models.
    • E. An approval flow is a business process tool and is not a supported access-control mechanism for opening a model-driven app.
    • F. A table relationship defines how records relate to each other and has no bearing on who can access the app itself.

    Subdomain 2.1: Create model-driven apps

    15.Which statements correctly describe interactive experience dashboards in model-driven apps? (Select 2)(Select 2)

    1. A.Multi-stream interactive dashboards can be configured as home page dashboards or as table-specific dashboards
    2. B.Single-stream interactive dashboards display real-time data driven by a single table view
    3. C.Interactive dashboards prevent users from taking any action directly on a row
    4. D.Standard dashboards and interactive dashboards are the same feature under different names
    5. E.Multi-stream dashboards are limited to two data streams maximum
    Show answer & explanation

    Correct answers: A, B — Multi-stream interactive dashboards can be configured as home page dashboards or as table-specific dashboards; Single-stream interactive dashboards display real-time data driven by a single table view

    • A. Multi-stream interactive dashboards can be set up either as home page dashboards or as dashboards scoped to a specific table.
    • B. Single-stream interactive dashboards display real-time data based on one table view, which is the defining characteristic that separates them from multi-stream dashboards.
    • C. Interactive dashboards specifically enable users to act on a particular row directly from the dashboard, which is the opposite of preventing row-level action.
    • D. Standard dashboards only support adding unrelated components, while interactive dashboards add row-level interactivity, so the two are distinct dashboard types, not the same feature.
    • E. There is no documented two-stream cap on multi-stream dashboards; they support multiple real-time data streams without that specific limit.

    Subdomain 2.2: Create canvas apps

    16.A canvas app must render correctly on phones, tablets, and desktop browsers without the maker building separate layouts for each screen size. Which design approach best satisfies this requirement?

    1. A.Use responsive layout containers with relative sizing so controls reflow and resize based on available screen space
    2. B.Set every control's Width and Height to fixed pixel values that match the most common desktop resolution
    3. C.Create a duplicate copy of every screen for phone, tablet, and desktop and use App.OnStart to detect device type
    4. D.Lock the app's orientation to landscape so all devices display the same fixed layout dimensions
    Show answer & explanation

    Correct answer: A — Use responsive layout containers with relative sizing so controls reflow and resize based on available screen space

    • A. Layout containers combined with relative (percentage or fill-based) sizing let controls automatically reflow to fit the available screen space, which is the recommended way to support many screen sizes with one layout.
    • B. Fixed pixel sizing tuned to one desktop resolution will overflow or leave large gaps on phones and tablets, defeating responsiveness.
    • C. Duplicating every screen per device class is possible but creates a large maintenance burden and duplicated logic, which is not the recommended responsive design approach.
    • D. Locking orientation does not address differing screen dimensions across phones, tablets, and desktops, and removes a usability option for portrait-preferring users.

    Subdomain 2.2: Create canvas apps

    17.A maker builds a canvas component that is intended for use only within a single specific app and needs the component to directly reference that app's global variables and Dataverse-connected controls. What must the maker enable on the component?

    1. A.The Access app scope switch on the component's property pane
    2. B.The Export and import components upcoming feature toggle
    3. C.The Formula-level error management upcoming feature toggle
    4. D.The Git version control experimental feature for the app
    Show answer & explanation

    Correct answer: A — The Access app scope switch on the component's property pane

    • A. Turning on Access app scope allows a component to directly read global variables, collections, and controls in its host app, which is otherwise blocked to keep components encapsulated and reusable.
    • B. Export and import components is a retired mechanism for moving component definitions between apps and has no effect on whether a component can read the host app's variables.
    • C. Formula-level error management governs how IfError, IsError, and related error-handling functions behave in the app and is unrelated to a component's ability to access app-level state.
    • D. Git version control manages source control integration for the app and does not affect whether a component can directly access the host app's variables and controls.

    Subdomain 2.2: Create canvas apps

    18.A maker tries to record a Test Studio test case for a screen that heavily uses a custom canvas component with nested galleries inside it, and the recorder fails to capture interactions correctly. What is the most likely explanation?

    1. A.Components and nested galleries are known limitations that Test Studio does not currently support
    2. B.Test Studio only supports apps published to production and cannot record against apps in Studio
    3. C.The maker lacks co-owner or creator permission on the app, which blocks the recorder entirely
    4. D.Test Studio requires the app to first be exported to a component library before recording
    Show answer & explanation

    Correct answer: A — Components and nested galleries are known limitations that Test Studio does not currently support

    • A. Components, code components, and nested galleries are documented known limitations of Test Studio, so interactions inside them may not record or play back correctly.
    • B. Test Studio can be used while building and testing an app and is not restricted to apps already published to production.
    • C. Lacking creator or co-owner permission would prevent using Test Studio at all rather than causing a partial recording failure specific to components and nested galleries.
    • D. Test Studio does not require an app to be exported into a component library before it can be tested; testing works directly against the app.

    Subdomain 2.2: Create canvas apps

    19.Which of the following design choices directly support accessibility for canvas app users relying on assistive technology? (Select all that apply.)(Select 3)

    1. A.Setting AccessibleLabel on controls that lack visible descriptive text, such as icon-only buttons
    2. B.Configuring a logical TabIndex order so keyboard navigation follows the visual flow of the screen
    3. C.Ensuring sufficient color contrast between text and background so low-vision users can read content
    4. D.Using ZIndex to make sure decorative background shapes visually overlap interactive controls
    5. E.Relying only on color, without text or icons, to indicate validation errors on form fields
    Show answer & explanation

    Correct answers: A, B, C — Setting AccessibleLabel on controls that lack visible descriptive text, such as icon-only buttons; Configuring a logical TabIndex order so keyboard navigation follows the visual flow of the screen; Ensuring sufficient color contrast between text and background so low-vision users can read content

    • A. AccessibleLabel gives assistive technology like screen readers text to announce for controls that would otherwise have no readable label, directly supporting accessibility.
    • B. A logical TabIndex order lets keyboard-only users move through the screen in a predictable sequence that matches what sighted users see visually.
    • C. Sufficient contrast between text and its background is a core accessibility requirement so users with low vision can read content without additional aids.
    • D. Using ZIndex to overlap decorative shapes on top of interactive controls can obscure or block those controls, harming usability rather than supporting accessibility.
    • E. Relying only on color to convey validation state excludes users with color vision deficiencies, so accessible design pairs color with text or icons as well.

    Subdomain 2.2: Create canvas apps

    20.A maker is deciding whether to build a piece of reusable logic as a Power Fx user-defined function (UDF). Which of the following statements about UDFs in canvas apps are correct? (Select all that apply.)(Select 3)

    1. A.Parameter types and the function's return type must be explicitly declared rather than inferred
    2. B.A UDF is defined alongside named formulas in the app's formulas definitions rather than inside a single screen
    3. C.A UDF can be marked as a behavior function to perform side-effecting actions like Patch or Notify
    4. D.A UDF automatically inherits the TabIndex and AccessibleLabel properties of the control that calls it
    5. E.A UDF can only be called from App.OnStart and cannot be referenced from any control property formula
    Show answer & explanation

    Correct answers: A, B, C — Parameter types and the function's return type must be explicitly declared rather than inferred; A UDF is defined alongside named formulas in the app's formulas definitions rather than inside a single screen; A UDF can be marked as a behavior function to perform side-effecting actions like Patch or Notify

    • A. Unlike regular formulas where types are inferred from context, UDFs require explicit type annotations for their parameters and return value.
    • B. UDFs live alongside named formulas in the app's formulas definitions, making them available app-wide rather than scoped to one screen's properties.
    • C. A UDF can be declared as a behavior function so it is allowed to call side-effecting operations such as Patch or Notify, similar to how OnSelect formulas work.
    • D. TabIndex and AccessibleLabel are control properties related to UI accessibility and have no relationship to a Power Fx function's definition or behavior.
    • E. A UDF can be referenced from many places once defined, including control properties, not only from App.OnStart.

    Domain 3: Build business application logic and automation

    Subdomain 3.1: Create cloud flows

    21.The approvals connector in Power Automate is a premium connector that requires a Power Apps per-user license for every approver.

    1. A.True
    2. B.False
    Show answer & explanation

    Correct answer: B — False

    • A. This is incorrect because the approvals capability is delivered through a standard connector.
    • B. This is correct because approvals actions are standard connector actions, so any license that covers standard connectors, such as a Power Automate or Microsoft 365 license, is enough to build and run them.

    Subdomain 3.1: Create cloud flows

    22.A team adds a single Dataverse action into an otherwise all-standard-connector flow that previously ran under a Microsoft 365 license with no extra cost. What is the licensing consequence?

    1. A.Only the Dataverse step requires a premium license while the rest of the flow stays standard
    2. B.The entire flow becomes a premium flow because Dataverse is a premium connector
    3. C.Nothing changes, because Dataverse ships with every Microsoft 365 license at no extra cost
    4. D.The flow becomes premium only if the Dataverse step writes data rather than only reading it
    Show answer & explanation

    Correct answer: B — The entire flow becomes a premium flow because Dataverse is a premium connector

    • A. Licensing applies at the whole-flow level, not per action, so this partial-premium outcome does not reflect how flow licensing actually works.
    • B. Because Dataverse is a premium connector, adding even one Dataverse step elevates the entire flow to premium, requiring every runner to hold a license that covers premium connectors.
    • C. Dataverse is a premium connector and is not included in a base Microsoft 365 license at no extra cost.
    • D. The premium classification applies to the connector itself, not to whether a given step reads or writes data, so read-only usage still elevates the flow.

    Subdomain 3.1: Create cloud flows

    23.A support desk wants any one of three on-call managers to be able to authorize an emergency access request, with the flow continuing as soon as the first manager responds. Which approval type fits?

    1. A.Approve/Reject - Everyone must approve
    2. B.Approve/Reject - First to respond
    3. C.Sequential approval
    4. D.Custom Responses - Wait for all responses
    Show answer & explanation

    Correct answer: B — Approve/Reject - First to respond

    • A. This type requires a response from every assigned approver, which would delay the flow instead of continuing after the first response.
    • B. This type lets the actions after the approval run as soon as any one of the assigned approvers responds, exactly matching the requirement for the fastest available manager to authorize.
    • C. This type asks approvers one at a time in order rather than letting whichever manager is available respond first.
    • D. This type still requires every approver to respond before continuing, just with custom labels, which does not fit a first-response requirement.

    Subdomain 3.1: Create cloud flows

    24.Which Run after statuses can be used to control when a downstream action executes relative to the action before it?

    1. A.Is successful, has timed out, is skipped, has failed
    2. B.Is queued, is running, is paused, is cancelled
    3. C.Is premium, is standard, is custom, is deprecated
    4. D.Is triggered, is scheduled, is instant, is automated
    Show answer & explanation

    Correct answer: A — Is successful, has timed out, is skipped, has failed

    • A. These are the four outcome states that Run after can be configured against to control exactly when a downstream action is allowed to execute.
    • B. These describe general execution states of a run rather than the specific set of predecessor outcomes that Run after can be configured against.
    • C. These describe connector licensing tiers, not action outcome states, and have no relationship to the Run after setting.
    • D. These describe trigger categories, not the outcome states of a preceding action, so they are not options available in Run after.

    Subdomain 3.1: Create cloud flows

    25.A maker testing a new flow wants failures to be visible and easy to diagnose during development, rather than silently retried and hidden. Which testing practices support this? (Select all that apply — choose 2)(Select 2)

    1. A.Temporarily set the retry policy on key actions to None while testing
    2. B.Manually trigger a test run and immediately inspect the run history for each action's status
    3. C.Disable run history entirely so old runs do not clutter the view
    4. D.Set every action's Run after to include all four statuses so nothing is ever skipped
    5. E.Rename the flow to include the word test without changing any action settings
    Show answer & explanation

    Correct answers: A, B — Temporarily set the retry policy on key actions to None while testing; Manually trigger a test run and immediately inspect the run history for each action's status

    • A. Turning off retries during testing means a failure surfaces on the very first attempt instead of being masked by automatic retries, making problems easier to spot.
    • B. Manually running the flow and immediately reviewing each action's status in run history is the direct way to see exactly what succeeded or failed during that test run.
    • C. Disabling run history removes the exact diagnostic record a maker needs to see what happened during a test run, which works against the goal rather than supporting it.
    • D. Allowing every action to run regardless of the predecessor's outcome would make actions execute even after a failure, which obscures rather than clarifies where the real failure occurred.
    • E. Changing only the flow's display name has no effect on how failures behave or how visible they are during testing.

    Subdomain 3.2: Create prompts and models in AI Hub

    26.A prompt originally built for a canvas app is later added as a tool inside a Copilot Studio agent, and both the app and the agent use it regularly. What changes about how usage of that prompt is billed?

    1. A.The canvas app usage draws on prompt builder credits, while the Copilot Studio agent usage draws on Copilot Credits
    2. B.Both the canvas app and the Copilot Studio agent always draw from the same shared prompt builder credit pool
    3. C.Only the canvas app usage is billed at all, since prompts running inside Copilot Studio agents are always free
    4. D.Both surfaces convert automatically to AI Builder credits regardless of whether Power Apps or Copilot Studio is used
    Show answer & explanation

    Correct answer: A — The canvas app usage draws on prompt builder credits, while the Copilot Studio agent usage draws on Copilot Credits

    • A. Using prompts in Power Apps or Power Automate consumes prompt builder credits, while using the same prompt inside Copilot Studio consumes Copilot Credits, so the billing source depends on which surface runs it.
    • B. The two surfaces draw from different credit pools rather than one shared prompt builder pool, since Copilot Studio usage is metered through Copilot Credits instead.
    • C. Copilot Studio usage is not free; it is metered through Copilot Credits rather than being exempt from billing.
    • D. AI Builder credits are consumed first when prompts are used in Power Apps or Power Automate if available, but Copilot Studio usage is billed through Copilot Credits, not a universal AI Builder credit conversion.

    Subdomain 3.2: Create prompts and models in AI Hub

    27.A maker adds an Account table as a knowledge source and wants the prompt to reference data three relationship levels deep, such as an account's contact's related opportunity's owner. What should the maker expect?

    1. A.Only up to two levels of one-to-many or many-to-one relationships are supported, so this three-level chain isn't possible
    2. B.Any number of relationship levels can be traversed as long as every table involved is a Dataverse table itself
    3. C.Three-level relationships work only when a many-to-many relationship replaces one of the standard lookups used
    4. D.Relationship depth has no limit for knowledge objects, only the total number of selected columns is restricted
    Show answer & explanation

    Correct answer: A — Only up to two levels of one-to-many or many-to-one relationships are supported, so this three-level chain isn't possible

    • A. Only one-to-many or many-to-one relationships are supported for Dataverse knowledge, and a table added as a data source can select up to two levels of relationships, so a three-level chain exceeds the supported depth.
    • B. Even when every table is a Dataverse table, the documented limit caps relationship traversal at two levels rather than allowing unlimited depth.
    • C. Many-to-many relationships are not the supported relationship type for Dataverse knowledge; only one-to-many or many-to-one relationships are supported, so substituting one doesn't remove the depth limit.
    • D. The two-level relationship restriction is a separate, documented limitation from any column-count consideration, so column count alone does not describe the relationship-depth limit.

    Subdomain 3.2: Create prompts and models in AI Hub

    28.A maker notices a knowledge-grounded prompt only seems to consider a subset of a large Dataverse table when generating answers, even though filtering isn't the cause. What is the most likely explanation?

    1. A.The knowledge source retrieves 30 records by default, and that limit needs raising in prompt settings for larger sets
    2. B.Dataverse tables larger than 1,000 rows cannot be used as a knowledge source under any configuration at all
    3. C.The prompt automatically excludes any table that has more than one related table configured as knowledge
    4. D.Knowledge objects only consider records created in the current calendar month unless a date filter is added
    Show answer & explanation

    Correct answer: A — The knowledge source retrieves 30 records by default, and that limit needs raising in prompt settings for larger sets

    • A. The number of records retrieved from a knowledge source is 30 by default, and this can be increased up to 1,000 in the settings, which explains why a large table appears to be only partially considered.
    • B. Tables larger than 1,000 rows can still be used as a knowledge source; the record retrieval setting simply caps how many records are pulled in per run, up to that 1,000 maximum.
    • C. There is no documented rule that excludes a table entirely just because more than one related table is added as knowledge; the two-level relationship limit governs depth, not exclusion of related tables.
    • D. Knowledge objects do not automatically restrict results to the current calendar month; any such date-based narrowing would need to be configured explicitly as a filter.

    Subdomain 3.2: Create prompts and models in AI Hub

    29.Which of the following are documented limitations of image or document input in a prompt? (Select all that apply.)(Select 3)

    1. A.Files passed to the prompt must total less than 25 MB combined across the run
    2. B.Uploaded documents must have fewer than 50 pages in total to be processed
    3. C.Processing an image or document is restricted to a maximum of 100 seconds
    4. D.Only one file can be uploaded per prompt run, no matter which file type is used
    5. E.Video files up to 25 MB in size are fully supported as a document input type
    6. F.Image or document input is fully supported for prompts added as Copilot Studio agent tools
    Show answer & explanation

    Correct answers: A, B, C — Files passed to the prompt must total less than 25 MB combined across the run; Uploaded documents must have fewer than 50 pages in total to be processed; Processing an image or document is restricted to a maximum of 100 seconds

    • A. Files passed to the prompt must total less than 25 MB for all files combined, which is a documented size limitation for image or document input.
    • B. Documents must have fewer than 50 pages, which is a documented limitation on image or document input.
    • C. Image or document processing is restricted to a maximum of 100 seconds, terminating with a timeout error if this is exceeded.
    • D. There is no documented one-file-per-run restriction; the limitation described is a combined size cap across the files, not a fixed file count.
    • E. Video files are not a supported format for image or document input; supported formats are generally limited to PNG, JPG, JPEG, and PDF.
    • F. Image or document input isn't yet supported for prompts added as tools in Copilot Studio agents, which is the opposite of full support.

    Subdomain 3.2: Create prompts and models in AI Hub

    30.Which of the following are documented limitations when using your own data as knowledge in a prompt? (Select all that apply.)(Select 3)

    1. A.Only one-to-many or many-to-one Dataverse relationships are supported, up to two levels deep
    2. B.Virtual tables aren't yet supported as a knowledge source for grounding a prompt's answers
    3. C.Connectors such as Salesforce or Zendesk can only be used for prompts running in Power Automate
    4. D.Any Dataverse column, including multiple-lines-of-text columns, can be used as a filter attribute
    5. E.Knowledge sources support unlimited levels of related-table traversal with no depth restriction
    Show answer & explanation

    Correct answers: A, B, C — Only one-to-many or many-to-one Dataverse relationships are supported, up to two levels deep; Virtual tables aren't yet supported as a knowledge source for grounding a prompt's answers; Connectors such as Salesforce or Zendesk can only be used for prompts running in Power Automate

    • A. Only one-to-many or many-to-one relationships are supported for Dataverse, with up to two levels of relationships selectable for the table added as a data source, matching this statement.
    • B. Virtual table use isn't yet supported as a documented limitation of using your own data in a prompt.
    • C. Connectors are documented as usable only for prompts running in Power Automate, so this correctly describes a real platform restriction.
    • D. Only attributes with specific data types, such as text, number, date and time, choice, currency, and unique identifier, are available as filter attributes, so multiple-lines-of-text columns are not usable this way.
    • E. Relationship traversal is capped at two levels for supported relationship types, so there is a documented depth restriction rather than unlimited traversal.

    Subdomain 3.3: Implement business and process logic

    31.A formula column's Power Fx expression cannot exceed ____ characters in length.

    1. A.500
    2. B.1,000
    3. C.2,000
    Show answer & explanation

    Correct answer: B — 1,000

    • A. 500 characters is too low; the documented maximum formula expression length for a formula column is higher than this.
    • B. 1,000 characters is the documented maximum expression length for a Dataverse formula column.
    • C. 2,000 characters exceeds the documented maximum; the actual limit for a formula column expression is lower.

    Subdomain 3.3: Implement business and process logic

    32.A consultant needs to build a business rule but must configure it so the condition and action apply everywhere, including through the API. Which column types cannot be referenced in a business rule's conditions or actions? (Select all that apply)(Select 3)

    1. A.Unique identifier columns
    2. B.Rollup columns
    3. C.Multiselect choice columns
    4. D.Whole number columns
    5. E.Text columns
    Show answer & explanation

    Correct answers: A, B, C — Unique identifier columns; Rollup columns; Multiselect choice columns

    • A. Unique identifier columns are explicitly listed as unsupported in business rule conditions and actions.
    • B. Rollup columns are explicitly listed as unsupported in business rule conditions and actions.
    • C. Multiselect choice columns don't work with business rules, per Microsoft's documented limitation.
    • D. Whole number columns are fully supported in business rule conditions and actions, with no documented restriction.
    • E. Text columns are fully supported in business rule conditions and actions, with no documented restriction.

    Subdomain 3.3: Implement business and process logic

    33.Which of the following business rule actions are available regardless of whether the rule's scope is Entity or a client-side scope like All Forms? (Select all that apply)(Select 3)

    1. A.Set Field Value
    2. B.Set Default Value
    3. C.Show Error Message
    4. D.Lock/Unlock
    5. E.Set Visibility
    6. F.Recommendation
    Show answer & explanation

    Correct answers: A, B, C — Set Field Value; Set Default Value; Show Error Message

    • A. Set Field Value applies to every scope, including Entity, so it also runs server-side.
    • B. Set Default Value applies to every scope, including Entity, so it also runs server-side.
    • C. Show Error Message applies to every scope, blocking the save either on the form or by returning an error to the calling process server-side.
    • D. Lock/Unlock only applies to model-driven app forms and is ignored when the rule runs server-side under Entity scope.
    • E. Set Visibility only applies to model-driven app forms and is ignored when the rule runs server-side under Entity scope.
    • F. Recommendation only applies to model-driven app forms and is ignored when the rule runs server-side under Entity scope.

    Subdomain 3.3: Implement business and process logic

    34.A business rule needs to run against a table used only through a canvas app and the Dataverse Web API, with no model-driven form defined for that table. Which scope must be chosen for the rule to have any effect?

    1. A.Entity
    2. B.All Forms
    3. C.Specific Form
    4. D.None of the scopes will run without a model-driven form
    Show answer & explanation

    Correct answer: A — Entity

    • A. Entity scope executes server-side and is enforced across canvas apps, the API, and any model-driven forms, so it works even without a form present.
    • B. All Forms scope is client-side and only runs on model-driven forms, so it would have no effect for a table with no such form.
    • C. Specific Form scope is also client-side and tied to a particular model-driven form, so it can't apply here either.
    • D. A scope does exist that runs without any model-driven form present, since Entity scope enforces logic server-side.

    Subdomain 3.3: Implement business and process logic

    35.A maker needs a guided, multi-record process where a deal moves from an Opportunity record to a Quote, then an Order, and finally an Invoice, each with its own set of required fields before advancing. Which tool is designed for this?

    1. A.A multi-table business process flow spanning Opportunity, Quote, Order, and Invoice
    2. B.Four separate business rules, one per table
    3. C.A single rollup column summarizing all four tables
    4. D.A calculated column chained across four tables
    Show answer & explanation

    Correct answer: A — A multi-table business process flow spanning Opportunity, Quote, Order, and Invoice

    • A. A multi-table business process flow is designed to tie together rows from multiple tables, such as opportunity through invoice, into a single guided process with required steps at each stage.
    • B. Four separate business rules would each be scoped to a single table and couldn't provide one continuous guided process spanning the four tables together.
    • C. A rollup column produces a single aggregated value and has no capability to guide users through sequential stages across tables.
    • D. Calculated columns are also limited to spanning two tables at most, and their purpose is computing a display value, not guiding users through a process.

    Want the full experience?

    These are just samples. Practice the full Microsoft Intelligent Applications Builder (AB-410) question bank in quiz mode — free, no signup, with domain practice and exam simulation.

    Practice all 357 questions in quiz mode