Microsoft Certified: Intelligent Applications Builder Associate (AB-410) (AB-410) (AB-410) — Questions 226–278

The Wait for an approval action pauses the flow run and blocks until a designated approver responds, then resumes along the approve or reject branch. This suspends execution pending human input, which is exactly what Human in the Loop requires.

Before invoking the AI Builder invoice processing model, you must verify that the file is a PDF because the model supports only PDF and JPEG formats. Adding a Condition action immediately after the trigger to check the file extension ensures that non-PDF files are not processed, preventing errors and unnecessary consumption of AI Builder credits.

A collection holds the cart in memory, and the label's Text property recalculates automatically whenever the collection changes. Sum over the Price times Quantity expression produces the running total, and because the label depends on the collection, adding or removing an item updates the display immediately without extra code.

Record ownership defaults to the creating user, regardless of their table-level access. Because the user holds only User access, they cannot assign ownership to another user or team; that requires at least the same User access plus share privileges on the record. Creation therefore sets the creator as owner, satisfying the stem's constraint.

A one-to-many relationship allows a single project to be associated with many tasks, and each task to be associated with one project. Configuring cascade delete to 'Cascade All' ensures that when a project is deleted, all related task records are also deleted automatically. This matches the requirement.

The Filter function with a delegation-friendly condition on Dataverse will retrieve only the matching records from the server. User().Email returns the current user's email, which can be compared to the Owner column. This is efficient and ensures only relevant data is loaded.

Enabling Concurrency Control on 'Apply to each' allows multiple iterations to run in parallel, significantly speeding up processing for large batches. Setting a degree of parallelism (e.g., 10) balances speed and resource usage. However, you must ensure the actions inside the loop are idempotent and can handle concurrent access to shared resources. This configuration is recommended for batch processing to reduce overall run time.

The SharePoint trigger 'When a file is created (properties only)' fires when a new file is added to the specified library. It provides file properties, including the file identifier, which the flow can use to retrieve the file content and pass it to AI Builder. This is the correct trigger for reacting to new files in SharePoint.

A rollup field aggregates values from related child records, so it can sum invoice amounts linked to a customer and display the total automatically. It recalculates as related invoices change, satisfying the stem's requirement for an automatically calculated total revenue value.

Moderation settings filter unsafe or unsupported responses, while grounding settings constrain answers to your indexed organisational data. Adjusting both satisfies the requirement to reduce hallucination, as grounding anchors output in retrieved sources rather than unconstrained model generation.

The Currency data type is designed to store monetary values and supports currency symbols and precision. The Two Options data type (also known as Boolean) stores a yes/no choice. These are the correct data types for the specified requirements. Using them ensures proper formatting, calculations, and user interface controls.

Environment limits in a tenant are governed by the assigned licence SKUs and capacity add-ons, not a fixed platform ceiling. The maximum therefore varies with the organisation's licensing, making it dependent on licensing rather than a constant number.

After adding the 'Invoke an agent' action, you must map the output from the AI Builder model (e.g., sentiment or extracted text) to the agent's input parameters. This ensures the agent receives the necessary data to generate a personalized response. Without this mapping, the agent would not have the context needed.

Sum aggregates numeric values across a table. Using Sum(Products, Price) calculates the total of the Price column for all records. This is the correct function for displaying a total in a label, provided the data is already loaded or the function can delegate to Dataverse.

The Date picker control provides a calendar interface for selecting one date, and its SelectedDate property can be referenced in the gallery's Items filter. This directly satisfies the requirement of picking a single date that drives a filter, without extra configuration for multi-date or time selection.

SortByColumns is specifically designed for delegation with Dataverse. Combining it with a delegable Filter on Status ensures both filtering and sorting occur server-side. This returns all matching rows in the correct order and performs well with large datasets, meeting the requirements.

Dataverse enforces row-level security through security roles, ownership, and sharing, and this applies to every client including canvas apps. When a user queries the table, the platform returns only records they are permitted to see. This guarantees consistent enforcement regardless of how the app is built, satisfying the requirement.

Triggers must be events that initiate a flow, such as a new email arriving, a scheduled recurrence or a SharePoint item change. Sending an email is an output action performed by the flow, so it cannot start one.

A Choice column restricts input to a predefined set of options, so users select only valid Status values rather than typing free text. This enforces consistent categorisation of Project records, which a text or lookup column cannot guarantee.

Run history records each flow run's action-level inputs and outputs, including the exact payloads passed between steps. Opening the failed run and expanding each action reveals the specific data, which is what troubleshooting requires; the designer only shows configuration, not runtime values.

The Copilot control reads its knowledge from the Data sources property. Adding your table or connector there grounds responses in your specific data, which is the supported mechanism for linking Copilot to a particular dataset within the canvas app.

Document Processing is the AI Builder model type designed for extracting structured fields such as totals, dates and vendor names from varied invoice layouts. It satisfies the varied invoices constraint by learning from labelled samples rather than requiring a fixed template.

A Choice column (also known as Option Set) allows users to select from a predefined list of values. It can be a local choice specific to the table or a global choice shared across tables. This is ideal for storing priority levels like High, Medium, Low, as it restricts input to valid options and ensures data consistency.

Dataverse supports delegation for Filter with equality and comparison operators on columns like Status and CreatedOn, so both conditions are translated into a server-side query. This returns only the matching rows and avoids downloading the full table, satisfying the 100,000-row requirement efficiently.

Importing the solution into the target environment transfers the AI components, flows and connection references from development into the production environment, satisfying the stem's requirement to move the solution to production. Without this deployment step, the built artefacts remain confined to the source environment and cannot run in production.

A subgrid on a form displays a list of related records from a table, such as contacts related to an account. It provides built-in filtering, sorting, and view selection. This directly meets the need to show active contacts with quick filter and sort capabilities without adding unnecessary editing features.

A Power Automate flow triggers on the Dataverse change event for the specified column, then dispatches the email action. This satisfies the requirement to automate sending an email whenever a specific field value changes, without custom plug-in code.

The Add Picture control renders a native image picker that captures or selects a file and exposes it as an image value, which can be written directly to a Dataverse image column. Other controls cannot produce image data for upload.

A Parallel branch splits the flow so that each contained action runs concurrently rather than sequentially, reducing total elapsed runtime when steps are independent. This is distinct from conditions or switches, which select a single path instead of executing all branches simultaneously.

Table visibility depends on the security role's Read privilege at table level; solution membership alone grants nothing. Without Read on the custom table, the user cannot see it in the app, making missing privileges the likely cause.

An action is a single operation a flow performs, such as sending an email, creating an item or calling a connector. Together with the trigger that starts the flow, actions form the fundamental building blocks of every Power Automate flow.

Delegation pushes filtering, sorting and aggregation to the data source rather than pulling every record to the client. This limits transferred row counts, so large tables stay responsive and avoid the truncation and slow-loading problems that undermine canvas app performance.

Training is the phase where the model's parameters are adjusted by exposing it to datasets, letting it learn patterns and relationships. This satisfies the AI Hub definition of teaching the model with data, distinct from inference or deployment, where the model applies what it has already learned.

Plugins let Copilot Studio invoke external systems through connectors or APIs, so business logic can retrieve or act on data beyond the conversation itself. This satisfies the integration requirement by extending the copilot's reach into line-of-business services.

An on-premises data gateway brokers connectivity from cloud services to resources behind your firewall without inbound ports, satisfying the requirement to reach an on-premises SQL Server. It relays queries over outbound TLS, so Microsoft Entra ID-authenticated apps can access the database securely without exposing it directly to the internet.

Solutions package flows with their components and support managed and unmanaged layers, enabling export, import and deployment across environments. This satisfies the stem's requirement by providing Application Lifecycle Management, unlike standalone flows created outside a solution.

A solution acts as the container that packages flows, apps, connection references, environment variables and other components together, so the whole unit can be exported and imported across environments. Saving components individually prevents this portability, which the scenario explicitly requires.

This models a many-to-many relationship with additional attributes using an intersect table. The Prescription table links a Visit and a Medication, and the dosage and frequency columns belong to this specific combination. This accurately represents that a visit can have multiple prescriptions, each with its own dosage and frequency.

The 'Apply to each' action has a concurrency control setting that, when enabled, runs multiple iterations in parallel. You can set the degree of parallelism. This allows the flow to process multiple documents simultaneously, reducing overall time. It is the correct action for dynamically processing an array of documents in parallel.

DLP policies are created and edited centrally in the Power Platform admin center, satisfying the requirement for tenant-wide governance of connectors across environments. This administrative surface enforces which connectors can coexist, blocking data movement between business and non-business groups, unlike maker-level or environment-level settings.

Document processing models in Azure AI Document Intelligence handle form-like and invoice layouts, extracting fields across multiple pages within a single document. This directly satisfies the multi-page invoice constraint, unlike fixed-template or single-page extraction approaches.

Recurrence triggers are subject to platform throttling limits, so a five-minute interval can be delayed or skipped when the flow executes too frequently. This constrains the achievable schedule, meaning the trigger cannot be relied upon for guaranteed precise five-minute execution.

Adding the flow via the Power Automate pane binds it to the app, then invoking it from the button's OnSelect property triggers execution on click. This is the supported mechanism for canvas app to Cloud Flow invocation.

A business rule applies client-side logic that makes a field required based on another column's value, directly satisfying the conditional mandatory constraint. Unlike column-level required settings, which are unconditional, business rules evaluate conditions at runtime, so the field becomes mandatory only when the dependent field is populated.

Canvas apps are first-party Power Apps components, so they can be packaged into a solution and transported between environments alongside their dependencies. This satisfies the stem's requirement to identify component types that a Power Apps Solution can contain.

A subgrid is a form component that displays a list of related records in a grid. It can be configured to show records from a related table, such as Contacts related to an Opportunity. The subgrid is read-only by default but can allow opening records. It appears directly on the main form and supports sorting, filtering, and viewing multiple records, which matches the requirement.

A one-to-many relationship from Agent to Property matches the business rule that one agent lists many properties. Referential, Remove Link sets the Listing Agent lookup on related Property records to null when the Agent is deleted, preserving the Property records. This is exactly the required deletion behavior.

Screens appear as top-level nodes in the Tree View pane, letting makers select, reorder and navigate between them while building an app. This satisfies the stem's requirement for items visible in that pane, since Tree View exposes the app's screen hierarchy alongside controls nested within each screen.

A gallery renders only what its Items property supplies, so an empty Items value produces no rows regardless of an active Dataverse connection. Populating Items with the table or a filtered collection restores the display.

AI Builder prediction models require you to first create the model, then select the Dataverse table and target column before training. Creating the Prediction model initiates the workflow that consumes the historical churn data as training input.

An autonumber column automatically generates a unique, sequential number when a new record is created. It is read-only for users, ensuring the number cannot be edited. Autonumber columns support formats like 'ORD-{SEQNUM:5}' and are ideal for order numbers that need to be easily referenced. This directly meets the requirement.

The 'Start and wait for an approval' action sends approval requests that can be acted upon in Microsoft Teams, email, or the Power Automate mobile app. It pauses the flow until a response is received, making it ideal for human-in-the-loop business processes. This action supports Teams integration out of the box, allowing approvers to respond directly within Teams.

Security roles define the privilege matrix controlling which users can create, read, update, and delete records within Dataverse environments and tables. They are a core component of Dataverse's role-based access model, alongside business units and teams.