You are deploying Copilot in Power Apps for end-users within a Model-driven app. The users need to use the Copilot sidecar to ask natural language questions to summarize data and find specific records. Which of the following prerequisites must be met for this built-in agent to function properly and access the data? (Select two)
Select all that apply
check_circle
Correct AnswersA, B
Copilot for Model-driven apps relies heavily on the Dataverse Search infrastructure to quickly retrieve contextual data and synthesize natural language answers. Therefore, Dataverse Search must be enabled at the environment level. Furthermore, Copilot can only reason over data that has been indexed; if a specific table (and its relevant columns) is not added to the Dataverse Search index configuration, Copilot will not be able to 'see' or query that data to answer user questions.
Question 92
A logistics company manages a Dataverse solution where a single 'Fleet' record has a 1:N relationship with thousands of related 'Trip' records. Users need the 'Current Status' column on the 'Fleet' record to automatically switch to 'In Transit' if ANY related 'Trip' record is marked as 'Active', and switch to 'Available' if ALL related trips are 'Completed'. To ensure this updates as close to real-time as possible without causing severe system performance degradation, what is the best architectural approach?
check_circle
Correct AnswerD
Rollup columns (Option 1) calculate asynchronously via a background job (defaulting to every 12 hours), so they cannot provide real-time updates. Formula columns (Option 3) are excellent for intra-record math, but Dataverse formula columns currently have limitations regarding complex cross-table aggregations and filtering of 1:N child records to set a parent state. The standard, reliable architectural pattern for real-time aggregation across child records that drives a parent update is an event-driven approach: a Power Automate flow (or synchronous plugin) triggered on the child 'Trip' record that evaluates the dataset and updates the parent 'Fleet' record.
Question 93
A healthcare provider needs an end-to-end solution to manage patient intake. External patients must submit complex medical history forms online. Internal clinical staff will review these submissions in a data-dense, relationship-heavy UI. Upon approval, an automated backend process must extract structured text from attached PDF medical records, summarize the findings, and post the summary to a secure Microsoft Teams channel. Which architectural combination represents the Microsoft-recommended best practice for this scenario?
check_circle
Correct AnswerD
Power Pages is the purpose-built product for external-facing, secure websites where authenticated or unauthenticated users (patients) can submit Dataverse records. Model-driven apps are the architectural standard for internal staff needing highly structured, data-dense back-office UIs. To orchestrate the backend automation, a Power Automate cloud flow should be used. Within the flow, AI Builder Document Processing extracts structured data from the PDFs, and an AI Hub custom prompt (GPT text generation) can summarize that text before the flow posts it to Microsoft Teams.
Question 94
A custom Copilot Studio agent deployed on a company intranet is configured to execute a Power Automate flow. This flow retrieves an employee's Paid Time Off (PTO) balance from Dataverse. To ensure the flow securely retrieves the balance ONLY for the specific employee chatting with the bot, what security implementation is required?
check_circle
Correct AnswerD
For an AI agent to perform sensitive actions or retrieve personal data on behalf of a user, End-User Authentication must be implemented. By configuring the agent to authenticate users (e.g., via Microsoft Entra ID), the agent obtains a secure authentication token. This token verifies the user's identity and is passed to the backend Power Automate flow, ensuring the flow's actions and Dataverse queries execute strictly in the security context of that authenticated individual, preventing data leakage.
Question 95
You embed the native Copilot control in a Canvas app to allow users to query a custom 'Equipment' table. However, when users ask 'Which equipment requires maintenance?', Copilot provides generic web-based advice on how to maintain equipment instead of querying the actual Dataverse records. What must be configured on the Copilot control to anchor it to your data?
check_circle
Correct AnswerC
The native Copilot control in Canvas apps needs to be explicitly directed on which Dataverse table to reason over. By configuring the 'Data source' (Items) property of the Copilot control to point to the specific Dataverse table (e.g., 'Equipment'), you anchor the AI strictly to that data set. It will then use Dataverse Search to query that specific table to answer user prompts, rather than falling back to general generative AI knowledge.
Question 96
A maker creates a custom prompt in AI Hub designed to rewrite text. The instructions are: `Rewrite the following text to sound professional: [InputText]`. When the maker attempts to test this prompt in the AI Hub designer pane, it fails to execute or expose a place to enter the text. What is the cause of this issue?
check_circle
Correct AnswerA
In the AI Hub custom prompt designer, simply typing brackets like `[InputText]` into the text box does not automatically generate a dynamic variable. The system treats it as literal text. To create a variable that can be tested in the sample data pane (and eventually mapped to dynamic content in Power Automate or Power Apps), the maker must explicitly use the 'Add input' (or 'Dynamic value') button/menu to formally define the parameter and insert its token into the prompt.
Question 97
A Power Automate cloud flow is designed with a 'Do until' loop to poll a third-party API every 12 hours to check if a long-running background rendering task is 'Complete'. The rendering task sometimes takes up to 45 days. Users report that the flow consistently stops running and is marked as 'Failed/Timed Out' after exactly 30 days, even if the task is not complete. What is the fundamental cause of this?
check_circle
Correct AnswerA
Regardless of how you configure loop limits, action timeouts, or wait conditions, a fundamental architectural limitation of Power Automate cloud flows is that a single flow run has an absolute maximum execution duration of exactly 30 days. If a flow run remains active or is waiting/sleeping for longer than 30 days, the platform forcefully terminates it. For processes lasting longer than 30 days, you must re-architect the solution to use disconnected state-machine patterns or Dataverse background workflows.
Question 98
An enterprise has a complex Dataverse-centric solution containing tables, cloud flows, and AI models. To democratize development, they allow citizen developers to build apps in a shared environment. You must enforce a security rule that strictly prevents makers from using the Twitter (X) connector alongside Dataverse in the same flow or app to prevent data exfiltration, while still allowing Twitter to be used in isolated personal productivity flows. How should you apply this governance best practice?
check_circle
Correct AnswerB
Data Loss Prevention (DLP) policies govern which connectors can be used together. By categorizing connectors into 'Business', 'Non-Business', and 'Blocked' groups, you dictate data flow. Connectors in the 'Business' group cannot share data with connectors in the 'Non-Business' group within the same app or flow. By placing Dataverse in Business and Twitter in Non-Business, you prevent them from being used together (stopping data exfiltration) without completely blocking the Twitter connector for isolated, non-business use cases. Tenant isolation controls cross-tenant connections, not connector combinations.
Question 99
A business requirement states that when a 'Project' record is deleted from Dataverse, its associated 'Time Entry' records must NOT be deleted. Instead, the system should simply clear the lookup field referencing the deleted Project on those Time Entry records, leaving them as orphaned records for auditing purposes. How should you configure the table relationship behavior to achieve this?
check_circle
Correct AnswerA
In Dataverse, relationship behaviors dictate how actions on a parent record cascade down to related child records. 'Parental' cascading forces the deletion of child records when the parent is deleted. To keep the child records ('Time Entry') but remove the reference to the deleted parent ('Project'), you must use 'Configurable cascading' and set the 'Delete' action to 'Remove Link'. 'Restrict' would prevent the deletion of the parent record entirely if child records exist. 'Referential' is similar but defaults to Remove Link; configuring it explicitly via Configurable cascading provides absolute control over all cascading actions.
Question 100
A custom 'Feedback' table needs to capture comments from users. The feedback might be regarding an 'Event', a 'Product', or a 'Training Course'. The business wants a SINGLE lookup field on the Feedback form where users can select a record from any of these three distinct tables. What Dataverse feature must you implement to achieve this?
check_circle
Correct AnswerD
Dataverse supports Multi-Table (Polymorphic) Lookup columns. These specialized lookup columns allow a single field on a form to reference records from multiple different tables. This perfectly solves the scenario where a single 'Feedback' record needs to relate to either an Event, a Product, or a Training Course without cluttering the schema and form with multiple separate 1:N lookup columns.
