Question 11 — AB-620 Practice Exam 5
A Copilot Studio topic must collect information from several systems, process paginated results, apply business rules, and then provide the consolidated information to a generative answers node. The user's query must be available to the retrieval process. Which implementation meets these requirements? (Select one answer)
Use an HTTP request node and pass the combined raw responses directly into Custom data
Use a Power Automate flow that accepts topic input, assembles the results, returns the required Table, and maps that output to a topic variable used by Custom data
Use a generative answers node to call each system directly and store the combined results in its response variable
Use a Power Automate flow that returns an arbitrary JSON object and assign that object directly to Custom data
Correct answers
Use a Power Automate flow that accepts topic input, assembles the results, returns the required Table, and maps that output to a topic variable used by Custom data
Explanations
Correct:
Use a Power Automate flow that accepts topic input, assembles the results, returns the required Table, and maps that output to a topic variable used by Custom data
Power Automate is appropriate when retrieval requires aggregation across systems, pagination, complex filtering, business rules, or Power Platform connectors. The flow can receive the user's query or other topic variables, perform the required processing, and return a Table that Copilot Studio maps to a topic variable for the Custom data field.
Wrong:
Use an HTTP request node and pass the combined raw responses directly into Custom data
HTTP requests are well suited to straightforward retrieval that can be reshaped easily, but raw responses cannot be supplied directly when they do not conform to the required Custom data Table structure. The complex aggregation and pagination requirements also favor a flow.
Use a generative answers node to call each system directly and store the combined results in its response variable
The generative answers node consumes the prepared Custom data rather than performing the described multi-system retrieval and assembly logic itself.
Use a Power Automate flow that returns an arbitrary JSON object and assign that object directly to Custom data
Using a flow is appropriate for the complex processing requirements, but its result must reach the Custom data field as the required Table containing Content, ContentLocation, and Title columns. An arbitrary JSON object does not satisfy that input contract.