Enforcing Approved Models for Prompt-Driven Copilot Studio Responses
A company uses an agent in Copilot Studio that produces structured responses using a reusable instruction template. The responses follow the required tone and format. During a compliance review, the team discovers that the responses were generated using a model that is not on the approved list. The company requires that the reusable instruction template continues to be used and the model used for prompt-driven responses meets governance requirements. The solution must ensure that prompt-driven responses are generated using an approved model. You need to configure the solution that meets the requirements. Which two actions should you perform? Each correct answer presents part of the solution. NOTE: Each correct selection is worth one point.
Community Insight
The question tests keeping the reusable instruction template while enforcing model governance; the trap is assuming publishing, citations, or Azure AI Search grounding silently switches or constrains the underlying model.
In Copilot Studio, structured, tone-compliant responses rely on a reusable instruction template, and model governance requires that the template in use carries an approved-model requirement and is actually the template applied to the agent. This page explains why adding an approved-model line to the template plus verifying the correct reusable template is applied (answers C and E) is the required two-part configuration.
The most common wrong pick is A (publish the agent so the model updates automatically), because learners assume publishing refreshes the generative model — publishing only deploys a version and does not repoint prompt-driven responses to an approved model. Citations (B) and Azure AI Search grounding (D) improve traceability or retrieval but do not govern which model generates the answer.
Community Discussion (3 comments)
Comments & Corrections
No comments yet — spotted an error or have a note? Share it below.
Expert Analysis
Why the Answer Is Correct
Answer C adds an explicit approved-model governance rule to the reusable instruction template, which keeps the template in service (a stated requirement) while encoding the compliance constraint directly where prompt-driven responses are shaped. Answer E closes the loop by confirming that the reusable template with that governance rule is the one actually bound to the agent that produces the structured response — otherwise the instruction never reaches the generative pipeline. Together the two actions satisfy both stated requirements: the reusable template continues to be used, and prompt-driven responses are produced under the approved-model governance rule. This is the classic Copilot Studio pattern of enforcing policy at the instruction-template layer and validating template assignment on the agent.Why the Other Options Are Wrong
Answer A is incorrect because publishing an agent deploys the current configuration; it does not automatically swap the model used by prompt-driven responses to an approved one, so it does not enforce governance. Answer B is incorrect because enabling citations provides traceability and source transparency for compliance evidence, but citations say nothing about which model generated the response. Answer D is incorrect because configuring Azure AI Search as the grounding source governs where answers are retrieved from, not which model performs generation, so it cannot satisfy the approved-model requirement. None of A, B, or D preserve-and-extend the reusable instruction template the way C and E do.Community Comment Notes
All three visible commenters independently selected CE; Snoopy70 and Joselito79 both asserted "The correct answers are: C... E", with Joselito79 explaining that "The reusable instruction template must remain in use, and adding the model-governance instruction ensures prompt-driven responses..." meet the requirement, and massnonn simply concurred that it is the correct answer. This unanimous learner consensus aligns with the requirement wording: keep the template, add the approved-model rule, and verify the template is the one applied to the responding agent.Official Reference
Exam Strategy
When a question pins a requirement to an existing artifact (here, the reusable instruction template), favor answers that modify and verify that artifact rather than answers that introduce unrelated features like citations or a new grounding source. For governance questions, remember that model selection and policy live in configuration and instructions, not in publishing actions.
Frequently Asked Questions
Why doesn't publishing the agent update the model used by prompt-driven responses?
Publishing deploys the agent's current configuration as a version; it does not repoint generative answers to an approved model. Model and governance settings are configured separately and must be verified before publishing.
Why are enabling citations and configuring Azure AI Search wrong for enforcing an approved model?
Citations add source traceability for audit evidence, and Azure AI Search controls the grounding/retrieval source. Neither constrains which model generates the response, so neither satisfies the approved-model requirement.