Q1.Which framework should an organization adopt to meet the requirements of an AI agent that strengthens the sales cycle?
Select the best option in the answer area.
Show answer
Microsoft Copilot Studio is part of Power Platform and must be designed from the perspectives of reliability, security, cost optimization, operational excellence, and performance efficiency.
Therefore, the Microsoft Power Platform Well-Architected framework, which systematizes these design principles, is the most suitable set of best practices.
For the conversational user experience as well, the framework’s experience-optimization pillar provides “Recommendations for designing a conversational user experience (XO:10),” which directly addresses conversational AI experience design and is therefore optimal.
Success by Design is an implementation methodology for Dynamics 365 and is not direct guidance for conversational UX design; the ALM Accelerator is a development lifecycle management tool; and the Cloud Adoption Framework targets an overall adoption strategy for Azure, so none of these are the direct answer to this question.
Power Platform Well-Architected
Recommendations for designing a conversational user experience
Q2.A company uses Microsoft Dynamics 365 Finance to manage accounts payable and to collect customer receivables.
You are designing an AI-driven financial process that must meet the following requirements.
– Identify overdue vendor invoices and unpaid balances with AI-generated insights.
– Reduce the time required to review overdue invoices and payment history.
Recommend the Microsoft Copilot capabilities to include in the design.
Select the appropriate option in the answer area.
Show answer
This question asks which Copilot capabilities address two business requirements in Dynamics 365 Finance: accounts payable and receivables collection.
First, AI summaries with Copilot summarizes invoice and balance information so you can grasp the content quickly, making it ideal for identifying overdue invoices and unpaid balances.
You can understand the situation without reviewing each transaction in detail.
Next, the Collections coordinator summary specializes in receivables collection and reduces review time by organizing and presenting overdue status and payment history.
Agent management is for administrative purposes, the Account Reconciliation agent is for reconciliation, the Supplier Communication agent is for external communication, and Analyze demand plans is for demand analysis, so none of these fit the requirements.
Microsoft Dynamics 365 Finance documentation
Q3.A company runs an AI business solution.
You need to extend the solution so that Microsoft 365 Copilot can call external logic hosted on Azure services.
What should you include in the solution?
Show answer
This question asks how to integrate Microsoft 365 Copilot so it can call external logic (APIs or services) on Azure.
This requirement needs a standardized means of connecting to external systems.
Therefore, Microsoft Power Platform connectors provide an integration layer that links to external APIs and Azure services, allowing Copilot to call external logic securely and easily, which makes them optimal.
Connectors let Copilot invoke Azure Functions, web APIs, and more.
In addition, they offer the advantage of low-code extensibility while ensuring reusability, governance, and security.
Copilot Studio skills are for extending agent capabilities, and custom engine agents are for advanced, bespoke orchestration, so neither is the best answer.
Extend Microsoft 365 Copilot with actions
Q4.When building a custom AI agent, which components should you incorporate into the application lifecycle management (ALM) process?
Each correct answer presents part of the solution.
Show answer
This question asks about the ALM composition of a custom AI agent that combines Dynamics 365 and Power Platform.
First, a Microsoft Power Platform solution is the standard ALM unit for migrating and managing Copilot Studio and Power Platform components across environments, and it is essential for deploying the agent.
Meanwhile, custom business logic for Dynamics 365 Finance and Supply Chain Management is implemented in X++, so an X++ model is required as the ALM component for managing and distributing business logic outside the application.
A ZIP package is too generic to be an ALM standard, and an Azure package is not part of the core composition.
A CSU package is a distribution unit for Commerce and is not suitable for this scenario.
Application lifecycle management (ALM) with Microsoft Power Platform
Q5.You need to propose a Microsoft Power Platform solution for customer support.
The solution must incorporate the AI capabilities of Microsoft Power Automate and meet the following requirements.
– Use a centralized workspace for AI models.
– Generate short summaries from large volumes of unstructured text such as case notes and transcripts, without additional training or coding.
Select the appropriate option for each requirement.
Show answer
This question asks about the building blocks for using AI in Power Platform.
First, Azure OpenAI Foundry (formerly Azure AI Studio) is a workspace that centralizes model management, deployment, and prompt management, making it optimal as the core of the AI foundation.
The name has recently evolved to Azure AI Foundry, positioned as an integrated enterprise AI development platform.
Next, an AI Builder prebuilt prompt uses predefined prompts to perform text summarization without additional training or coding, making it optimal for the requirement.
It is easy to use from Power Automate and quickly summarizes unstructured data.
Copilot Studio is for agent development, Dataverse is a data foundation, and Azure OpenAI is a standalone foundation service, so none fit simple usage.
Microsoft Foundry documentation
AI Builder overview
Q6.You are designing a testing solution for a Microsoft Copilot Studio agent.
To ensure the agent generates accurate, context-appropriate responses, you need to validate prompt engineering best practices.
Which prompt validation technique and metric should you include in the solution?
Select the appropriate option in the answer area.
Show answer
This question asks how to validate prompt quality in Copilot Studio.
First, using prompts with varied phrasing tests the same intent expressed in different ways, which validates the agent’s ability to handle the diversity of natural language and is therefore optimal.
Because real users use a variety of wording, this validation is important.
Next, response relevance and accuracy is the central quality metric that evaluates whether a response is appropriate and accurate for the user’s intent.
Word count and response time are performance metrics and are insufficient for evaluating quality validity.
Excluding jargon and single-word prompts do not reflect real-world operation and are not appropriate.
Microsoft Copilot Studio documentation
Prompt engineering techniques
Q7.A company uses Microsoft Dynamics 365 Finance and Operations apps.
The company plans to use Microsoft Copilot in-app help and guidance to generate answers for internal users.
You need to add an additional knowledge source for the business process.
The solution must not add a new topic to the Copilot agent of the Finance and Operations app.
Which knowledge source should you add?
Show answer
This question asks how to add existing business knowledge to Copilot while not creating a new topic.
In this case, a method that reuses existing documents as-is, rather than a structured conversational design, is appropriate.
Therefore, a file upload directly ingests documents such as existing business manuals and procedure guides as a knowledge source, allowing Copilot to reference them and generate answers, which makes it optimal.
This makes it possible to extend knowledge without creating a new topic.
In addition, it is also important that you can add knowledge quickly on a document basis and generate answers aligned with internal business processes.
Dataverse is for structured data, a public website is for external information, and Azure AI Search is an advanced search platform that is excessive for this requirement.
Add knowledge to generative help and guidance with Copilot
Q8.You plan to implement an AI business solution for a consumer goods manufacturer.
You need to create agents that meet the following requirements.
– Orchestrate the fulfillment and shipment of sales orders to customers.
– Analyze historical data and trends for inventory replenishment.
Select the appropriate agent type for each requirement.
Drag each option to the matching requirement’s drop zone, or tap to place it. Click a filled zone to send it back. (The same option may be used more than once.)
Prompt-and-response
Task
Show answer
This question asks you to choose the appropriate agent type based on the agent’s role.
First, an Autonomous agent can independently execute and coordinate multi-step business processes, making it optimal for end-to-end orchestration such as sales order fulfillment and shipment.
These involve decision-making and workflow control, so an autonomous type is appropriate.
Meanwhile, a Task agent specializes in a specific task (such as data analysis or pattern detection), making it suitable for limited, well-defined processing like historical data and trend analysis.
Prompt-and-response is for dialog-centric uses and is not suited to controlling business processes or analytical processing.
Microsoft Copilot Studio documentation
AI architecture design
Q9.A company is designing a Microsoft Power Platform solution that deploys existing AI models to reduce manual work in business processes.
You need to identify the solution’s metadata and telemetry and calculate the return on AI investment (ROAI).
What should you use?
Show answer
This question asks which tool to select to measure the return on AI investment (ROAI).
The Business value toolkit is the official tool for measuring the value of Power Platform and AI adoption, and it is important that it can quantitatively evaluate ROAI by leveraging metadata, telemetry, and usage data.
It visualizes the effects of operational efficiency and automation, clarifying the business impact.
In addition, it provides templates for value tracking, cost reduction, and ROI analysis, allowing you to systematically organize the metrics needed for decision-making.
The Cloud Adoption Framework and Success by Design are adoption methodologies, and the admin center is for operational management, so none fit this question.
Capture value with the Business value toolkit and communicate it to stakeholders
Q10.A retail company uses Microsoft Copilot Studio agents for its customer support operations.
The team is reviewing and validating the following two prompts.
– A prompt containing the instruction “Assist the customer as much as possible.”
– A prompt that helps retrieve product information from the knowledge base.
You need to ensure the agent returns consistent, accurate responses.
Which action should you choose for each prompt?
Select the appropriate option in the answer area.
Show answer
This question asks about the basic principles for improving prompt quality in Copilot Studio.
First, vague instructions lead to inconsistent output, so it is important to rewrite them as clear, specific task instructions.
This stabilizes the model’s behavior and produces the expected responses.
Next, when using a knowledge base, constraining responses to only the source information prevents hallucination and ensures accuracy.
Limiting the scope eliminates unnecessary guessing and improves reliability.
Designs that leave ambiguity or allow open-ended answers reduce consistency and accuracy and are therefore inappropriate.
Microsoft Copilot Studio documentation