Microsoft AI-500 Exam Prep
Designing and Implementing Multi-Agent AI Solutions (Page 4 )

Updated On: 17-Oct-2026
View Related Case Study

DRAG DROP (Drag and Drop is not supported)
You have a Microsoft Agent Framework solution for customer support that includes the following agents:
● A triage agent
● A refund specialist agent that runs in the same process as the triage agent
● A compliance review agent that is exposed by a partner team as a remote service
You need to identify which integration components the triage agent will use to coordinate interactions with the other agents.
What should you identify for each agent? To answer, drag the appropriate components to the correct agents.
Each component may be used once, more than once, or not at all. You may need to drag the split bar between panes or scroll to view content.
Note: Each correct selection is worth one point.
Select and Place:

  1. See Explanation section for answer.

Answer(s): A

Explanation:





Box 1: Agent-as-tools
The agent-as-tools component can be used to coordinate interactions between the triage agent and the refund specialist agent within the Microsoft Agent Framework.
In the agent-as-tools pattern (often implemented via methods like .AsAIFunction() in the SDK), you expose the sub-agent (the refund specialist) directly to the primary agent (the triage agent) as an invocable tool. Because they run in the same process, this architectural pattern provides distinct advantages over alternative patterns like explicit handoffs:
Central Control: The triage agent retains full ownership and overall context of the user's chat session.
Execution Delegation: When a customer asks for a refund, the triage agent triggers the refund specialist just like a typical function or plugin.
Loop Closure: The refund specialist processes the request within the same runtime process and returns its result back to the triage agent. The triage agent then decides how to phrase the final response to the user).
Box 2: The Agent-to-Agent (A2A) protocol
To coordinate interactions between a triage agent and a remote compliance review agent across team or partner service boundaries, the necessary capability within the Microsoft Agent Framework is Agent-to-Agent
(A2A) collaboration.
Specifically, the system requires the following core architectural component:
An Orchestrator (utilizing an Agent-to-Agent Handoff protocol)
The triage agent relies on an Orchestrator—specifically configured with either an explicit code/graph-based
Workflow (such as a sequential pipeline) or a Handoff Orchestration pattern—to coordinate the request.
Because the partner team's compliance review agent is exposed as a remote service, the orchestrator utilizes the A2A protocol layer to safely discover, send standardized payloads, and bridge the network/team boundary without needing direct access to the partner's internal code or model.


Reference:

https://learn.microsoft.com/en-us/agent-framework/overview https://learn.microsoft.com/en-us/agent-framework/journey/agent-to-agent



View Related Case Study

DRAG DROP (Drag and Drop is not supported)
You are designing an enterprise automation solution that has a web portal for users and the following types of workflows:
● Routine workflows that use prompt-defined agents, approved knowledge stores, and HTTPS tools
● Modernization workflows that use custom Microsoft Agent Framework code packaged as container images with predefined handoffs and task states
You need to recommend Azure services that meets the following requirements:
● The routine workflows must run agents in a fully managed environment.
● The user portal must be deployed as a managed platform as a service (PaaS) web app.
● The modernization workflows must run agents in serverless containers that support automatic scaling.
What should you recommend for each requirement? To answer, drag the appropriate services to the correct components. Each service may be used once, more than once, or not at all. You may need to drag the split bar between panes or scroll to view content.
Note: Each correct selection is worth one point.
Select and Place:

  1. See Explanation section for answer.

Answer(s): A

Explanation:






Box 1: Microsoft Foundry Agent Service
To fulfill these specific requirements, use the Microsoft Foundry Agent Service.
Fully Managed Environment: The requirement specifies that the workflows must run agents in a fully managed environment. Microsoft Foundry Agent Service manages the underlying hosting, compute resources, scaling,
security, and session isolation completely out of the box, requiring no infrastructure maintenance from your development team.
Prompt-Defined Agents: The service directly supports Prompt Agents. You define these entirely through configuration (instructions, model selection, and parameters) directly within the portal or programmatically via
SDKs/APIs, matching your workflow strategy.
Approved Knowledge Stores & HTTPS Tools: It natively integrates with enterprise knowledge bases (such as
Azure AI Search, SharePoint, and Fabric IQ) and enables seamless interaction with external systems, APIs, or
Model Context Protocol (MCP) servers using secure HTTPS communication.
Box 2: Azure Container Apps
The component you should use for the modernization workflow is Azure Container Apps.
Serverless Container Platform: Azure Container Apps is purpose-built for running custom container images without the overhead of managing complex Kubernetes infrastructure.
Automatic Scaling: It natively supports event-driven automatic scaling (via KEDA), which can scale your containerized agents down to zero when idle or scale them out horizontally based on workload demands.
Microsoft Agent Framework Integration: It provides the flexible hosting environment required to run custom pro-code Microsoft Agent Framework runtimes, allowing you to control dependencies while leveraging
Microsoft Foundry for multi-agent orchestration and observability.
Box 3: Azure App Service
Azure App Service is the correct component to use for deploying a managed platform-as-a-service (PaaS) web portal.
Managed PaaS: It provides a fully managed platform for building, hosting, and scaling web apps without managing underlying infrastructure.
Web App Focus: It is specifically built for web applications and front-end portals, whereas options like Logic
Apps are for workflow orchestration and Container Apps or Foundry Agent Services serve different backend/AI
agent layers.


Reference:

https://learn.microsoft.com/en-us/azure/foundry/agents/overview https://azure.microsoft.com/en-us/products/app-service



View Related Case Study

You have a Microsoft Foundry agent that handles customer support chats. The agent has two tools named get_contract_terms and calculate_refund.
On turns where refunds must be calculated, the agent inconsistently calls the tools.
You need to reliably force refund-estimate turns to call calculate_refund.
Which request configuration should you use?

  1. tool_choice: {"type": "function", "function": {"name": "calculate_refund"}}
  2. tool_choice: "auto"
  3. instructions: "Use calculate_refund for refund estimates."
  4. tool_choice: "required"

Answer(s): A

Explanation:

To reliably force a refund-estimate turn to call a specific tool, you must explicitly configure the tool_choice parameter in your agent request configuration to target that function instead of using the default "auto" setting.
Python SDK (azure-ai-projects)
When creating and processing the run for a refund-estimate turn, define the tool_choice payload specifying the exact function name:
run = project_client.agents.runs.create_and_process(
thread_id=thread.id,
agent_id=agent.id,
tool_choice={"type": "function", "function": {"name": "calculate_refund"}}
)


Reference:

https://learn.microsoft.com/en-us/azure/foundry/agents/concepts/tool-best-practice



View Related Case Study

You have a multi-agent solution in Microsoft Foundry. Every request begins with the same 1,800-token instruction block and Model Context Protocol (MCP) tool definitions, and then appends a unique user message.
Input-token costs and Time to First Token (TTFT) increase during peak hours.
You need to reduce the input-token costs and TTFT for requests that share common instructions and tool definitions. The solution must meet the following requirements:
● Reuse cached work only when the common prefix matches exactly.
● Generate a new completion for each user request.
● Minimize application changes.
Which type of caching should you use?

  1. retrieval caching
  2. semantic caching
  3. response caching
  4. prompt caching

Answer(s): D

Explanation:

The best option for your requirements is Prompt Caching.
Prompt Caching is specifically designed to recognize when multiple API requests share an identical starting block of text. It meets all your constraints out of the box
Exact Prefix Matching: Prompt caching checks the sequence of tokens from the very beginning of the prompt. It matches and reuses the Key-Value (KV) cache only if the initial segment (including your system instructions and Model Context Protocol/MCP tool definitions) is 100% identical.
New Completion Per Request: Unlike traditional application-level caching or semantic caching, which intercept the request to return an entirely pre-cooked answer—Prompt Caching only caches the pre-evaluation of the input tokens. The model still executes a fresh, unique completion for each user's appended message
Minimize Application Changes: Prompt caching is enabled by default in Microsoft Foundry for supported models and deployments. It requires zero code changes to activate. You only need to ensure that the static components (instructions and MCP tool schemas) are consistently placed at the very beginning of the prompt array, with the dynamic user message appended at the absolute end.


Reference:

https://learn.microsoft.com/en-us/azure/foundry/openai/how-to/prompt-caching



View Related Case Study

You have a Microsoft Foundry project that contains an incident triage agent.
You have a Model Context Protocol (MCP) server registered in the organizational tool catalog. The MCP server exposes two tools named docs_search and deployment_delete.
You need to ensure that the agent can only invoke docs_search.
What should you configure?

  1. the project details
  2. the agent run configuration
  3. the agent tool configuration
  4. the agent instructions

Answer(s): C

Explanation:

To ensure that your incident triage agent can only invoke tool one tool, and is restricted from invoking the second, from the registered Model Context Protocol (MCP) server, you must configure an allowlist policy during the tool integration process.
Here is exactly what must be set up within Microsoft Foundry:
Tool Association with an Allowlist
When adding or attaching the MCP server from the organizational tool catalog (managed via Azure API Center)
to your project's agent, you must specify which tools the agent is permitted to use.The Parameter to Configure:
You must define the allowed_tools property.
The Value: Set allowed_tools=["docs_search "], or explicitly select only docs_search in the Azure AI Foundry portal UI under the agent's tool configuration settings.
By setting this up, Microsoft Foundry's runtime gateway enforces governance by default, preventing the underlying large language model (LLM) from discovering or executing the other tool even though the remote
MCP server exposes both.


Reference:

https://learn.microsoft.com/en-us/azure/foundry/agents/how-to/tools/model-context-protocol



Viewing page 4 of 23
Viewing questions 16 - 20 out of 107 questions


Post your Comments and Discuss Microsoft AI-500 exam prep with other Community members:

AI Tutor AI Tutor 👋 I’m here to help!