Free AI-103: Microsoft Azure AI Apps and Agents Developer Associate Exam Questions and Answers

You are developing a Python app that uses the azure-ai-projects library to connect to a Microsoft Foundry project named Project1. The Foundry resource is named contoso-ai. The app runs with a managed identity and must not use API keys. You write the following code. from azure.ai.projects import AIProjectClient from azure.identity import DefaultAzureCredential project_client = AIProjectClient( endpoint="<value>", credential=DefaultAzureCredential(), ) Which value should you use for endpoint?

• AA. https://contoso-ai.services.ai.azure.com/
• BB. https://contoso-ai.openai.azure.com/openai/deployments/gpt-4o-mini/chat/completions
• CC. https://contoso-ai.services.ai.azure.com/api/projects/Project1
• DD. https://ai.azure.com/projects/Project1/overview

You call a chat model deployment in a Foundry project by using the Python OpenAI client and the chat.completions.create method. Which two arguments are required in the call? Each correct answer presents part of the solution. NOTE: Each correct selection is worth one point.

Select 2 answers.

• AA. temperature
• BB. max_completion_tokens
• CC. model
• DD. messages

An insurance company is designing a hybrid claims workflow in Foundry. You need to assign a component to each requirement. Components: - Deterministic rules engine - Small language model deployment - Multimodal model deployment - Embedding model deployment Each component can be used once, more than once, or not at all. Match the component to each requirement. Target 1: Require manager approval for every claim over $10,000. Target 2: Classify short claim notes into six fixed categories at high volume with low latency and cost. Target 3: Read photos of vehicle damage and write a narrative description for the adjuster.

Drag or tap each item onto its target, then reveal.

You are designing a support assistant that must answer questions by using 40,000 product manuals as grounding data. For each of the following statements, select Yes if the statement is true. Otherwise, select No. NOTE: Each correct selection is worth one point. Statement 1: Azure Translator in Foundry Tools can generate the vector embeddings that are stored in the search index. Statement 2: Azure AI Search can run a single query that combines keyword search and vector search over the same index. Statement 3: An agent in Foundry Agent Service can use an Azure AI Search index as a knowledge tool.

Answer each statement, then reveal.

1. Azure Translator in Foundry Tools can generate the vector embeddings that are stored in the search index.YesNo
2. Azure AI Search can run a single query that combines keyword search and vector search over the same index.YesNo
3. An agent in Foundry Agent Service can use an Azure AI Search index as a knowledge tool.YesNo

A Foundry resource named contoso-ai contains two projects named Project1 and Project2. Key-based authentication is disabled on the resource. A web app uses a system-assigned managed identity to run agents and call model deployments in Project1 only. Security requires that the identity can neither create or delete model deployments nor access Project2. What should you do?

• AA. Assign the Azure AI User role to the managed identity at the scope of Project1
• BB. Assign the Contributor role to the managed identity at the scope of contoso-ai
• CC. Assign the Azure AI Project Manager role to the managed identity at the scope of the subscription
• DD. Assign the Cognitive Services Contributor role to the managed identity at the scope of the resource group

You have a Microsoft Foundry project named Project1 that contains a gpt-image-1 deployment. An app generates product images for a catalog. Designers need 1024x1024 draft images with a transparent background so that they can place each product over different page colors. Drafts are regenerated many times, so the cost per image must be kept low. Which parameter values should you pass to the images.generate call?

• AA. background="transparent", output_format="png", quality="low"
• BB. background="auto", output_format="png", quality="high"
• CC. background="transparent", output_format="jpeg", quality="low"
• DD. background="opaque", output_format="webp", quality="medium"

A law firm stores thousands of contracts. Attorneys want a summary of each contract that consists of the most important sentences copied exactly from the document, so that each sentence can be traced to its position in the source. You plan to use Azure Language in Foundry Tools. What should you use?

• AA. Abstractive summarization
• BB. Extractive summarization
• CC. Entity linking
• DD. Key phrase extraction

You use the images.edit method of a gpt-image-1 deployment to replace only the sky in a landscape photo named photo.png. You will supply a separate mask file. How should you prepare the mask?

• AA. A grayscale JPEG at half the resolution of photo.png, with the sky pixels set to white
• BB. A JSON file of bounding box coordinates around the sky, passed in the mask parameter
• CC. A PNG with an alpha channel and the same dimensions as photo.png, with the sky pixels fully transparent
• DD. A PNG with an alpha channel and the same dimensions as photo.png, with the sky pixels fully opaque

You are designing a multi-agent solution by using Microsoft Agent Framework. When a contract is submitted, a legal agent, a finance agent and a security agent must each review it. None of the reviews depends on the output of another reviewer. A summarizer agent then combines the three reviews into one report. Total processing time must be minimized. Which orchestration pattern should you use?

• AA. Concurrent orchestration that sends the contract to all three reviewers and aggregates their results
• BB. Group chat orchestration in which a manager selects one reviewer to speak in each round
• CC. Handoff orchestration in which each reviewer transfers control of the conversation to the next reviewer
• DD. Sequential orchestration that passes the contract through the legal, finance and security agents in turn

A company is designing a Foundry agent for its HR help desk. The agent must: - Remember each employee's preferred language and office location across separate conversations. - Answer questions from approved HR policy documents that are stored in Azure Blob Storage and change monthly. - Create tickets in an internal system that publishes an OpenAPI 3.0 specification. Which combination of capabilities should you use?

• AA. Conversation thread history for preferences, Azure AI Search for policy documents, OpenAPI tool for ticket creation
• BB. Foundry memory for preferences, Azure AI Search for policy documents, Code Interpreter for ticket creation
• CC. Foundry memory for preferences, Azure AI Search for policy documents, OpenAPI tool for ticket creation
• DD. Foundry memory for preferences, Grounding with Bing Search for policy documents, OpenAPI tool for ticket creation