Microsoft · AI-103
Build generative apps with Foundry
AI-103 · Lesson 5 of 15 · 38:47
YouTube loads only after you press play. This chapter opens at its timestamp in the full course. Watch on YouTube ↗
In this lesson
Prepare for the AI-103 exam (Azure AI Apps and Agents Developer Associate). In this lesson: the Foundry model catalog and deployments (model = deployment name, Global Standard, provisioned throughput), the project endpoint and the Foundry SDK vs the Azure OpenAI v1 endpoint and the OpenAI SDK, keyless authentication with Entra ID, the Responses API in Python (AIProjectClient, get_openai_client, previous_response_id) vs Chat Completions, and RAG in an application with Azure AI Search. Includes three original practice questions.
Preview the topics
- A developer calls responses.create with model set to gpt-5-mini, the catalog name. It fails, but the deployment named support-chat works in the playground. What should model contain?
- An app must create agents and run cloud evaluations, and the security team has disabled API keys. Which combination should it use?
- A support bot must answer from HR policies that change every week, and cite the policy it used. What’s the best approach?