Microsoft · AI-103
Tools, workflows and evaluation
AI-103 · Lesson 6 of 15 · 45:41
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: tool-augmented flows (tool_choice, code interpreter, web search, file search, remote MCP servers, the function calling loop with call_id), multistep reasoning (structured outputs vs JSON mode, prompt chaining and routing, reasoning models and reasoning effort), workflows (Foundry workflows and Microsoft Agent Framework orchestrations), and evaluating models and apps (quality, RAG and safety evaluators, ground truth metrics, continuous evaluation, red teaming). Includes three original practice questions.
Preview the topics
- A model with a function tool answers a question by returning a function_call item. The app shows that item to the user, and the conversation stops there. What’s missing?
- A pipeline parses the model’s JSON into an invoice object, and it crashes whenever a field is missing. What guarantees that every field is there?
- A RAG assistant sometimes adds details that appear in none of the retrieved documents. Which evaluator measures this problem?