Content Understanding for documents
AI-103 · Lesson 15 of 15 · 1:53:54
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: content extraction with Azure Content Understanding in Foundry Tools (text, tables, figures, sections, prebuilt-read and prebuilt-layout without a model, prebuilt-documentSearch for RAG, business analyzers), custom analyzers (base analyzers, field schemas, the extract, classify and generate methods, model names and default deployments, content categories), grounded output (Markdown for language models, spans, sources and confidence scores for human review), and the Python SDK (ContentUnderstandingClient, begin_analyze_binary, begin_create_analyzer). Includes three original practice questions.
Preview the topics
- A team needs the text of scanned manuals, with their headings and tables, as Markdown, without paying for model tokens. Which analyzer fits?
- A custom analyzer must return a contract’s supplier name exactly as printed, with where it was found and a confidence score. How should you define the field?
- A new custom analyzer lists gpt-5.2 as its completion model. The first analysis fails: no deployment is available for that model. What should you configure?