AI-103 Study Guide

If you're preparing for AI-103, Developing AI Apps and Agents on Azure, you're in the right place. This is the written companion to the practice exams: a plain-language walk through everything the test covers, in one place. Read it front to back, or drop into the domain you're least sure of, and the chapter builds the picture from the ground up.

AI-103 rewards judgment more than recall. Its scenarios hand you a requirement and ask which instrument owns it, and the answer favors restraint: the lightest, purpose-built option that already does the job, over a heavier one that looks plausible. You'll weigh whether a single agent holding several tools beats splitting the work across many, and whether a purpose-built service or a general chat model should read an image. Match the tool to the requirement, prefer the identity-first, least-privilege path, and measure before you change anything.

The guide follows the five official domains, weighted as the real exam weights them: implementing generative AI and agents is the largest at 30–35%, planning and managing an Azure AI solution at 25–30%, and computer vision, text analysis, and information extraction each at 10–15%. Every chapter builds the mental model in plain language, separates the look-alike options with comparison tables and decision trees, and ends with a cheat sheet. Start at the top, or pick a domain from the list beside this page.