Pass AI-103, and know Azure AI for real
Study material for AI-103: Developing AI Apps and Agents on Azure. Lessons, exam-format practice, timed mocks and hands-on labs.
| The exam | |
|---|---|
| Certification | Azure AI Apps and Agents Developer Associate (replaced AI-102, retired 30 June 2026) |
| Format | Typically 40–60 questions, 100 minutes; pass mark 700; Python |
| Question types | Multiple choice, multiple response, drag and drop, hot area, build list, case studies, problem/solution sets (no going back) |
| Open book | Microsoft Learn is available during the exam (not Q&A or practice assessments); the clock keeps running |
| Renewal | Yearly, with a free online assessment |
Start here
Roadmap, exam weights, the services map by exam depth, all 19 lessons, 30 practice questions, glossary of renames.
Decision guideChoose the right modelMind map of eight model categories, a decision tree, criteria, scenario table and category comparison.
Cost calculatorWhat does an agent cost to run?Interactive TCO: users, turns and tokens in; an itemised daily bill, one-time costs and what moves the total out.
Lab · Week 1Keyless call to a Foundry modelSetup, expected output, four break-it exercises.
Domains
Lessons and a 34-item practice bank in every exam format.
D2 · 30–35%Generative AI and agents16 teaching cards with CCA-F bridges, and a 22-item practice bank.
D3 · 10–15%Computer vision8 teaching cards and a 16-item practice bank.
D4 · 10–15%Text analysis8 teaching cards and a 15-item practice bank.
D5 · 10–15%Information extraction7 teaching cards and a 16-item practice bank.
Mock · 100 minMock Exam #150 items, all formats, two long case studies, exam-level stems.