AI-103
AI-103: Developing AI Apps and Agents on Azure
Everything you need for Microsoft exam AI-103: format, passing score, domain weights, a study plan and free sample questions.
- Duration
- 120 min
- Passing score
- 700 / 1000
- Price (US)
- $165
- Replaces
- AI-102
AI-103 is Microsoft’s exam for developers who build generative AI apps and agents on Azure with Microsoft Foundry and Python. Passing it earns the Microsoft Certified: Azure AI Apps and Agents Developer Associate credential. It succeeded AI-102, which Microsoft retired on June 30, 2026.
Compared with its predecessor, the exam puts much more weight on agents, retrieval-augmented generation (RAG) and running AI solutions in production. Classic services such as vision and language are still tested, but they make up a smaller share of the score.
What the exam covers
| Skill area | Weight |
|---|---|
| Plan and manage an Azure AI solution | 25–30% |
| Implement generative AI and agentic solutions | 30–35% |
| Implement computer vision solutions | 10–15% |
| Implement text analysis solutions | 10–15% |
| Implement information extraction solutions | 10–15% |
The first two areas together account for well over half of your score, so that is where most of your study time should go.
AI-103 guides
- AI-103 exam format: questions, passing score, time and costHow many questions AI-103 has, the passing score, time limit, price, languages and how to book it.
- AI-103 vs AI-102: what changed?The differences between the new AI-103 and the retired AI-102 exam, and what it means for your preparation.
- AI-103 study plan: pass in 4 weeksA practical four-week AI-103 study plan weighted by the official exam domains, with hands-on tasks for each week.
- AI-103 sample questions: 5 worked examplesFive worked AI-103 sample questions across the exam domains, each with an explained answer.
- AI-103 plan and manage an Azure AI solution explainedWhat the 25–30% plan and manage domain of AI-103 covers: choosing Foundry services, deployment options, security, monitoring and responsible AI.
- AI-103 generative AI and agentic solutions explainedThe largest AI-103 domain explained: building generative apps with RAG, agents with tools and memory, and operationalising them in production.
- AI-103 computer vision solutions explainedWhat the AI-103 computer vision domain covers: image and video generation, multimodal understanding, Content Understanding and responsible AI for visuals.
- AI-103 text analysis solutions explainedWhat the AI-103 text analysis domain covers: entity and sentiment extraction, structured JSON output, translation, and speech as an agent modality.
- AI-103 information extraction solutions explainedWhat the AI-103 information extraction domain covers: ingestion and indexing, hybrid and vector search for grounding, and document field extraction.
- Is AI-103 worth it? Who should take the examIs AI-103 worth taking? Who benefits, what it costs in time and money, and when another Azure certification is the better choice.
- Free AI-103 practice questions: 20 with answers20 free AI-103 practice questions weighted like the real exam, with an explanation for every answer.
- AI-103 study resourcesWhich AI-103 resources are worth your time: official Microsoft material, hands-on practice, and what to skip.
- AI-103 cheat sheetA one-page AI-103 reference: domain weights, key concepts, the distinctions the exam tests and the traps to avoid.
- Is AI-103 hard? Exam difficulty explainedIs AI-103 hard? Yes, it is a real associate-level developer exam: Python is assumed and 55-65% of the score is Foundry plus generative AI and agents.
- AI-103 career: Azure AI developer rolesWhich jobs AI-103 maps to: Microsoft tags it AI Engineer and Developer. What the Azure AI Apps and Agents credential proves, and what to take next.
- AI-103 renewal, validity and retake rulesAI-103 is valid one year, renewed free by an unproctored Learn assessment in a six-month window. Retake limits, expiry and the AI-102 retirement.
- AI-103 study scenarios: three example profilesStudy AI-103 from an AI-102 background, as a Python back-end developer or as a notebook data scientist: domain order, hours and the trap for each.
- AI-103 prerequisites: what you need firstWhat to know before AI-103: Microsoft's audience profile, the Python and Azure background that helps, what you can skip, and a self-check before you start.