Microsoft Certified: Azure AI Apps and Agents Developer Associate (AI-103) Practice Tests

Microsoft retired AI-102 and the Azure AI Engineer Associate certification on June 30, 2026, and AI-103 (Developing AI Apps and Agents on Azure) is the confirmed successor, leading to the new Microsoft Certified: Azure AI Apps and Agents Developer Associate credential. If you searched for AI-102 practice material, AI-103 is the exam Microsoft now points candidates toward, and it is live and schedulable through Pearson VUE today. The center of gravity has shifted hard toward generative AI: per the skills outline dated April 16, 2026, implementing generative AI and agentic solutions is the single largest domain at 30-35%, ahead of planning and managing an Azure AI solution at 25-30%. Computer vision, text analysis, and information extraction solutions round out the outline at 10-15% each. The real differentiator versus AI-102 is Microsoft Foundry: AI-103 tests building generative applications and agents inside Foundry projects, retrieval-augmented generation (RAG) pipelines, multi-agent orchestration, tool-augmented workflows, and agent evaluation and observability, none of which existed in the old AI-102 blueprint. This 409-question bank is built to that current AI-103 outline, weighted to match the generative AI and agentic emphasis rather than treating it as a minor add-on.

AI-103 runs 120 minutes, 20 minutes longer than AI-102's 100-minute slot, and keeps Microsoft's familiar scaled scoring model: you need 700 out of a possible 1,000 to pass. Microsoft does not publish an exact number of exam-day questions for AI-103, but its associate-level exams typically run in the 40-to-60 question range, mixing multiple-choice, drag-and-drop, and scenario-based case studies delivered through Pearson VUE, either at a testing center or via online proctoring. The credential expires annually, like other Microsoft Associate certifications, and renews through a free online assessment on Microsoft Learn, so budget time each year to stay current as Foundry and the underlying services keep shipping features. Expect the exam to lean heavily on applied, scenario-based judgment: choosing the right Foundry service or model for a task, designing RAG ingestion and retrieval pipelines, wiring multi-agent orchestration with tool access controls, and configuring responsible AI guardrails, rather than testing service names in isolation. Because this is a newer exam with a limited public track record of candidate reports, treat any third-party claims about exact question counts or item difficulty with caution and rely on the official Microsoft study guide as ground truth.

AI-103 has no formal prerequisite exam or certification, but Microsoft's candidate profile is explicit: you should have hands-on experience building applications with Python, and unlike AI-102, which accepted either Python or C#, the AI-103 audience profile names Python specifically. You should also be comfortable with general AI and generative AI concepts and familiar with core Azure services, since the exam expects you to collaborate conceptually with solution architects, data scientists, DevOps engineers, and cloud security engineers. The exam is commonly listed at $165 USD in the United States through Pearson VUE, though Microsoft prices exams by the country or region where you sit the test, so check your local price before scheduling. If you are coming from AI-102, expect real overlap on Azure AI service fundamentals, computer vision, and text analysis, but plan to add dedicated study time for Foundry projects, agent building, and RAG, the topics AI-102 never covered. Start with the 30 free questions in this 409-question bank to gauge where you stand, then work through the full set until your accuracy holds steady across all five domains, with extra reps on the generative AI and agentic solutions domain given its 30-35% weight.

The Microsoft Certified: Azure AI Apps and Agents Developer Associate credential, earned by passing exam AI-103 (Developing AI Apps and Agents on Azure), validates the ability to build, manage, and deploy AI-infused applications and agents on Azure using Microsoft Foundry. It is Microsoft's confirmed successor to the Azure AI Engineer Associate certification (exam AI-102), which was retired on June 30, 2026 along with its renewal assessment. AI-103 is live and schedulable through Pearson VUE, and its skills outline, dated April 16, 2026, reflects a substantial rewrite of the old AI-102 blueprint rather than a light refresh.

Where AI-102 organized its content around individual Azure AI services (Vision, Language, Speech, Document Intelligence) plus a small agentic slice, AI-103 restructures around Microsoft Foundry as the unifying platform: choosing and deploying models, building generative applications and RAG pipelines, constructing and orchestrating multi-agent systems, and applying responsible AI controls across all of it. Computer vision, text analysis, and information extraction remain on the exam, but each is now scoped to Foundry-based, multimodal, and generative workflows rather than standalone service calls. Candidates who held or studied for AI-102 will recognize the underlying Azure AI services but should expect a meaningfully different exam.

AI-103 targets Azure AI engineers and developers who use Python and Azure services to plan, implement, and manage AI solutions and agents built on Microsoft Foundry. Per Microsoft's own candidate profile, this person collaborates with business stakeholders, solution architects, data scientists, DevOps engineers, and cloud security engineers to design, implement, and maintain production AI solutions, not just prototype them.

The certification suits developers already comfortable with general AI and generative AI concepts who want to formalize skills in agent building, RAG, and multi-agent orchestration, as well as AI-102 holders whose certification is aging out and who need a current, recognized credential now that AI-102 and its renewal path are retired. It is less suited to complete beginners with no application-development background, since the exam assumes real Python coding experience rather than conceptual familiarity alone.

AI-103 has no mandatory prerequisite exam or certification and no minimum years-of-experience requirement enforced at registration. That said, Microsoft's official candidate profile is specific about the experience it expects: hands-on experience developing applications using Python, plus familiarity with the capabilities of general AI, generative AI, and core Azure services. This is a change from AI-102, whose candidate profile accepted either Python or C#; AI-103 names Python specifically, so C#-only developers will need to build Python fluency before sitting the exam.

Beyond language skills, candidates should be comfortable with foundational Azure AI concepts (models, endpoints, deployments) and should have some exposure to concepts like retrieval-augmented generation, vector search, and agent-based architectures, since the exam tests applying these patterns in Foundry rather than defining them from scratch. Microsoft's official AI-103 study guide and the AI-103T00-A instructor-led course are the recommended starting points for candidates who need to build this baseline before attempting practice questions.

AI-103 runs 120 minutes, 20 minutes longer than AI-102's 100-minute window, and is delivered as a proctored, computer-based exam through Pearson VUE, either at a physical testing center or via online proctoring. Microsoft has not published an exact question count for AI-103, but its Associate-level exams have historically run in the 40-to-60 question range, mixing multiple-choice, drag-and-drop, and scenario-based case-study formats. A score of 700 or higher on a scale of 100 to 1,000 is required to pass, the same scaled-scoring model Microsoft uses across its certification portfolio.

The exam is commonly listed at $165 USD when proctored in the United States, though Microsoft prices exams by the country or region in which you sit the test, so candidates outside the US should confirm local pricing on the official exam page before registering. Like other Microsoft Associate and Expert certifications, this credential expires annually and renews through a free online assessment on Microsoft Learn rather than a full exam retake, provided you renew before expiration. Retakes of a failed attempt are available 24 hours after the first attempt, with longer waiting periods for subsequent retakes.

AI-103 positions developers for Azure AI engineer, AI application developer, and generative AI / agent engineer roles, the same career lane AI-102 served before retirement, now updated for Foundry-based, agentic development. Typical job titles include Azure AI Engineer, AI Solutions Developer, Generative AI Engineer, and Conversational AI / Agent Developer. Third-party salary trackers put Azure AI Engineer pay in the United States at roughly $90,000 to $129,500 for the middle 50% of earners, with a reported average near $111,500 and top earners around $145,000 annually as of September 2026; more experienced specialists in generative AI and agent engineering can command $150,000 or more depending on role and location.

Because AI-103 is the only current Associate-level path for Microsoft Foundry, agentic, and generative AI development skills, it also functions as a forward-looking credential for developers who do not hold AI-102 at all and want to enter Azure AI development directly at the current, Foundry-centered blueprint. Since the certification renews annually through a free Microsoft Learn assessment rather than a full retake, maintaining it costs time rather than money, which keeps it attractive as a resume signal that stays current with a fast-moving platform.