Microsoft Certified: Azure AI Apps and Agents Developer Associate

---

Exam at a Glance

AI-103 measures whether you can do the day-to-day work of an AI developer on Azure: choosing and deploying models in Microsoft Foundry, building generative AI apps and agents in Python, wiring in retrieval, tools, speech and vision, and keeping the solution secure, evaluated and observable.

 

What the Exam Covers

Five domains with unequal weights: generative AI and agentic solutions plus planning and managing the solution make up most of the exam; the three service domains are 10 to 15 percent each, all framed around language models and Foundry Tools rather than the older standalone services.

 

Plan and manage an Azure AI solution — 25-30%

Choosing models and Foundry services for each task, including agent memory, tool and knowledge integrations; infrastructure design, deployment options and CI/CD for Foundry projects; quotas, scaling, rate limits and cost; monitoring performance, drift, safety events, grounding quality and index health; managed identity, private networking, keyless credentials and role policies; and responsible AI: safety filters, guardrails, evaluators, trace logging, approval workflows and agent oversight controls.

 

Implement generative AI and agentic solutions — 30-35%

Deploying and consuming LLMs, small, code and multimodal models; RAG and multistep workflows; evaluating for fabrications, relevance, quality and safety; connecting apps to a Foundry project; agents with defined roles, tool schemas, retrieval, function calling and memory; integrating APIs, knowledge stores and custom functions; multi-agent and autonomous workflows with safeguards and approval controls; prompt engineering, model parameters, self-critique loops and observability.

 

Implement computer vision solutions — 10-15%

Image and video generation, inpainting and mask-based edits; multimodal understanding: captions, visual question answering, accessibility alt text, Azure Content Understanding in Foundry Tools for images and video; and responsible AI for visual content, including indirect prompt injection through text in images.

 

Implement text analysis solutions — 10-15%

Language-model text analysis (entities, topics, summaries, structured JSON), sentiment, tone, safety and sensitive-content detection, translation with Azure Translator in Foundry Tools or LLM-powered flows; and speech: speech to text and text to speech for agents, custom speech models, audio reasoning, speech translation.

 

Implement information extraction solutions — 10-15%

Retrieval and grounding pipelines: ingesting and indexing documents, images, audio and video, semantic, hybrid and vector search, enrichment skills, RAG ingestion with OCR, retrieval wired into agent tools; and document extraction with OCR, layout analysis and field extraction, plus Content Understanding analyzers producing structured or markdown output.

 

The full bullet-level list lives in the official study guide. Treat it as your checklist:

The Official AI-103 Study Guide

 

How This Maps to CertiAce Practice Modules

The CertiAce question bank follows the four learning paths of the official AI-103 course, not the five skills-measured areas. Plan-and-manage and information-extraction skills are drilled where they are taught:

 

Where AI-103 Fits

AI-103 has no certification prerequisites: the certification page lists none. If you are new to AI on Azure, starting with AI-901 is strongly recommended; if you already build AI solutions on Azure with Python, go straight to AI-103. Earning Azure AI Apps and Agents Developer Associate (AI-103) satisfies the prerequisite for Microsoft Certified: Multi-Agent AI Solutions Expert. You must also pass AI-500: Designing and Implementing Multi-Agent AI Solutions, currently in beta as of September 8, 2026.

Certification Page: Azure AI Apps and Agents Developer Associate

Microsoft Certified: Multi-Agent AI Solutions Expert (beta)

 

Before You Start

The exam assumes real development experience: Python plus generative AI and Azure familiarity, per the audience profile. Anything unfamiliar below is where your preparation should start:

 

---

Step-by-Step Study Plan

How long you need depends on where you start. Treat these estimates as planning guidance and adjust them to your starting knowledge and weekly study hours:

 

Step 1: Read the Official Study Guide

Skim the full skills-measured list once, and mark every bullet you could not explain to a colleague. That marked-up list is your personal syllabus: everything else in this plan exists to clear it. Note the naming: the current list says Microsoft Foundry and Foundry Tools where older material says Azure AI Foundry and Azure AI services, and older standalone services such as Custom Vision, Video Indexer and custom question answering do not appear at all.

The Official Study Guide

 

Step 2: Schedule Your Exam

Choose a realistic target date after reviewing the skills list and your available study time. A date on the calendar turns studying into a countdown, so pick one using the prep-time table above and plan backwards from it.

Certification and Exam Details Page

 

Step 3: Work Through the Official Learning Path

Complete the AI-103 course modules on Microsoft Learn: four learning paths that map one-to-one to the CertiAce practice modules above. Start with the generative AI apps path, then the agents path, the largest and the heaviest skill area. Flag anything that needs hands-on practice. You will come back to those in Step 5.

Official Learning Path Course Page

 

Step 4: Watch the Study Cram

No free full-length AI-103 course exists yet: the exam is young. The best substitute is John Savill's AI-103 Study Cram: about two hours of whiteboard explanations across every domain, published in August 2026. Watch it after the learning path as a second pass, and again in your final days; it is a cram, not a course, so it complements the modules and labs rather than replacing them.

John Savill's AI-103 Study Cram

 

Step 5: Get Hands-On Practice

AI-103 rewards real development experience with Microsoft Foundry: the questions test configurations, trade-offs and failure modes you only meet by building. Every learning path has a free, self-paced Microsoft lab site (the agents site is linked below, the rest are in the resources table). Make sure you personally touch:

• Deploying models with different deployment types and watching quota and rate-limit behavior
• Building an agent with tools (file search, Azure AI Search, OpenAPI, MCP) and a knowledge store
• Implementing RAG against an Azure AI Search index and connecting it to an agent
• Running evaluations for groundedness, relevance and safety, and reading the traces
• Creating an Azure Content Understanding analyzer for a document schema
• Calling speech to text, text to speech, translation and image generation (with mask-based edits) from the SDK

Official Lab Exercises: Develop AI Agents in Azure

 

Step 6: Benchmark Your Knowledge

Use CertiAce to benchmark your readiness module by module. The practice modules map to the exam domains as shown above, so use the module mapping and the topics behind your missed questions to identify the skills-measured areas to revisit. Microsoft also offers a free official practice assessment on AI Skills Navigator (sign in required). Aim for consistent performance across every module, not one lucky high score; if a topic is unstable, go back to learning plus hands-on practice.

CertiAce AI-103 Exam Practice

Official Practice Assessment on AI Skills Navigator

 

Step 7: Take the Exam

The day before, review only your weak topics (no brand-new material); deployment types, agent tool choices and evaluator names are the usual last-minute wins. On exam day, read each question for what it is truly asking, eliminate wrong options first, and watch for wording that implies constraints such as data residency, cost, least privilege, or minimal development effort.

Worth knowing before you sit down:

• Microsoft Learn is available during the exam in a split screen. Use it to check details you genuinely need: the timer keeps running, and the exam is deliberately too long to look up every answer. Browsing is limited to Microsoft Learn itself (no Q&A, practice assessments, or profile), and personal notes and other websites are unavailable.
• You can take unscheduled breaks, but the clock keeps running and you cannot return to questions you already saw.
• Case studies lock when you leave them: finish each one before moving on, because you cannot revisit its questions afterwards.
• Labs may appear: AI-103's 120-minute allotment is the one Microsoft uses for role-based exams that may contain labs. The introduction screen tells you whether yours does; if it does, budget extra time for them.
• Try the exam sandbox beforehand so the question formats and interface hold no surprises: Exam Sandbox

 

---

Additional Learning Resources