AI-102 Retired Exam Practice & Next Steps — Microsoft Azure AI Engineer
Review the final AI-102 outline on iPhone or iPad, preserve the Azure AI skills you learned, and compare them with the current AI-103 requirements. Core study stays on-device and works offline; optional sync uses your private iCloud account.
The exam
What is the AI-102 exam?
AI-102 was the exam for Microsoft Certified: Azure AI Engineer Associate. Microsoft retired both the exam and certification on 30 June 2026, so AI-102 can no longer be scheduled. AI-103 is the current role-based route for developing AI apps and agents on Azure. This page and its question bank remain available as reference material for the final AI-102 outline.
AI-102 is hands-on and code-aware. It validates that you can plan and manage Azure AI solutions (resource provisioning, RBAC, monitoring, deployment), implement generative AI solutions using Azure OpenAI (prompt engineering, RAG, content filters), implement agentic solutions (Azure AI Foundry agents, tool calling), build computer-vision solutions (image analysis, OCR, custom models with AI Vision), implement NLP solutions (Azure AI Language for entity recognition, sentiment, summarisation; Azure AI Speech for transcription and synthesis), and build knowledge-mining and information-extraction solutions (Azure AI Search, Document Intelligence). Expect scenario questions with REST API calls, Python SDK snippets, and config JSON.
Microsoft last updated the AI-102 skills outline on 23 December 2025, adding the "Implement an agentic solution" domain and rebalancing weights toward generative AI. Every question in Azure Mastery's AI-102 bank is mapped to that final outline. Read the archived official outline at learn.microsoft.com.
Questions40–60 multiple choice
Duration100 minutes (120 min seat)
Pass score700 / 1000
CostNot schedulable
CredentialAzure AI Engineer Associate retired
StatusRetired 30 June 2026
Skills measured · April 2026
AI-102 exam objectives
Six domains, with weights set by Microsoft's December 2025 update. Every domain summary below is paraphrased from the official skills outline; bullet-level objectives in Azure Mastery are tagged so you always know which domain you're being tested on and where your weak spots cluster.
AI & agent skill mapDataModelsAgentsResponsible AI
Plan and manage an Azure AI solution20–25%
Explore key topics
Tied for the largest domain. Plan and provision Azure AI services resources — multi-service vs single-service accounts, RBAC roles, customer-managed keys, regional and SKU choices. Manage authentication (keys, RBAC, managed identities), networking (VNet integration, private endpoints), and content moderation. Plus Responsible AI considerations, monitoring with Azure Monitor and Application Insights, and deployment strategies (containers, ARM/Bicep). Around 8–15 questions per sitting.
Implement generative AI solutions15–20%
Explore key topics
Azure OpenAI end-to-end. Provision Azure OpenAI resources, deploy models, choose the right model (GPT family, embeddings, image, audio), prompt engineering basics (system prompts, few-shot, chain-of-thought), retrieval-augmented generation (RAG) with Azure AI Search and embeddings, content filters and abuse monitoring, fine-tuning workflows, evaluating model outputs (groundedness, relevance, fluency). Around 6–12 questions.
Implement an agentic solution5–10%
Explore key topics
New domain in the December 2025 outline — smallest by weight but distinct. Cover Azure AI Foundry agents, tool calling and function definitions, agent grounding via knowledge sources, multi-agent orchestration patterns, agent observability and tracing. Around 2–6 questions.
Implement computer vision solutions10–15%
Explore key topics
Azure AI Vision for image analysis (tagging, captioning, OCR via Read API, smart cropping), Custom Vision for image classification and object detection (training, evaluation, deployment), Face service (detection, identification, verification — note Limited Access policy), and Video Analyzer for Media. Around 4–9 questions.
Implement natural language processing solutions15–20%
Explore key topics
Azure AI Language for entity recognition, sentiment analysis, key phrase extraction, language detection, summarisation, custom text classification, custom named-entity recognition, conversational language understanding (CLU), and question answering. Azure AI Speech for speech-to-text, text-to-speech, speech translation, custom speech, and pronunciation assessment. Azure AI Translator for text and document translation. Around 6–12 questions.
Implement knowledge mining and information extraction solutions15–20%
Explore key topics
Tied for the largest domain in scope. Azure AI Search end-to-end — indexers, indexes, skillsets, semantic search, vector search, hybrid search. Document Intelligence (formerly Form Recognizer) — prebuilt models (invoice, receipt, ID, business card, layout) and custom models (template, neural). Around 6–12 questions.
Designed for AI-102
How Azure Mastery preserves AI-102 coverage
Azure Mastery ships with 334 AI-102 practice questions, every one written specifically against the current (December 2025) skills outline. Each question carries a domain tag mapped to the official six domains (Plan/manage, generative AI, agentic, vision, NLP, knowledge mining), so you always know which area you're being tested on and where your weak spots are clustered. REST API calls, Python SDK snippets, Azure OpenAI prompt configurations, and Foundry project setups appear throughout — matching the format of the live exam.
The on-device Exam IQ engine can still estimate your performance against the final AI-102 bank for reference. After roughly 30 questions it identifies the specific retired-outline topics that need review, helping you retain Azure AI knowledge before moving to AI-103.
The adaptive study plan rebuilds itself from your answer history. Get a scenario question wrong? The engine surfaces another question in the same domain in your next session. Master a topic across three sessions and it backs off, prioritising the next-highest-leverage gap. The plan optimises for the gap between where you are and the 700 pass score, not for blind volume.
Knowledge decay tracking matters more for AI-102 than for foundational exams — five domains is a lot to retain, and the topic you mastered three weeks into your study window is the topic you'll forget by exam day if you stop revising. Azure Mastery tracks every topic's decay curve and flags topics approaching expiry. The padlock icon on the Today screen is your "revisit before you forget" cue, and weak-spot drills automatically pull from decayed topics first.
Historical simulation mode can still run a 40–60-question set from the full 334-question bank with the former 100-minute timing. It is retained for reference and does not represent an exam that can still be booked.
Answer Coach turns each missed answer into a private, grounded lesson: the misconception, key distinction, and rule to remember. It always uses authored certification guidance; on supported devices, an optional on-device model may rewrite the note only when it passes grounding checks.
During your first week, Aura adapts the next step as you go. Every session ends with a concise recap of what changed, what to focus on, and the best follow-up.
Everything essential runs on-device. Your answer history, readiness gauge, and coaching stay private. Optional sync uses your private iCloud account; there is no Azure Mastery account, tracking, or external processing server.
6-week study plan
Suggested AI-102 study plan
The retained six-week plan maps onto the six domains in AI-102's final outline and remains useful for historical reference. New candidates should use the current AI-103 study plan and question bank for the active Azure AI apps and agents route.
Plan, manage, and generative AI
Week 1: Plan and manage Azure AI solutions — resource provisioning (multi-service vs single-service), RBAC, customer-managed keys, networking, content moderation, Responsible AI considerations, monitoring, deployment via containers and ARM/Bicep.
Week 2: Implement generative AI solutions — Azure OpenAI provisioning, model deployment, prompt engineering (system prompts, few-shot, chain-of-thought), RAG with Azure AI Search and embeddings, content filters, fine-tuning, evaluation metrics.
Vision and NLP
Week 3: Computer vision — Azure AI Vision (analysis, OCR via Read API), Custom Vision (classification and object detection), Face service (Limited Access policy), Video Analyzer for Media.
Week 4: NLP — Azure AI Language (entity recognition, sentiment, summarisation, CLU, question answering), Azure AI Speech (STT, TTS, translation, custom speech), Azure AI Translator.
Knowledge mining, agentic, sharpen, simulate
Week 5: Knowledge mining — Azure AI Search (indexers, indexes, skillsets, semantic and vector search), Document Intelligence (prebuilt and custom models). Plus agentic — Azure AI Foundry agents, tool calling, multi-agent orchestration.
Week 6: Use Focus Weak Spots and historical simulator runs for reference, then move to the current AI-103 bank before scheduling an exam.
Inside the app
Every Microsoft question type, on iPhone
AI-102's question bank uses the same formats Microsoft puts on the live exam — not just multiple choice. Each visualisation below is a faithful mock of how the type renders inside Azure Mastery on iPhone and iPad. Exam-simulator mode runs all of them at full 100-minute length with no flag-and-review jumps, mirroring Pearson VUE.
What is the name of the object that represents one execution of an agent against a thread?
A run
A repository
A workbook
An indexer
Multiple choice
A real AI-102 question-bank example with one correct answer. The app explains every option after you answer.
Exam-specific sample
Which THREE message roles are used in the Azure OpenAI chat completions API? (Choose three.)
System
User
Assistant
Admin
Moderator
Multi-select
A real AI-102 multi-select item. Every required selection must be correct to earn the mark.
All-or-nothing
Arrange the steps to create and invoke an agent using the Microsoft Foundry Agent Service SDK in the correct order.
⋮⋮1Create an agent with instructions and tools
⋮⋮2Create a thread for the conversation
⋮⋮3Add a user message to the thread
⋮⋮4Create a run on the thread to invoke the agent
⋮⋮5Retrieve the agent's response from the completed run
Drag-and-drop
A real AI-102 interactive-format prompt, rendered for touch on iPhone and iPad.
Interactive item
You are reviewing the architecture diagram for an Azure AI solution. Identify the correct placement of services. Select the service that provides network isolation for…
Hotspot
A real AI-102 prompt that tests recognition inside a visual or contextual interface.
Tap target
Woodgrove Bank AI Assistant Woodgrove Bank has 500 financial advisors across 50 branches. They currently rely on a keyword-based search tool to find information in product documentation, which is…
1Referring to the Woodgrove Bank case study: Which architecture pattern should be…
2Referring to the Woodgrove Bank case study: How should the assistant look up…
3Referring to the Woodgrove Bank case study: Which TWO configurations ensure the…
Case studies
A real AI-102 case-study scenario with linked questions that share the same requirements and environment.
Multi-question
✕Your answer: A repository
✨ Answer Coach:A repository is a version-controlled store of source code and files in services like Azure Repos or GitHub; it holds artifacts at rest and never represents a single agent invocation against a thread.
— grounded in authored certification guidance
Answer Coach
Answer Coach uses the bank's authored rationale to explain the misconception, key distinction, and rule to remember. On supported devices, an optional on-device model may rewrite the note only when it passes grounding checks.
App exclusive
Frequently asked
AI-102 FAQs
Can I still take the AI-102 exam?
No. Microsoft retired AI-102 and the Azure AI Engineer Associate certification on 30 June 2026. The exam can no longer be scheduled.
What replaced AI-102?
AI-103 is the current role-based route for developing AI apps and agents on Azure. It leads to the Azure AI Apps and Agents Developer Associate credential and assesses Python, Microsoft Foundry, generative and agentic solutions, vision, text analysis, and information extraction.
What happens if I already earned Azure AI Engineer Associate?
A certification earned before retirement remains on your Microsoft Learn transcript. AI-102 and its renewal assessment are retired, so the exam cannot be retaken or newly earned.
Are AI-102 practice questions still useful?
Yes, as reference material for Azure AI services, generative AI, vision, language, search, and document intelligence. New candidates should study the current AI-103 outline and use its exam-specific bank.
What is the current path after AI-102 retirement?
AI-901 is the current Azure AI Fundamentals exam, and AI-103 is the current role-based route for developing AI apps and agents on Azure. AI-102 content remains useful as historical reference, but new candidates should use the AI-901 and AI-103 outlines rather than planning an AI-102 certification path.
What replaced the retired AI-102 and DP-100 routes?
Microsoft replaced the retired AI-102 route with AI-103 for Azure AI apps and agents, and identifies AI-300 as the replacement for the retired DP-100 data-science route. AI-103 focuses on application and agent development; AI-300 focuses on operationalizing machine learning and generative AI systems.
Where AI-102 fits
AI-102 retirement and current certification paths
AI-102 and the Azure AI Engineer Associate certification retired on 30 June 2026. The diagrams below separate the historical AI-102 route from Microsoft's current AI-901 and AI-103 pathway.
AI-102 is retained for historical reference. New candidates should use AI-901 for Azure AI fundamentals and choose AI-103 for AI apps and agents or AI-300 for machine-learning and generative-AI operations.
Review AI-102 or move to AI-103
Keep 334 final-outline AI-102 questions for reference, or use Azure Mastery's current AI-103 bank and adaptive study plan for the active Azure AI apps and agents route.
