AI-102 study guide · 2026

The 6 AI-102 domains, explained

6 domains and 17 official skill groups. Plan and manage an Azure AI solution carries the most weight at 20-25%. Each guide explains what the domain covers, how its questions are framed and where candidates lose marks.

Start with Domain 1 See all domains
Questions
Not published by Microsoft (typically 40–60)
Time
100 minutes
Passing score
700 or greater (scaled score, 1–1000)
Outline
AI-102 study guide, skills measured as of December 23, 2025 (final version; exam retired June 30, 2026)

Exam blueprint

Bar widths follow the official weights, scaled to fit the bar. Weight ranges are drawn at their midpoint.

  1. 1 20-25%
  2. 2 15-20%
  3. 3 5-10%
  4. 4 10-15%
  5. 5 15-20%
  6. 6 15-20%
  1. 1 Plan and manage an Azure AI solution
  2. 2 Implement generative AI solutions
  3. 3 Implement an agentic solution
  4. 4 Implement computer vision solutions
  5. 5 Implement natural language processing solutions
  6. 6 Implement knowledge mining and information extraction solutions

All 6 domains

17 official skill groups 80–110 hrs suggested study

  1. Domain 01 20-25%

    Plan and manage an Azure AI solution

    The heaviest AI-102 skill area: choosing the right Foundry service, deploying and securing it, controlling cost, and applying content safety. Almost every scenario elsewhere on the exam assumes you can make these platform decisions correctly.

    Inside this domain

    • Select the appropriate Microsoft Foundry Services
    • Plan, create and deploy a Microsoft Foundry Service
    • Manage, monitor, and secure a Microsoft Foundry Service
    • Implement AI solutions responsibly

    Watch for Key Vault as the zero-secret answer

    Skill groups
    4
    Questions est.
    8–15
    Study time est.
    20–25 hrs
  2. Domain 02 15-20%

    Implement generative AI solutions

    Building, grounding, evaluating and tuning generative AI apps in Microsoft Foundry and Azure OpenAI. The exam tests whether you pick the right technique, parameter or deployment for the problem described, not whether you can write clever prompts.

    Inside this domain

    • Build generative AI solutions with Microsoft Foundry
    • Use Azure OpenAI in Foundry Models to generate content
    • Optimize and operationalize a generative AI solution

    Watch for Fine-tune for fresh knowledge

    Skill groups
    3
    Questions est.
    6–12
    Study time est.
    15–20 hrs
  3. Domain 03 5-10%

    Implement an agentic solution

    The newest and smallest AI-102 skill area: building agents that reason, call tools and work together. Questions test when an agent is the right design, which tool fits a task, and how to orchestrate and control several agents safely.

    Inside this domain

    • Create custom agents

    Watch for Agent for everything

    Skill groups
    1
    Questions est.
    2–6
    Study time est.
    5–10 hrs
  4. Domain 04 10-15%

    Implement computer vision solutions

    Analyzing images, training custom image models and extracting insights from video. Questions turn on choosing the right feature or project type and reading the response or metrics correctly, so precise vocabulary matters more than deep math.

    Inside this domain

    • Analyze images
    • Implement custom vision models
    • Analyze videos

    Watch for Classification for counting

    Skill groups
    3
    Questions est.
    4–9
    Study time est.
    10–15 hrs
  5. Domain 05 15-20%

    Implement natural language processing solutions

    Text analytics, translation, speech and custom language models in one wide skill area. Most questions hinge on picking the right Language, Translator or Speech feature and knowing its exact setup steps, outputs and limits.

    Inside this domain

    • Analyze and translate text
    • Process and translate speech
    • Implement custom language models

    Watch for FAQ bot for actions

    Skill groups
    3
    Questions est.
    6–12
    Study time est.
    15–20 hrs
  6. Domain 06 15-20%

    Implement knowledge mining and information extraction solutions

    Turning files into searchable, structured data with Azure AI Search, Document Intelligence and Content Understanding. Expect pipeline design, index and query details, and the choice between prebuilt, custom and composed extraction models.

    Inside this domain

    • Implement an Azure AI Search solution
    • Implement an Azure Document Intelligence in Foundry Tools solution
    • Extract information with Azure Content Understanding in Foundry Tools

    Watch for Searchable is enough

    Skill groups
    3
    Questions est.
    6–12
    Study time est.
    15–20 hrs

Study the heaviest domains first

Marks follow weight. This order puts the most marks behind you first, across about 80–110 hrs of study.

  1. 1 Plan and manage an Azure AI solution 20-25% · 20–25 hrs
  2. 2 Implement generative AI solutions 15-20% · 15–20 hrs
  3. 5 Implement natural language processing solutions 15-20% · 15–20 hrs
  4. 6 Implement knowledge mining and information extraction solutions 15-20% · 15–20 hrs
  5. 4 Implement computer vision solutions 10-15% · 10–15 hrs
  6. 3 Implement an agentic solution 5-10% · 5–10 hrs

Try one

Practice questionDomain 1: Plan and manage an Azure AI solution

A company's web app calls Azure AI Language. Security policy says no keys or passwords may be stored in code, configuration or any vault, and the app must have only the access needed to call the API. What should you configure?

One trap from every domain

The answer that looks right, and the clue that gives it away.

  • Domain 1

    Key Vault as the zero-secret answer

    Clue Phrases such as 'without storing any secrets' or 'without managing credentials' point to a managed identity with an RBAC role.

  • Domain 2

    Fine-tune for fresh knowledge

    Clue Content that changes often, or a requirement to cite sources.

  • Domain 3

    Agent for everything

    Clue A single input, a single output and no action to take.

  • Domain 4

    Classification for counting

    Clue Words such as where, how many, locate or mark.

  • Domain 5

    FAQ bot for actions

    Clue Verbs like book, transfer, cancel or order with parameters.

  • Domain 6

    Searchable is enough

    Clue Requirements to filter, facet or sort on a field.

Other Artificial Intelligence & Machine Learning certifications

Study guides for exams that overlap with AI-102.

Practise what you read

AI-102 domains: common questions

How many domains are in the AI-102 exam?

AI-102 has 6 domains with 17 official skill groups between them: Plan and manage an Azure AI solution, Implement generative AI solutions, Implement an agentic solution, Implement computer vision solutions, Implement natural language processing solutions and Implement knowledge mining and information extraction solutions.

Which AI-102 domain carries the most weight?

Plan and manage an Azure AI solution is the heaviest at 20-25%, followed by Implement generative AI solutions (15-20%). The lightest is Implement an agentic solution (5-10%). Weights come from Microsoft's official outline.

What is the AI-102 exam format?

Questions: Not published by Microsoft (typically 40–60). Time allowed: 100 minutes. Passing score: 700 or greater (scaled score, 1–1000).

How long should I study for AI-102?

Our planning estimate is 80–110 hrs across all domains, split by weight. It is an editorial estimate, not an official figure, so adjust it to your experience and to how you score on practice questions.

Where should I start?

Start with Plan and manage an Azure AI solution, because it carries the most marks, then work down by weight. Use the checklist on each domain page to confirm you can do what the domain asks before moving on.

Which outline is this guide based on?

This guide follows AI-102 study guide, skills measured as of December 23, 2025 (final version; exam retired June 30, 2026), published by Microsoft. It was last reviewed in October 2026.

Domain names, weights and skill groups come from Microsoft's official outline (AI-102 study guide, skills measured as of December 23, 2025 (final version; exam retired June 30, 2026)). Question counts and study hours are editorial planning estimates, not official figures. Last reviewed October 2026.