About the Azure AI Engineer Associate (AI-102) Exam

Quick Answer: Exam AI-102 has about 40-60 questions, a 100-minute time limit, and needs a scaled score of 700/1,000 to pass. It measures six skill areas led by Plan and Manage an Azure AI Solution (20-25%). It costs $165 USD in the US, has no required prerequisite certification, and retires June 30, 2026 in favor of AI-103.

The Microsoft Azure AI Engineer Associate (AI-102) certification, officially titled Designing and Implementing a Microsoft Azure AI Solution, validates that you can plan, build, secure, and operate AI solutions on Azure. You are expected to combine classic Azure AI services — Vision, Language, Speech, AI Search, and Document Intelligence — with generative AI through Azure OpenAI and the new agent tooling, using Python or C#, REST APIs, and SDKs.

A Moving Target: The December 2025 Refresh

Microsoft last revised the Skills Measured on December 23, 2025. The biggest structural change is the addition of an Implement an Agentic Solution domain and the rebranding of the developer surface to Microsoft Foundry (formerly Azure AI Foundry). If a study resource still lists a separate "Content Moderation" domain or claims NLP is the largest area, it is describing a retired blueprint. Content safety now lives inside Plan and Manage as a responsible-AI sub-objective.

Exam Format at a Glance

Current Domain Breakdown (as of Dec 23, 2025)

On the Exam: The single heaviest area is now Plan and Manage (20-25%), which folds in service selection, deployment, cost/identity management, and responsible AI (content safety, content filters, blocklists, prompt shields). Generative AI, NLP, and Knowledge Mining each sit at 15-20%. The slim Agentic (5-10%) band is new but high-yield because few candidates study it. Computer Vision dropped to 10-15%.

Question Types Explained

• Single-select multiple choice: one correct answer among four. Typically "which service/SDK class do you use?"
• Multi-select: choose all that apply (often two or three of five-plus options). Partial credit is not given — every correct box must be checked.
• Drag-and-drop / build-list: order API calls or pipeline steps (for example, create resource -> create index -> attach skillset -> run indexer).
• Hot area: click the correct option in a configuration screenshot or code block.
• Case studies: a 2-4 page scenario followed by several linked questions. Once you leave a case study you cannot return, so answer every linked item before moving on.

Who Should Take This Exam?

• AI engineers shipping Azure AI features to production.
• Software developers embedding Vision, Language, Speech, or OpenAI into apps.
• Solution architects designing AI-powered systems on Azure.
• Data scientists and cloud engineers moving from modeling toward deployment and integration.

Prerequisites and Recommended Skills

There is no mandatory prerequisite exam, but Microsoft's audience profile assumes you can:

• Write Python or C# and read both in code-completion items.
• Construct REST requests with correct endpoint, Ocp-Apim-Subscription-Key or Entra token, and JSON body.
• Provision resources via the Azure portal, Azure CLI, and ARM/Bicep.
• Apply responsible AI principles (fairness, content safety, transparency).

How to Register

1. Open the AI-102 page on Microsoft Learn and select Schedule exam.
2. Sign in with the Microsoft account tied to your Learn profile.
3. Choose Pearson VUE, then a test center or online proctoring.
4. Select a date and pay $165 USD (or local equivalent).

Vouchers and discounts: Microsoft Enterprise Skills Initiative (employer-funded free vouchers), Cloud Skills Challenge promo vouchers, verified-student pricing, and employer reimbursement.

Scoring and Renewal

Scores are scaled 1-1,000; 700 passes. The scaled score is not a raw percentage, so do not assume 70% correct equals a pass. You see pass/fail immediately on screen and a domain-level bar report appears in your Learn profile shortly after. The associate certification expires after one year; renew free within the six-month window before expiry by passing an unproctored online assessment on Microsoft Learn — no retake of AI-102 required.

Important: AI-102 retires June 30, 2026 (11:59 PM Central). Its successor, AI-103 — Developing AI Apps and Agents on Azure — leads to the Azure AI Apps and Agents Developer Associate certification and leans further into generative AI, RAG, and agent orchestration. If you have studied the current AI-102 blueprint, sit the exam before the deadline; your earned credential stays renewable afterward.