- "AI-102" was the exam code for Designing and Implementing a Microsoft Azure AI Solution, tied to Azure AI Engineer Associate.
- Both the exam and credential retire June 30, 2026 at 11:59 PM Central, with no renewal path afterward.
- The live successor is AI-103, Developing AI Apps and Agents on Azure, awarding Azure AI Apps and Agents Developer Associate.
- Implement generative AI and agentic solutions is the heaviest domain at 30-35% of exam content.
What "AI-102" Actually Stands For
"AI-102" is Microsoft's internal exam code, and the exam it identifies is titled Designing and Implementing a Microsoft Azure AI Solution. Passing it earned the credential Microsoft Certified: Azure AI Engineer Associate. The code itself doesn't spell out an acronym in the traditional sense - it's simply the numbering Microsoft used to track this particular certification track within its "AI" exam family, distinguishing it from beginner-level AI fundamentals exams and from role-based exams in other product areas.
If you're searching for what the letters and digits mean today, the honest answer has a second layer: the AI-102 exam and its associated credential are being retired. Anyone researching this certification in 2026 needs to understand both what AI-102 historically covered and what has replaced it, because the practical, career-relevant answer has shifted. We break down the naming and the transition in more depth in AI-102 Meaning and What Is AI-102?, but this article focuses specifically on what the term stands for and why that definition is changing.
The Exam Behind the Name
Structurally, AI-102 was a Microsoft-owned, Pearson VUE-delivered assessment, available at physical test centers or via online proctoring. Candidates paid $165 USD at standard US pricing, with regional pricing varying by the country or region where the exam was proctored. The exam ran 120 minutes, was proctored, and could include interactive question components rather than only static multiple choice. Scoring used Microsoft's familiar 100-1000 scale, with 700 required to pass - Microsoft never published a fixed item count or an official pass rate for this exam.
There were no formal prerequisites listed for registration, but the intended audience profile assumed real app development experience, familiarity with Python, and working exposure to general AI, generative AI concepts, and core Azure services. For a full breakdown of who was expected to sit this exam and what background actually helped, see AI-102 Requirements 2026.
Key Takeaway
The exam name told you the scope: "Designing and Implementing" meant candidates needed hands-on build experience, not just conceptual familiarity with Azure AI services.
Why AI-102 Is Retiring and What Replaces It
This is the status change that matters most right now. Exam AI-102 and the Azure AI Engineer Associate credential - along with the associated renewal assessment - both retire on June 30, 2026 at 11:59 PM Central. After that date, the exam can no longer be scheduled, the credential can no longer be newly earned, and there is no renewal path even for people who hold it. If you certified before the retirement date, the credential stays visible in the Active Certifications section of your Microsoft transcript until its printed expiration date - it just can't be renewed once that date passes.
The credential's replacement is already live: Exam AI-103, Developing AI Apps and Agents on Azure, which awards Microsoft Certified: Azure AI Apps and Agents Developer Associate. At the training level, the four-day AI-103T00-A course has replaced the older AI-102T00 course. Practically speaking, if you're planning your certification path for 2026 and beyond, AI-103 is the exam you should be preparing for - the "AI-102" name now mostly matters for search context, migration questions, and understanding what the older credential covered. Our AI-102 Certification overview and Is the AI-102 Certification Worth It? analysis both go deeper into how this transition affects existing holders and new candidates.
| Attribute | AI-102 (Retiring) | AI-103 (Current) |
|---|---|---|
| Exam Title | Designing and Implementing a Microsoft Azure AI Solution | Developing AI Apps and Agents on Azure |
| Credential | Azure AI Engineer Associate | Azure AI Apps and Agents Developer Associate |
| Status | Retires June 30, 2026, 11:59 PM Central | Live, active exam |
| Training Course | AI-102T00 | AI-103T00-A (four days) |
| Renewal After Cutoff | None available | Standard annual free renewal via Microsoft Learn |
What the Five Domains Cover
Whether you're studying the retiring exam for historical context or the live AI-103 successor, the skills-measured structure is organized into five weighted domains. Understanding these weights is essential for allocating study time, and we cover each domain in full detail in the AI-102 Exam Domains 2026 guide.
Domain 1: Plan and manage an Azure AI solution (25-30%)
Covers resource provisioning, security, monitoring, and cost management decisions candidates must make before writing any application code.
- Selecting and configuring Azure AI resources appropriately
- Securing access with keys, managed identities, and role-based controls
- Monitoring usage and containerizing services where relevant
Domain 2: Implement generative AI and agentic solutions (30-35%)
The heaviest-weighted domain by a clear margin, and the one that most reflects how far this certification track has moved beyond classic "cognitive services" work into agentic application development.
- RAG (retrieval-augmented generation) implementation patterns
- Multi-agent orchestration and function calling with tool schemas
- Conversation memory, agent evaluation, error analysis, tracing, and token analytics
- Autonomous workflows built with approval controls
Domain 3: Implement computer vision solutions (10-15%)
Focuses on image analysis, object detection, and vision-based service integration within Azure AI.
Domain 4: Implement text analysis solutions (10-15%)
Covers natural language processing tasks such as sentiment, key phrase extraction, and language detection.
Domain 5: Implement information extraction solutions (10-15%)
Involves pulling structured data out of unstructured documents and content using Azure's extraction tooling.
Beyond the domain names, candidates are expected to have working knowledge of specific services and frameworks: Microsoft Foundry, Foundry Agent Service, Microsoft Agent Framework, Azure OpenAI in Foundry Models, Azure AI Search, Azure Content Understanding in Foundry Tools, Azure Document Intelligence in Foundry Tools, Azure Speech, and Azure Translator. Skills measured are current as of April 16, 2026, and while most questions target generally available features, commonly used preview features may still appear on the exam.
Who Actually Earns This Credential
This certification track is aimed squarely at developers rather than data scientists or IT administrators. The audience profile Microsoft describes assumes Python application development experience combined with familiarity in general AI, generative AI, and Azure services - not deep machine learning theory. In practice, that means the people pursuing this path are usually building chatbots, document processing pipelines, search-augmented applications, or agentic workflows on top of Azure AI services, rather than training models from scratch.
Given the domain weighting toward generative AI and agentic solutions, the strongest candidates already have some hands-on exposure to building with large language models, orchestrating multiple agents, and wiring up tool-calling patterns in real applications. If you're evaluating whether this fits your career direction, our AI-102 Jobs article looks at the kinds of roles that reference this credential family, and AI-102 Training covers the formal course options, including the AI-103T00-A successor course.
Registration, Fees, and Retake Rules
Registration happens through Pearson VUE, with the choice of an in-person test center or online proctoring. Standard US pricing sits at $165, though the exact fee varies depending on the country or region where you sit the exam - always confirm local pricing during scheduling rather than assuming the US figure applies. A full pricing breakdown, including how regional variance works, is available in AI-102 Certification Cost 2026.
If you don't pass on your first attempt, Microsoft enforces a mandatory 24-hour wait before you can retake the exam. Each subsequent failed attempt extends the required waiting period further, so a failed attempt isn't something to treat casually - plan your study cycle assuming one serious attempt rather than a rapid-fire retake strategy. Once earned, Microsoft associate certifications expire annually and are renewed for free through an unproctored online assessment on Microsoft Learn - though for AI-102 specifically, that renewal path disappears once the June 30, 2026 retirement takes effect, since the renewal assessment retires alongside the exam itself.
For a numbers-only reference on what score you actually need and how scaled scoring works, see AI-102 Passing Score 2026.
Mapping Study Time to the Domain Weights
Because Domain 2 (Implement generative AI and agentic solutions) carries the largest share of exam content at 30-35%, it deserves proportionally more study time than any other single domain - more than Domains 3, 4, or 5 combined in most study plans. A sensible allocation treats Domain 1 (25-30%) as a close second priority, since planning and managing the underlying Azure AI solution is foundational to everything built on top of it.
Domain 1 Foundations
- Provisioning and securing Azure AI resources
- Cost, monitoring, and identity configuration
Domain 2 Deep Dive
- RAG pipelines and multi-agent orchestration
- Function calling, tool schemas, conversation memory
- Agent evaluation, tracing, and token analytics
Vision, Text, and Extraction
- Computer vision solution patterns (Domain 3)
- Text analysis fundamentals (Domain 4)
- Information extraction workflows (Domain 5)
Practice and Review
- Timed practice questions under exam-length constraints
- Review weak domains flagged during practice
Running realistic timed questions against each domain, especially the generative AI and agentic solutions material, is one of the best ways to confirm you're ready before spending $165 on an attempt. You can work through domain-weighted practice questions on the main practice test site to gauge readiness before you schedule anything through Pearson VUE.
Key Takeaway
Don't split study time evenly across five domains - Domain 2 alone is worth roughly a third of the exam, so it should consume roughly a third (or more) of your prep hours.
Frequently Asked Questions
It's Microsoft's exam code for "Designing and Implementing a Microsoft Azure AI Solution," the exam that led to the Azure AI Engineer Associate credential. It isn't an acronym with individually meaningful letters - it's a Microsoft exam numbering convention.
It's being retired. The exam, its associated credential, and its renewal assessment all retire on June 30, 2026 at 11:59 PM Central. After that date it can't be earned or renewed, though people who already certified keep the credential on their transcript until its printed expiration.
Exam AI-103, Developing AI Apps and Agents on Azure, is the live successor. It awards Microsoft Certified: Azure AI Apps and Agents Developer Associate and is supported by the four-day AI-103T00-A training course.
Implement generative AI and agentic solutions, weighted 30-35%, is the largest domain and covers RAG, multi-agent orchestration, function calling, and agent evaluation - it deserves the most study hours of the five domains.
Standard US pricing is $165 USD, though it varies by proctoring region. The exam itself runs 120 minutes, is proctored, may include interactive question types, and requires a score of 700 out of 1000 to pass.