AI-102 Retired: What Replaces the Azure AI Engineer Certification

AI-102 was the exam for the Microsoft Certified: Azure AI Engineer Associate. Microsoft retired it on June 30, 2026, together with the certification and its renewal assessment. If you’re searching for it, the exam to take now is AI-103, for the Azure AI Apps and Agents Developer Associate. It’s a 120-minute proctored exam aimed at the same job role.

Is AI-102 Still Available?

No. AI-102 is retired, and you can’t register for it or renew it. Plenty of study guides and course listings still mention it as if it were open, so check the date on anything you study from.

Microsoft’s own page for the certification carries a warning that reads: “This certification and the renewal assessment are retired.” The page is still online, but the exam isn’t.

If you already earned the credential, you haven’t lost it. Microsoft’s retirement page says a certification you earned or renewed before it retired “will remain on your transcript in the Active Certifications section until it expires.”

What Should You Take Instead of AI-102?

Start with AI-103. Microsoft’s current certification for an Azure AI engineer is the Azure AI Apps and Agents Developer Associate, and its page describes the candidate as “an Azure AI engineer who builds, manages, and deploys agents and AI solutions that take advantage of Microsoft Foundry.”

Microsoft doesn’t call it a one-for-one replacement on the pages we read, so treat it as the successor in role, not in name. The role tag is the same, AI Engineer, and the skills list covers much of the same ground. It’s an intermediate-level credential, and you have 120 minutes for the exam, up from 100 for AI-102.

A second option fits a narrower path. The Azure AI Cloud Developer Associate, exam AI-200, focuses on back-end services and components. Its page says you should be proficient in “Vector databases,” “Python programming” and “Implementing containerized applications on Azure.”

How Do AI-102 and AI-103 Compare?

The two exams overlap, but the new one leans harder on agents and drops the older framing around Azure AI services. Here’s how the skills lists line up on Microsoft’s pages.

Two things stand out. Agents are now part of the generative AI section instead of standing alone, and the platform name is Microsoft Foundry. The AI-103 page names Python as the language you should know. The old AI-102 page named Python and C#.

If you studied for AI-102 before it retired, most of that work carries over. Your gap is likely agents and Foundry, so start the AI-103 study guide there.

What Does the Azure AI Engineer Certification Cover?

Both versions test whether you can build working AI features on Azure, not whether you can design a model from scratch. That’s a real difference from a machine learning role, and it matters for how you plan your education.

On the current exam, Microsoft lists five areas:

• Planning and managing an Azure AI solution.
• Implementing generative AI and agentic solutions.
• Implementing computer vision solutions.
• Implementing text analysis solutions.
• Implementing information extraction solutions.

In plain terms, you’d be calling prebuilt and hosted models through code. You might wire a language model into an app, read text out of scanned forms or build an agent that uses tools. Microsoft says you collaborate with “business stakeholders, solution architects, data scientists, DevOps engineers, and cloud security engineers.” Notice that data scientists are a separate job on that list.

How Do You Prepare for AI-103?

Plan on hands-on practice more than reading. Microsoft expects Python experience and familiarity with general AI, generative AI and Azure services, so a first-timer should build those basics before booking the exam.

A reasonable order looks like this:

1. Get comfortable with Python. You should be able to read an SDK example, call a REST API and handle the response without copying blindly.

2. Learn the concepts first. The Azure AI Fundamentals exam, AI-901, covers the vocabulary. It’s optional, but it makes the associate material easier.

3. Open the study guide. Microsoft links the AI-103 study guide from the certification page. It lists what the exam measures.

4. Build something small. Make a chat feature, a document reader or a simple agent in Foundry. Breaking it teaches more than a video.

5. Take the practice assessment. It lives on AI Skills Navigator, and you need to sign in to use it.

Microsoft offers an exam sandbox so you can see the question types before test day. Try it once so the interface isn’t a surprise.

Retakes aren’t instant. Microsoft says you can retake after 24 hours following the first attempt, and the wait grows for later attempts.

Should You Pick AI-103 or AI-200?

They aim at different parts of the same job, so the choice comes down to what you like building.

AI-103 is the closer match to the old AI-102. It’s about using Foundry to build agents, language features, vision features and document reading. If you enjoy the “what should this AI feature do” side, start here.

AI-200 is for the plumbing around those features. Microsoft’s skills list for it includes developing containerized solutions, working with Azure data management services, connecting to and consuming Azure services, and securing, monitoring and troubleshooting solutions. If you like databases, containers and keeping production systems healthy, it fits better.

Neither one asks for a degree, and you don’t need to take them in order. Some people take both over a year, which gives them a broader story for an interview.

What If You Already Started Studying for AI-102?

Don’t throw the work away, but don’t finish the old plan either. Hands-on practice with Azure AI services, REST calls and Python SDKs still applies, because those skills show up on the new exam too.

Check the study material against the AI-103 skills list. Look for anything that treats agents as an add-on, or that names tools Microsoft has renamed. Treat those parts as background and replace them with the AI-103 study guide.

If you’d booked an AI-102 appointment that’s now canceled, contact Microsoft Credentials support through the link on the certification page. We can’t speak to how they handle individual cases.

Does a Certification Replace an AI Degree?

No, and it’s worth being plain about that. AI-103 shows you can use Microsoft’s current tools. A degree shows you’ve studied the math, programming and theory underneath them, which is what lets you move between tools as they change.

AI-102’s retirement is a good example. Within a few years, the exam you studied for was gone, and the platform got a new name. A bachelor’s or master’s keeps its value through that kind of change.

No degree is required for AI-103, so you can take it without one. But many AI engineer postings list a degree as a minimum, and a certificate won’t get a resume past that filter on its own. Our look at whether an AI degree is worth it goes through the trade-offs.

The best use of a Microsoft credential is on top of a degree or alongside real work experience. It tells an employer you already know their cloud.

Which Online AI Degree Lists Fit an Azure AI Engineer Path?

Three of our ranking lists match this career best. Pick based on where you’re starting.

• Best Online AI Engineering Degrees is the closest fit. It covers programs that teach you to build and deploy AI systems, which is the work AI-103 tests.
• Best Online Applied AI Degrees suits you if you want to put AI to work in a business setting and care less about research.
• Best Online Machine Learning Degrees goes deeper into models and math. Choose it if you want the option to move toward model work later.

If you’re still deciding between a computer science background and an AI-specific one, read how an AI degree compares to a computer science degree. And if you’re working your way up from a different field, our guide to becoming an AI engineer lays out the math and Python you’ll want first.

A degree takes years, and a certification takes weeks. That’s why many people do both: the degree for depth, the exam for a specific employer’s stack.

What Jobs Does an Azure AI Certification Help With?

It helps most with AI engineering roles on Azure, and with software jobs that are adding AI features. Our AI engineer career guide covers the daily work, pay and the degree employers expect.

Related paths are worth a look:

• Machine learning engineer roles lean toward training and deploying models.
• Software engineer roles often add AI work as part of a larger product.
• Data scientist roles focus on analysis and modeling, and the old DP-100 path fits there. Our DP-100 page covers that retirement.

No exam promises a job. Check a few current postings in your area and see whether they name Azure, AI-103 or any Microsoft credential. If they do, the exam is worth your time. If they don’t, spend that money on a portfolio project.

What Are the Downsides of Chasing a Microsoft AI Exam?

Three things are worth knowing before you pay.

Exams get retired. AI-102, AI-900 and DP-100 all left within about a month of each other in 2026. Whatever you study can change on short notice, so don’t build a long plan around one code.

It’s tied to one cloud. Employers on AWS or Google Cloud care less about a Microsoft credential.

The study guides lag the products. Microsoft’s own page for the retired data scientist exam says it was in the middle of renaming Azure AI Foundry to Microsoft Foundry. Expect names to shift, and check dates on anything you read.

None of these make AI-103 a bad choice. They’re reasons to pair it with a degree and a few projects.

How We Checked These Facts

We read Microsoft Learn’s pages for the Azure AI Engineer Associate, the Azure AI Apps and Agents Developer Associate and the Azure AI Cloud Developer Associate on October 1, 2026. The retirement date comes from Microsoft’s retired certification exams page, which lists “AI-102 Microsoft Certified: Azure AI Engineer Associate June 30, 2026.”

Microsoft’s pages we read didn’t show an exam price, so confirm it with Pearson VUE before you register. Prices vary by country. Microsoft can change names, codes and dates, so check the links in our sources before you plan around any of this.