AI-102 Is Retiring: Your AI-103 Successor Path for 2026

Last Updated: June 2026

TL;DR: Microsoft retires exam AI-102 on 30 June 2026, closing the route to the Azure AI Engineer Associate certification. Its place is taken by AI-103, Developing AI Apps and Agents on Azure, which is still in beta and built around Microsoft Foundry. If you have already prepared for AI-102, sit it before the deadline [1] [5].

Microsoft is retiring exam AI-102, the route to the Azure AI Engineer Associate certification, on 30 June 2026. In its place sits a new exam, AI-103: Developing AI Apps and Agents on Azure, which leads to the Azure AI Apps and Agents Developer Associate credential. The official notice is short and unambiguous, and it changes the plan for anyone who is mid-preparation right now [1].

This guide explains exactly what the retirement means, what happens to a certification you already hold, how the new AI-103 exam differs from AI-102, and the honest decision you face if you are studying today. Every factual claim here is checked against Microsoft Learn and the official skills outlines for both exams. Where Microsoft has not published a detail, this guide says so rather than guessing. If you are still mapping out your wider path, you can browse Examinotion's Microsoft AI exam preparation courses alongside this article.

When does AI-102 retire?

AI-102 retires on 30 June 2026, at 11:59 PM Central Standard Time [2]. After that moment you can no longer book the exam, sit it, or earn the Azure AI Engineer Associate certification through it. The certification page carries a plain warning banner [1]:

This certification, related exam, and renewal assessments will retire on June 30, 2026. You will no longer be able to earn or renew this certification after this date.

The full exam title is Designing and Implementing a Microsoft Azure AI Solution, and it has been the standard validation of Azure AI engineering skills for years [2]. It runs for 100 minutes, requires a passing score of 700, and is offered in ten languages, which makes its withdrawal a meaningful change for a large, global pool of candidates [2]. You can confirm the exact skills measured on the official AI-102 skills outline while it remains published [2].

The key practical point is the hard cut-off. There is no documented grace period and no soft landing. The last day to sit AI-102 is 30 June 2026, and Microsoft's guidance for anyone preparing is to take the exam before that date [4].

What happens to your Azure AI Engineer certification?

If you already hold the Azure AI Engineer Associate certification, it does not vanish on 30 June 2026. Microsoft's credential retirement policy is explicit that a certification you earned before its retirement stays on your record until it would normally expire [3]:

If you earned or renewed a Certification before it was retired, it will remain on your transcript in the Active Certifications section until it expires.

In practice this means your certification continues to appear as active on your Microsoft Learn transcript, and you can keep referencing it on your CV and LinkedIn profile, until its usual expiry date passes [3]. What you lose is the ability to renew it. Microsoft confirms you can no longer earn or renew the credential after the retirement date, so once your current certification lapses, there is no route back through AI-102 [1].

For anyone in the middle of studying, Microsoft's instruction is direct [4]:

If you're currently preparing for an exam or assessment lab that will be retired, we strongly recommend that you take the exam or assessment lab before the retirement date. You won't be able to take these exams, assessment labs, or earn the associated certification or credential after that date.

If your renewal window happens to open before 30 June 2026, complete the free online renewal assessment now to push your expiry date out by another year. If you are weighing up renewal timing more generally, the Microsoft certification renewal guide walks through how the annual renewal assessments work.

Meet AI-103: Developing AI Apps and Agents on Azure

The exam taking AI-102's place in the Azure AI engineering space is AI-103, and it is a different proposition. AI-103 leads to the Microsoft Certified: Azure AI Apps and Agents Developer Associate credential, and Microsoft describes it in a single line [6]:

This certification validates your expertise in designing, developing, and deploying advanced Azure AI solutions using Python and Microsoft Foundry.

That sentence signals the shift. Where AI-102 covered a broad spread of Azure AI services, AI-103 is built around Microsoft Foundry and centres on generative AI and agents. It is an intermediate, developer-focused credential, and the audience profile lists experience developing applications in Python as the expected starting point [6]. The table below summarises the published facts.

Two caveats matter before you build a study plan around it. First, AI-103 is still in beta, so the exam content can shift as Microsoft analyses early results, and a Microsoft practice assessment has not yet been released [6]. Second, it is available in English only for now, which narrows access for candidates who would normally sit an exam in another language [6]. If beta exams are new to you, the Microsoft beta exam guide explains how scoring and timing differ from a generally available exam. You can read the full requirements on the official Azure AI Apps and Agents Developer Associate page [6].

AI-103 vs AI-102: what actually changed?

AI-103 is not a renamed AI-102. The two exams sit at the same level and cover the same broad technology area, but the measured skills have been substantially modernised. Microsoft has not formally labelled AI-103 a direct replacement for AI-102, and the two credentials are distinct, so treat AI-103 as the current Azure AI engineering exam rather than a like-for-like swap [5] [6]. The clearest way to see the change is to compare the two skills outlines side by side.

Figures are taken from the official AI-102 and AI-103 skills outlines [2] [5]. Six shifts stand out, and together they explain why preparation does not simply carry over.

Agentic AI becomes the centre of gravity. Agentic work was a 5-10% sliver of AI-102. In AI-103 it merges into a single dominant domain, Implement generative AI and agentic solutions, worth 30-35% of the mark [5]. Multi-agent orchestration, autonomous and semi-autonomous workflows, tool and function calling, and agent monitoring are now core, not peripheral.

Microsoft Foundry is the platform throughout. Every AI-103 domain assumes you are building on Microsoft Foundry, from selecting models and setting up deployments to monitoring drift, grounding quality and safety events [5]. AI-102 spread across a wider set of standalone Azure AI services, so the centre of gravity has moved.

Python is the stated language. AI-102 accepted either C# or Python across much of its content. AI-103's audience profile specifies Python as the development language [6]. Microsoft does not state that C# is excluded from every question, so do not assume it is absent, but Python is now the expectation you should prepare around.

Classical natural language processing is largely gone. AI-102 staples such as intent recognition, custom question answering and conversational language understanding do not appear in the AI-103 outline. They are replaced by LLM-driven text analysis, with large language models doing the extraction, summarisation and translation work [2] [5].

Computer vision expands into generation. AI-102 focused on classification, object detection and optical character recognition. AI-103 keeps a computer vision domain but adds image and video generation from prompts, image editing workflows, and multimodal understanding such as visual question answering and captioning [5].

Knowledge mining reframes as grounding. The deep Azure AI Search coverage in AI-102 becomes, in AI-103, an information extraction domain centred on retrieval-augmented generation: ingesting and indexing documents, images and audio, then using semantic, hybrid and vector search to ground agents and generative applications [5]. The technology overlaps, but the framing is now agent-first.

The takeaway is that an AI-102 study plan gives you a useful Azure AI foundation, but it does not cover the agentic, Foundry-centric and generative material that carries the most marks in AI-103.

Should you take AI-102 now or wait for AI-103?

The right move depends on where you are today, and there are three honest scenarios.

You have already prepared for AI-102. Sit it before 30 June 2026. You have invested in the right material for that exam, the certification stays on your transcript until it expires even after retirement, and there is no benefit to throwing away a near-complete preparation [1] [3]. Do not assume your AI-102 revision transfers cleanly to AI-103, because the highest-weighted AI-103 domains cover material AI-102 barely touches [5].

You are starting from scratch with weeks, not days. AI-103 is the path forward, but be realistic about the work. It is a beta exam with no practice assessment yet, English only, and built on Foundry, agents and generative AI evaluation that you will need to learn hands-on [5] [6]. Budget genuine study time and lab practice rather than treating it as an AI-102 refresh.

You are unsure which Azure AI level even fits you. Step back before committing to either exam. If you are still building Azure AI fundamentals, the foundational Azure AI Fundamentals certification is a more sensible first rung than an intermediate developer exam, and our guide to which Microsoft AI certification to take first maps the options by role.

There is one more honest caveat. Because AI-103 is in beta, Microsoft has not confirmed a general availability date, and details can still change [6]. If a stable, widely supported exam matters more to you than being first, that uncertainty is a legitimate reason to either sit AI-102 before it goes or wait for AI-103 to reach general availability.

The wider June 2026 Microsoft retirement wave

AI-102 is not retiring alone. Several certifications reach end of life on the same date, so if you hold more than one Microsoft credential it is worth checking your whole portfolio [4]. The exams retiring on 30 June 2026 include the following.

Microsoft frames these withdrawals as routine portfolio maintenance rather than a downgrade of the subject area [4]:

To keep our credentialing program relevant, we continually review our Applied Skills scenarios and Certifications to ensure they reflect the latest skills and Microsoft technologies and retire those that are no longer relevant.

The fundamentals tier shows the same pattern as the engineering tier. AI-900 retires on the same day, and the Azure AI Fundamentals certification page now lists its replacement, AI-901, alongside it with a beta label [7]. If you are on the fundamentals track, the AI-900 retirement explainer and the AI-900 versus AI-901 comparison cover that transition in detail. The pattern across the wave is consistent: older exams retire, and modernised, often beta, successors take their place.

How to prepare for the Azure AI engineer path now

The practical plan splits cleanly by your deadline. If you can still sit AI-102, focus your remaining time on its existing outline and book your slot well before 30 June 2026, because test centre and online proctoring availability tightens as a retirement date approaches [2] [4]. If you are aiming at AI-103, start building Foundry, agent and generative AI skills directly, using the official AI-103 study guide as your specification [5].

Whichever exam you target, a solid Azure AI foundation makes the engineering material far easier to absorb. Examinotion does not yet sell AI-102 or AI-103 preparation, but if you need to shore up the fundamentals first, our AI-901 practice resources cover the Azure AI Fundamentals skills that underpin the developer track. To see how the engineer path fits the wider catalogue, the Microsoft AI certification roadmap lays out how the fundamentals, associate and expert tiers connect.

Above all, be honest with yourself about the gap. AI-103 asks for hands-on competence with agents and generative AI on Foundry, and no amount of multiple-choice revision substitutes for actually building and evaluating an agent. Treat the skills outline as a checklist of things to build, not just topics to read.

Frequently Asked Questions

When does AI-102 retire?

AI-102 retires on 30 June 2026, at 11:59 PM Central Standard Time. After that you cannot book or sit the exam, or earn the Azure AI Engineer Associate certification through it. Microsoft documents no grace period, so the final day to take AI-102 is 30 June 2026 [1] [2].

What replaces AI-102?

The current Azure AI engineering exam is AI-103, Developing AI Apps and Agents on Azure, which leads to the Azure AI Apps and Agents Developer Associate certification. Microsoft has not formally called it a direct replacement, and it is a distinct, modernised credential built around Microsoft Foundry, generative AI and agents rather than a renamed AI-102 [5] [6].

Is my Azure AI Engineer certification still valid after AI-102 retires?

Yes. If you earned the certification before 30 June 2026, it stays in the Active Certifications section of your transcript until it expires. What you lose is renewal: Microsoft confirms you cannot renew the credential after the retirement date, so once it lapses there is no route back through AI-102 [1] [3].

Should I take AI-102 or AI-103?

If you have already prepared for AI-102, sit it before 30 June 2026, because your revision suits that exam and the certification remains valid until it expires. If you are starting fresh with time to spare, prepare for AI-103 instead, but treat it as new material rather than an AI-102 refresh [1] [5].

Is AI-103 available now?

AI-103 is available in beta as of June 2026, in English only, with no Microsoft practice assessment released yet. Microsoft has not confirmed a general availability date, and beta exam content can change as early results are analysed. Check the official certification page for the current status before booking [6].

How is AI-103 different from AI-102?

AI-103 centres on Microsoft Foundry, generative AI and agents. Agentic work alone carries 30-35% of the mark, classical natural language processing has largely gone, computer vision now includes image generation, and Python is the stated development language. An AI-102 study plan covers only part of the AI-103 outline [2] [5].

How much does AI-103 cost?

Microsoft has not published a specific fee for the AI-103 beta on its certification page. Standard Azure associate exams are commonly listed at around USD 165, but the price varies by country and beta exams are sometimes offered at a reduced rate. Check the official exam page for the current price in your region [6].

Conclusion

AI-102 retiring on 30 June 2026 is a firm deadline, not a suggestion. If you are mid-preparation, the honest advice is simple: book and sit AI-102 before the date, bank the certification, and let it sit valid on your transcript until it expires. If you are starting fresh, AI-103 is the path forward, but it is a genuinely different, Foundry-centred, agent-heavy exam that is still in beta, so plan your study around the new outline rather than the old one.

Either way, a strong Azure AI foundation is the part that travels. If you want to build that base first, start with Examinotion's AI-901 practice resources, then browse the full Microsoft AI exam catalogue to plan the rest of your certification journey with confidence.

Sources

1. Microsoft Certified: Azure AI Engineer Associate — Microsoft Learn, accessed 2026-06-15
2. Study guide for Exam AI-102: Designing and Implementing a Microsoft Azure AI Solution — Microsoft Learn, accessed 2026-06-15
3. Credential retirement — Microsoft Learn, accessed 2026-06-15
4. Retired certifications and exams — Microsoft Learn, accessed 2026-06-15
5. Study guide for Exam AI-103: Developing AI Apps and Agents on Azure — Microsoft Learn, accessed 2026-06-15
6. Microsoft Certified: Azure AI Apps and Agents Developer Associate — Microsoft Learn, accessed 2026-06-15
7. Microsoft Certified: Azure AI Fundamentals — Microsoft Learn, accessed 2026-06-15