AI-103 guide
Azure AI apps and agents, Foundry, retrieval, tools, orchestration, safety, and monitoring.
AI-103 is the newer Azure AI apps and agents path and Microsoft currently labels it beta. AI-102 remains the established Azure AI Engineer Associate path, but the official study guide warns that the exam retires on June 30, 2026. Based on those official pages, the safer reading is that these are overlapping but distinct targets during a transition, not a clearly announced one-to-one replacement.
Important accuracy note: Microsoft Learn does not explicitly state on these pages that AI-103 replaces AI-102, so this page does not make that claim.
The official AI-103 certification page on Microsoft Learn labels the credential beta. The official AI-102 study guide carries a warning that the exam retires on June 30, 2026, at 11:59 PM Central Standard Time. Based on those official sources, Microsoft Learn currently presents AI-103 and AI-102 as separate certifications with overlapping subject matter during a transition period. That sentence is our inference from the pages above, not a Microsoft quote.
| Topic | AI-103 | AI-102 |
|---|---|---|
| Status | Beta on Microsoft Learn | Established certification path with retirement warning on June 30, 2026 |
| Audience | Azure AI engineers building, managing, and deploying agents and AI solutions that use Microsoft Foundry | Azure AI engineers building, managing, and deploying AI solutions across development, deployment, integration, maintenance, tuning, and monitoring |
| Language expectation | Official page specifically calls out Python experience | Official study guide says candidates can use languages such as Python and C# plus REST APIs and SDKs |
| Skills emphasis | Plan and manage Azure AI, implement generative AI and agentic solutions, plus vision, text analysis, and information extraction | Plan and manage Azure AI, implement generative AI, implement an agentic solution, plus vision, NLP, and knowledge mining or information extraction |
| Agent focus | Higher: generative AI and agentic solutions are 30-35% of the exam | Lower but still present: generative AI is 15-20% and agentic solutions are 5-10% |
| Practice link | AI-103 practice | AI-102 practice |
The official Microsoft Learn page says AI-103 validates expertise in designing, developing, and deploying advanced Azure AI solutions using Python and Microsoft Foundry. The study guide says the audience builds, manages, and deploys agents and AI solutions that take advantage of Microsoft Foundry, and the largest skill block is implementing generative AI and agentic solutions. Start with the AI-103 guide or jump into practice.
The official AI-102 study guide still maps well to learners who want the classic Azure AI Engineer Associate lane with broad Azure AI solution delivery, natural language processing, knowledge mining, and information extraction coverage. It also explicitly says candidates should use REST APIs and SDKs and may build with Python or C#. Start with AI-102 practice if this is your target and your timeline works before the June 30, 2026 retirement date.
AI-103 reads like the newer Foundry-centered apps-and-agents lane. AI-102 reads like the broader Azure AI engineer lane that still includes generative AI and agents but keeps more weight on natural language processing, knowledge mining, and classic service implementation breadth. If your day-to-day work is drifting toward agents, orchestration, grounding, and newer Foundry-first delivery, AI-103 is the cleaner topical fit. If your work is still broad Azure AI solution implementation and you want the established cert while it is active, AI-102 can still be the better target.
The daily practice pages are the fastest way to feel the difference in emphasis before you spend more study time.
Azure AI apps and agents, Foundry, retrieval, tools, orchestration, safety, and monitoring.
Compare Azure AI apps and agents against the GitHub agentic AI developer lane.
Microsoft does not explicitly say that on the official pages referenced here. As of June 9, 2026, Microsoft Learn shows AI-103 as a separate beta credential, while the AI-102 study guide carries a retirement warning for June 30, 2026.
Choose AI-103 for the newer Azure AI apps and agents lane. Choose AI-102 if you specifically want the current Azure AI Engineer Associate path and can finish before retirement.
Yes. Microsoft Learn currently labels the Azure AI Apps and Agents Developer Associate certification as beta.
The official AI-102 study guide warns that the exam retires on June 30, 2026, at 11:59 PM Central Standard Time.
AI-103 is more heavily weighted toward generative AI and agentic solutions, while AI-102 still includes agents but spreads more weight across NLP and knowledge mining.
AI-102 is broader across NLP, speech, knowledge mining, and older Azure AI service implementation patterns.
Microsoft, Azure, and related exam names are trademarks of Microsoft. dotCreds is not affiliated with Microsoft.
Flexible search understands Microsoft exam names, Azure AI, Foundry, agents, and related terms.