AI-103 is Microsoft’s current intermediate credential exam for developers and AI engineers who build Azure AI applications and agents with Microsoft Foundry. Passing it earns Microsoft Certified: Azure AI Apps and Agents Developer Associate. That positioning matters because Microsoft’s 2026 credential changes created several new AI paths that overlap in technology but target different responsibilities.
The AI-103 exam is centered on application engineering: planning Azure AI solutions, implementing generative and agentic systems, working with computer vision and text, extracting information, and managing security, evaluation, deployment, and monitoring. It is not a fundamentals credential, a general Azure administration credential, or a dedicated MLOps credential.
Candidates can therefore place AI-103 correctly by comparing role boundaries rather than assuming every Microsoft AI exam is a step on one mandatory ladder. Some credentials sit below it conceptually, some are adjacent specializations, and one current expert path can use AI-103 as an accepted associate-level prerequisite.
AI-103 replaced the retired AI-102 path in 2026
Microsoft retired AI-102 and Microsoft Certified: Azure AI Engineer Associate on June 30, 2026. The replacement credential, Azure AI Apps and Agents Developer Associate, became generally available in June 2026 with AI-103 as its exam. This was not merely a code change. The role was redesigned around the way Azure AI development had shifted toward Foundry, generative applications, agents, RAG, multimodal systems, and modern operational controls.
The historical AI-102 material can still explain older Azure AI services and the evolution of the role, but it should not be used as the current blueprint. AI-103 gives much more explicit weight to agent workflows, tool integration, grounding, model and application evaluation, Content Understanding, tracing, token analytics, and safeguards around autonomous behavior.
For professionals who earned or studied AI-102, the right approach is a gap analysis. Existing knowledge of Azure AI language, vision, speech, and search remains valuable. The new preparation work is connecting that background to Foundry-based application design and the generative/agentic patterns now at the center of the role.
AI-901 is the current fundamentals path below AI-103 in depth
Microsoft Certified: Azure AI Fundamentals now uses AI-901. It is aimed at people at the beginning of an AI solution-development career and covers core AI concepts together with foundational technical skills in Microsoft Foundry. The AI-901 exam can therefore be useful preparation for candidates who need to establish vocabulary and basic platform familiarity.
AI-901 is not listed as a prerequisite for AI-103. An experienced developer can move directly to AI-103 if the assumed knowledge is already present. The conceptual relationship is more important than a formal requirement: AI-901 explains what major AI capabilities are and how they appear in Azure; AI-103 asks candidates to build and operate applications that use them.
The depth difference is significant. A fundamentals candidate might identify responsible AI principles or describe computer-vision capabilities. An AI-103 candidate may need to choose a multimodal architecture, configure safety controls, handle prompt injection, connect retrieval, and evaluate the resulting application. The same topic appears at a different engineering level.
AI-200 is adjacent for developers who want broader Azure application depth
Microsoft Certified: Azure AI Cloud Developer Associate uses AI-200. Its current role profile emphasizes designing, building, and implementing AI solutions on Azure with a strong focus on back-end services, scalable architectures, and the full development lifecycle. It includes Azure SDKs, data management, monitoring, messaging, eventing, vector databases, containers, Python, and cloud application components.
That makes AI-200 an adjacent credential rather than a required next step. AI-103 goes deeper into AI apps, agents, Foundry, RAG, multimodal workloads, and information extraction. AI-200 broadens the developer’s ability to build the cloud services and back-end architecture around AI workloads.
A candidate whose AI-103 preparation revealed weaknesses in containerized deployment, messaging, scalable back-end design, or general Azure application engineering may find AI-200 particularly relevant. Someone already strong in those areas may prefer a different specialization after AI-103.
AI-300 is adjacent for engineers who want MLOps and GenAIOps depth
Microsoft Certified: Machine Learning Operations Engineer Associate uses AI-300. It focuses on the infrastructure and operational practices required for machine learning operations and generative AI operations on Azure. Its role includes MLOps infrastructure, model lifecycle operations, GenAIOps infrastructure, generative AI quality assurance, observability, and performance optimization.
The AI-300 exam therefore extends a part of AI-103 rather than replacing it. AI-103 expects candidates to monitor, evaluate, and operationalize generative systems as part of application engineering. AI-300 makes operations itself the main subject, with deeper emphasis on automation, lifecycle management, Azure Machine Learning, Foundry, GitHub Actions, and infrastructure as code.
This is a natural direction for candidates who enjoy the production side of AI: deployment automation, observability, quality gates, scaling, repeatability, and model/application lifecycle management. It is less about building one user-facing agent and more about establishing the systems that keep many AI assets reliable over time.
AB-100 creates an expert-level architecture route where AI-103 can count as the associate prerequisite
Microsoft Certified: Agentic AI Business Solutions Architect Expert uses AB-100. The current expert path requires AB-100 together with an accepted associate certification, and AI-103 is among the listed associate credentials that can satisfy that prerequisite. This creates a genuine progression route for professionals moving from implementation into broader solution architecture.
The AB-100 exam is not simply an “advanced AI-103.” Its focus is solution architecture across Microsoft AI apps, services, and business-application technologies. The architect role works at a wider decision level: translating business needs, designing end-to-end solutions, coordinating platform choices, and guiding delivery across teams.
AI-103 can provide the implementation credibility needed for that path, but experience matters. An expert architecture credential is most useful when the candidate has seen real trade-offs in security, integration, operations, data, adoption, and governance rather than only completing a sequence of exams.
Microsoft’s AI credentials now form a role map more than a single ladder
The current portfolio is easier to understand as a set of role directions. AI-901 establishes fundamental Azure AI literacy. AI-103 validates development of Foundry-based AI apps and agents. AI-200 emphasizes cloud application and back-end engineering around AI solutions. AI-300 emphasizes the operational lifecycle of machine learning and generative AI. AB-100 operates at an expert architecture level.
These roles overlap because real AI projects overlap. An AI-103 engineer should understand deployment and monitoring. An AI-300 engineer still needs to understand generative applications. An AI-200 developer may build services consumed by agents. An AB-100 architect needs enough implementation knowledge to make credible platform decisions. The credential boundaries describe primary responsibility, not sealed technology silos.
This is why candidates should not choose the next exam only by number. AI-200 is not necessarily “after” AI-103, and AI-300 is not automatically more advanced in every dimension. They validate different concentrations at a similar intermediate level.
AI-103 also sits inside the wider Azure and software-engineering ecosystem
AI applications depend on Azure identity, networking, compute, storage, monitoring, source control, and delivery. Professionals may therefore combine AI-103 with non-AI credentials when their jobs demand deeper cloud-platform or DevOps responsibility. That can be more useful than collecting several AI credentials that overlap heavily with work they already do.
The right path follows the role. A developer responsible for AI APIs and containerized back ends may prioritize AI-200. An engineer responsible for GenAIOps may prioritize AI-300. A cloud engineer with weak infrastructure knowledge may first deepen Azure administration. A senior consultant designing cross-platform business solutions may eventually move toward AB-100.
The broad Microsoft certifications inventory is therefore best used as a role map. AI-103 is the anchor for a specific role: intermediate Azure AI application and agent development with Foundry.
That also means a candidate does not need to turn every adjacent technology gap into another exam. Someone who already works comfortably with Azure identity, networking, source control, deployment automation, and monitoring may get more value from deeper project work than from adding an infrastructure credential. By contrast, an AI developer who can write prompts and agent logic but struggles to reason about managed identities, private connectivity, release controls, or production telemetry has a genuine platform gap worth closing. The broader Microsoft path helps expose those boundaries. AI-103 should strengthen a real working profile, while the next credential should add a capability that the candidate does not already demonstrate through day-to-day responsibility.
The strongest credential path is built around increasing responsibility
A useful progression is not “fundamentals, associate, expert” merely because the labels increase. It is a progression in responsibility. Fundamentals proves literacy. AI-103 proves that a candidate can build and operate an application. Adjacent associate credentials can broaden cloud development or operations. Expert architecture becomes appropriate when the professional is responsible for whole solutions and cross-team decisions.
Microsoft associate, expert, and specialty certifications also follow an ongoing renewal model, so earning AI-103 should be treated as the start of maintaining current skills rather than a permanent endpoint. The technology behind Foundry, agents, retrieval, and multimodal systems is changing quickly enough that continuous learning matters even when the credential remains active.
AI-103 occupies a useful center in this structure. It is technical enough to validate real application work and broad enough to connect naturally to several specialties. Candidates who understand that position can choose the next step based on the kind of AI engineer, developer, operator, or architect they actually want to become.