Exam prep · AI & Business Value

AI-103 — Developing AI Apps and Agents on Azure

Building AI applications and agents on Azure — generative and agentic solutions first, then vision, language and document intelligence.

Is this exam for you?

Who takes AI-103

Developers building AI into applications on Azure. Earns the Azure AI Apps and Agents Developer Associate certification.

Level: Intermediate · Official credential: Developing AI Apps and Agents on Azure

About the course

What AI-103 actually covers

The developer exam for AI on Azure, and its weighting tells you where the industry has moved: generative AI and agentic solutions carry the largest share, with planning and managing the overall Azure AI solution close behind. Computer vision, text analysis and information extraction share the remaining third between them — still there, no longer the main event.

Assumed knowledge: Professional development experience and familiarity with Azure. Comfort with REST APIs and at least one SDK language is assumed.

4 days · AI-103T00Course length
120 minutesExam
IntermediateTechnical level
YesHands-on labs

The problems it answers

Why people in this role end up sitting AI-103

  1. A proof of concept that impressed everyone and cannot be put in front of customers.
  2. Choosing between Foundry, the SDKs and direct API calls, then living with that choice.
  3. Grounding a model on company data without leaking it or hallucinating around it.
  4. Costs that scale with usage, discovered after the pilot rather than during it.
  5. Evaluating whether an AI feature is actually working, when the output is different every time.

Hands-on labs

What you actually build

Microsoft has not published a single AI-103 lab repository. The exercises are maintained as separate per-topic repositories covering agents, Foundry, vision and language, so the hands-on material exists but is not bundled under one course name yet.

For training partners

Selling AI-103

Everything a partner needs to put this course in front of the right customer. Lift any of it straight into a proposal.

“Their developers have built an AI demo. This is the course that turns it into something they can support in production.”

When to bring it up

  • An AI proof of concept that has stalled before release.
  • Developers choosing between Foundry, SDKs and raw API calls with no guidance.
  • An AI feature shipped and now producing costs or answers nobody can explain.
  • A team moving from consuming Copilot to building their own agents.
  • An AB-100 architecture that needs developers who can implement it.

Who signs: Development manager or head of engineering. A different budget from infrastructure training.
Who attends: Professional developers with Azure exposure. Not a course for makers or administrators.

What to sell next

  • AB-100 for the architecture above the build
  • AB-620 for the low-code agent builders alongside them
  • AZ-204 if the general Azure development grounding is missing
  • SC-500 once the AI workloads need securing

Pushbacks you will hear

And a straight answer for each

Our developers are already using the OpenAI APIs.

Then they have the easy third. Grounding, evaluation and cost management are where AI projects actually fail, and those are most of this exam.

The AI landscape changes too fast to certify against.

The services change; the problems do not. Grounding, evaluation and deployment are the durable parts, and they are what this covers.

Is there a lab environment?

Microsoft has not bundled AI-103 labs into one repository yet — the exercises are published per topic. Worth setting that expectation honestly up front.

Can our makers attend instead of developers?

No. Send makers to AB-620 for Copilot Studio. This one assumes code.

Instructor-led

Train for AI-103 with Alex

We deliver the official Microsoft course for AI-103 — in person or virtual, public or in-company. Real demos, real stories, and the whiteboards to take home.

Book this training

Prefer to browse first? See the full course catalog or the upcoming dates.

Same track

Related certifications

AB-100

Agentic AI Business Solutions Architect

Exam prep ›

AB-210

Accelerating Sales Pipelines with AI in Dynamics 365

Exam prep ›

AB-250

Transforming Contact Center Experiences with AI in Dynamics 365

Exam prep ›

AB-400

Extending Microsoft Power Platform Solutions with Code and AI

Exam prep ›

AB-410

Building Intelligent Applications

Exam prep ›

AB-620

Designing and Building Integrated AI Agent Solutions in Copilot Studio

Exam prep ›

AB-650

Administering Microsoft 365 and AI Services

Exam prep ›

AB-730

AI Business Professional

Exam prep ›

AB-731

AI Transformation Leader

Exam prep ›

AB-900

Microsoft 365 Copilot and Agent Administration Fundamentals

Exam prep ›

AI-300

Operationalizing Machine Learning and Generative AI Solutions

Exam prep ›

AI-500

Designing and Implementing Multi-Agent AI Solutions

Exam prep ›

AI-900

Microsoft Azure AI Fundamentals

Exam prep ›

Exam titles, study guides and practice assessments are published by Microsoft and linked here directly; the descriptions on this page are our own. Microsoft may update exam content at any time — the official study guide is always the authority.