AI-103 Study Guide: Microsoft Foundry, Agents, and RAG Explained

Most AI-103 material online falls into one of two traps: a repackaged AI-102 guide with a new exam code, or a two-paragraph overview that tells you the exam exists without telling you what's on it.

AI-103 is not AI-102 renamed. It is built around Microsoft Foundry, agentic AI, and RAG, and nearly a third of the content has no AI-102 equivalent. Here is what the exam tests, in the proportions it tests them.

Quick facts: The exam contains 40 to 60 questions, lasts 120 minutes, and requires a score of 700 out of 1000 to pass. It costs $165 USD and includes an annual free renewal.

Related: Azure AI Engineer Roadmap ↗

The Domain Weights (Read This First)

Domain	Weight
Generative AI and Agentic Solutions	30-35%
Plan and Manage Azure AI Solutions	25-30%
Computer Vision Solutions	10-15%
Text Analysis Solutions	10-15%
Information Extraction Solutions	10-15%

One domain is worth more than the other three combined. If your study plan gives all five equal time, you're underinvesting in the section that decides whether you pass.

Microsoft Foundry: The Platform Everything Sits On

Skip this and every later topic feels harder than it should.

Hub is the top-level container. It holds connections to Azure AI Search, Storage, Key Vault, and shared security/networking config.
Project sits inside a hub. This is where model deployments, agents, and evaluations occur.
Model deployment requires choosing from LLMs, small language models, multimodal models, or code models. The exam tests this as a decision, not a definition.
Connections define how a project reaches Azure AI Search, Storage, and Key Vault. Every RAG or agent question assumes you understand these are configured, not just present.

Do not treat Foundry as the new portal. Treat it as an architecture you can draw from memory.

Agents: The Domain That Decides Your Score

An agent is not a smarter chatbot. It runs a loop: reason → act → observe → repeat. Questions describe a requirement and ask what tools and instructions the agent needs because you are designing the loop, not recognizing it.

Know these cold:

Instructions define the agent's role and boundaries
Tools include function tools, Code Interpreter, Azure AI Search grounding, and Bing grounding
Threads manage how conversation memory persists across turns
Tool schemas provide the structured contract for what a tool accepts and returns

Two orchestration patterns get tested:

Orchestrated multi-agent uses one orchestrator to route to specialist agents, such as a billing agent or support agent
Autonomous/semiautonomous agents act with safeguards and human approval checkpoints

Governance matters as much as capability. Agents need Azure managed identities scoped with RBAC, not inherited user credentials or shared service accounts. This is scored content, not a nice-to-have.

The exam guide references "Foundry SDKs and connectors" without naming a specific orchestration framework. Study the concepts of plugins, function-calling, and planners, not any one SDK's syntax.

RAG: Shows Up Everywhere, Not Just One Domain

A model's knowledge is frozen at training time. RAG retrieves current, relevant data at query time and grounds the response in it. This is not confined to Information Extraction; it is core to the agents domain too.

Azure AI Search, the primary retrieval engine, tests three modes:

Vector search measures similarity by meaning
Hybrid search combines vector and keyword search, and is the default for most RAG scenarios
Semantic ranking reranks hybrid results for top relevance at the cost of latency

The ingestion pipeline: Document Intelligence (OCR/layout extraction) → chunking (semantic chunking usually beats fixed-size) → enrichment → index. The exam tests the whole flow from document to agent, not retrieval in isolation.

Evaluation is where candidates lose easy points. The Azure AI Evaluation SDK checks:

Groundedness checks if the answer is supported by retrieved content.
Relevance checks if retrieval pulled the right content.
Quality checks if the output is accurate and well-structured.
Safety checks for harmful or biased content.

A question describing a plausible-but-wrong RAG answer is testing whether you know which of these four to check first.

Content Understanding: The Service Most Study Guides Miss

Azure AI Content Understanding uses LLM-based extraction (not traditional OCR) to turn documents, images, audio, and video into structured output. It shows up in three places:

Computer Vision extracts visual characteristics from images and video
Information Extraction produces clean, grounded data for RAG and agents
Agents use it as a tool they can call directly

If your prep materials don't mention this service, they're incomplete for this exam.

Plan and Manage Azure AI Solutions (25-30%)

This domain represents the governance layer underneath everything else, which covers four areas:

Service selection requires choosing the right model, retrieval method, and agent tooling for the scenario.
Infrastructure and deployment covers Azure AI infrastructure design, deployment options, and CI/CD integration.
Monitoring and security includes managing quotas, costs, drift detection, managed identities, and private networking.
Responsible AI involves implementing safety filters, guardrails, trace logging, and agent oversight controls.

Questions here regularly combine two of these at once — service selection plus security, or monitoring plus responsible AI. Don't study this domain as isolated bullet points.

Related: How Hard Is AI-103 ↗

The Three Smaller Domains, in Brief

Computer Vision (10-15%) covers image and video generation from prompts, multimodal understanding, and unsafe content or prompt-injection detection
Text Analysis (10-15%) covers entity, sentiment, and summarization via generative prompting, plus speech framed as an agent modality
Information Extraction (10-15%) covers document extraction, RAG ingestion with OCR, and Content Understanding analyzers

These domains are individually smaller, but they collectively make up a third of the exam. Do not skip them, but do not over-invest either.

The Mistake Most Candidates Make

The biggest mistake is treating this like a vocabulary exam. Knowing that Azure AI Search does vector search, or that agents use tools, gets you through the easy questions. It won't get you through a scenario asking you to pick between three plausible architectures under a stated constraint (cost, latency, data sensitivity, safety).

Build these before your exam, not just read about them:

Build one working RAG pipeline.
Build one multi-agent system with tool-access governance.

Frequently Asked Questions (FAQ)

How is this different from AI-102?

The new exam is Foundry-centric instead of using separate service endpoints. Generative AI and agents (30-35%) barely existed on AI-102. Image/video generation, Content Understanding, and agent governance are all new. Expect 40-50% new content if you're coming from AI-102.

Do I need Python?

Yes. This is a developer exam — you'll read and reason about Foundry SDK code throughout.

How much hands-on practice do I need?

You need significant hands-on practice. Build a real RAG pipeline and a real multi-agent system in Foundry. Configure Content Understanding analyzers. Set up safety filters and guardrails. Scenario questions assume you have done these things.

Is Semantic Kernel on the exam?

Semantic Kernel is not named explicitly, but it is the dominant orchestration framework in this ecosystem. Study the concepts (plugins, function-calling, planners), not SDK-specific syntax.

What's the passing score?**

The passing score is 700 out of 1000. The exam uses scaled scoring, so do not try to back-calculate how many questions you can miss.

Can I use reference materials during the exam?**

Yes. You can use the integrated Microsoft Learn platform. Practice searching so that you do not waste too much time.

Want to pressure-test your understanding with scenario-based practice across all five domains? ExamOS offers the Daily Sprint, Mock Exam, and Readiness Score to build judgment for Foundry architecture, agentic solutions, and RAG design decisions.

AI-103 Study Guide: Microsoft Foundry, Agents, and RAG