AI-901 course
AI-901 free course: Azure AI Fundamentals
A free written AI-901 course in 12 lessons with diagrams: AI concepts, responsible AI and building with Microsoft Foundry, in exam order.
This free written course teaches the AI-901 skills outline from the ground up, in the order the exam is built. Each lesson takes about five to seven minutes to read, builds one idea with a diagram or a worked example, and ends with three check questions. No account is needed.
The lessons follow the skills outline as published on April 15, 2026, the version that applies to the exam now. The two skill areas map to the lessons like this:
| Lessons | Skill area | Weight |
|---|---|---|
| 1 to 5 | Identify AI concepts and capabilities | 40–45% |
| 6 to 11 | Implement AI solutions by using Microsoft Foundry | 55–60% |
| 12 | Both together: how the questions are built | – |
How to use the course
Read the lessons in order the first time, because the Foundry lessons rely on the vocabulary of the concept lessons: workloads, tokens, prompts and responsible AI come back in almost every implementation question. After a lesson, answer its check questions before reading the next one.
Because the larger skill area is about implementing, the course describes what a lightweight application or portal step does and in which order, rather than teaching syntax. That matches what a fundamentals exam asks: which capability fits a scenario, and what comes next.
When the order is clear, use the other guides for what a course cannot do. The exam format page explains the scaled score and the question types. The study plan spreads the work over four weeks. The cheat sheet is the page to reread the day before. The free practice test and sample questions show how scenario questions read.
Lessons
- The AI workload mapLearn how AI, machine learning, generative AI and agents relate, and match a business need to one of five AI workload families for AI-901.
- Responsible AI principlesLearn the six responsible AI principles with a concrete example each, and how AI-901 scenario questions hide the principle behind a described problem.
- How generative models workUnderstand tokens, context windows, prompts, temperature, model choice and deployment types, the generative AI basics behind AI-901 concepts questions.
- Text analysis and speechTell keyword extraction, entity detection, sentiment analysis and summarization apart, and match speech recognition and synthesis to scenarios.
- Vision and extractionLearn what computer vision, image-generation and information-extraction techniques do, and when pulling out fields beats generating new text or images.
- Foundry first stepsLearn what Microsoft Foundry is, how projects, the model catalog and deployments fit together, and how the playground and a chat client call a model.
- System and user promptsWrite effective system and user prompts: what each role does, the parts of a good prompt, grounding with your own data (RAG), and common prompt mistakes.
- Agents in FoundryLearn what an agent adds to a model in Foundry: instructions, tools and knowledge, how to create and test one in the portal, and how a client app calls it.
- Text and speech solutionsBuild text analysis and speech apps in Foundry: when to call a language service, a multimodal model with audio, or Azure Speech for spoken input.
- Vision and image appsSend images to a multimodal model, create pictures with an image-generation model, and shape a lightweight vision app in Microsoft Foundry.
- Content UnderstandingExtract fields from documents, images, audio and video with Azure Content Understanding analyzers and use the structured result in a lightweight app.
- Putting it togetherWork three cross-domain AI-901 scenarios, use a need-to-capability decision table for Foundry, and learn a routine for reading scenario questions.