Azure AI Fundamentals (AI-900) Study Guide: Pass in 3 Weeks

What Does the Azure AI-900 Exam Cover and How Long Does It Take?

Azure AI Fundamentals (AI-900) is Microsoft's foundational AI certification. It tests knowledge of AI and ML concepts, Azure AI services, computer vision, natural language processing, and generative AI on Azure. The exam costs $99, consists of 40 to 60 questions answered in 45 minutes of testing time (65 minutes total seat time), and requires a passing score of 700 out of 1,000. Full details are at learn.microsoft.com/en-us/credentials/certifications/azure-ai-fundamentals.

Three weeks is achievable for most candidates approaching this exam without a deep technical background, at roughly six to eight hours per week (18 to 24 hours total). Candidates with existing Microsoft Azure experience (AZ-900 or similar) can finish in two weeks.

How Are the Azure AI-900 Skill Areas Weighted?

The exam is organized into five skill areas with the following approximate weights:

• Describe AI workloads and considerations — 15–20%
• Describe fundamental principles of machine learning on Azure — 20–25%
• Describe features of computer vision workloads on Azure — 15–20%
• Describe features of natural language processing (NLP) workloads on Azure — 15–20%
• Describe features of generative AI workloads on Azure — 20–25%

Machine learning fundamentals and generative AI are the two largest skill areas—together they account for roughly 40 to 50% of the exam. The 3-week plan below allocates study time accordingly.

Week 1 — AI Workloads and Machine Learning Fundamentals (Skill Areas 1 and 2)

Days 1–2: Skill Area 1 — Describe AI Workloads and Considerations (15–20%)

This skill area covers AI use cases (prediction, classification, object detection, NLP, anomaly detection, conversational AI), responsible AI principles, and AI considerations for fairness, reliability, privacy, transparency, and inclusiveness. Microsoft's six responsible AI principles are tested by name and application—know them before moving to Week 2.

Primary resource: Microsoft Learn's free "Get started with AI on Azure" learning path covers this material in approximately three hours.

Days 3–7: Skill Area 2 — Fundamental Principles of Machine Learning on Azure (20–25%)

This is the largest skill area and the one most candidates under-prepare for. It covers:

• Core ML concepts: features, labels, training and validation data, supervised vs. unsupervised learning
• ML pipeline stages: data preparation, model training, evaluation, deployment
• Regression, classification, and clustering: when each applies
• Evaluation metrics: for regression (mean absolute error, root mean squared error, R²); for classification (accuracy, precision, recall, AUC)
• Azure Machine Learning service — Azure's ML platform. The exam tests: Azure ML Studio (visual designer), Automated ML (AutoML), and compute types (training clusters, inference endpoints)

Microsoft Learn's "Fundamentals of machine learning" learning path covers all of this and includes hands-on exercises in Azure ML Studio. Complete the exercises—the exam asks about the Azure ML interface and capabilities.

Week 1 target: Skill Areas 1 and 2 complete. Quiz yourself on: Microsoft's six responsible AI principles by name, the difference between supervised and unsupervised learning, and what Azure ML AutoML does vs. the Azure ML designer.

Week 2 — Computer Vision and Natural Language Processing (Skill Areas 3 and 4)

Days 1–3: Skill Area 3 — Computer Vision Workloads on Azure (15–20%)

This skill area tests knowledge of Azure computer vision services and what each one does:

• Azure AI Vision (formerly Computer Vision) — Image analysis, object detection, image classification, optical character recognition (OCR), spatial analysis
• Azure AI Custom Vision — Training custom image classifiers and object detectors with your own labeled images
• Azure AI Face — Face detection, verification, and identification
• Azure AI Video Indexer — Extracting insights from video: transcription, speaker identification, scene detection, object recognition in video

The exam matches services to use cases. A question describing a retail store that wants to count customers by age group in video footage points to Azure AI Video Indexer or Azure AI Face. A question about reading text from scanned forms points to Azure AI Vision (OCR) or Azure AI Document Intelligence.

Days 4–7: Skill Area 4 — Natural Language Processing Workloads on Azure (15–20%)

Azure NLP services tested on AI-900:

• Azure AI Language — Key phrase extraction, named entity recognition, sentiment analysis, language detection, summarization, question answering (built-in and custom knowledge bases)
• Azure AI Translator — Neural machine translation across 100+ languages
• Azure AI Speech — Speech-to-text, text-to-speech, speech translation, speaker recognition
• Azure AI Bot Service — Building conversational bots; integrates with Azure AI Language for question answering
• Azure AI Language Understanding (CLU) — Training custom models to understand user intent in conversational AI applications

Week 2 target: Skill Areas 3 and 4 complete. For each Azure AI service in both areas, you should be able to answer: what does it do, what input does it take, what output does it produce, and what use case should trigger choosing it?

Week 3 — Generative AI and Full Exam Review

Days 1–4: Skill Area 5 — Generative AI Workloads on Azure (20–25%)

Generative AI is one of the two largest skill areas and the newest addition to the AI-900 content. It tests:

• Foundation models and large language models — What they are, how they work, what training means for an LLM, and what fine-tuning does
• Azure OpenAI Service — Microsoft's managed access to OpenAI models (GPT-4, DALL-E, Whisper) via Azure. The exam tests what the service provides, how to access it, and what use cases it addresses
• Prompt engineering — Zero-shot, few-shot, and system prompts in the context of Azure OpenAI
• Responsible generative AI — Content filtering in Azure OpenAI, managing harmful outputs, Microsoft's content safety tools
• Copilot applications — Microsoft Copilot as an application of generative AI; how Copilot integrates into Microsoft 365 products

The Microsoft Learn "Fundamentals of generative AI" learning path covers this material. Complete the hands-on exercises with Azure OpenAI in Azure AI Studio—the exam tests the interface and workflow.

Days 5–7: Full Exam Review and Practice Tests

Spend the final days of Week 3 doing full practice exams and reviewing weak areas. Best practice resources for AI-900:

• Microsoft Learn official practice assessment — Free, available on the AI-900 certification page. Approximately 50 questions in the style of real exam questions.
• Measure Up AI-900 practice tests — Microsoft's official practice test partner. More representative than most third-party options.
• Tutorials Dojo AI-900 practice exams — Well-regarded third-party option with detailed explanations.

Score each practice exam by skill area. If you are below 75% on any area, go back to the Microsoft Learn learning path for that specific skill area and review before attempting the real exam.

Week 3 target: All five skill areas reviewed, practice exam scores at 750 or above, real exam booked and passed.

Exam details verified against learn.microsoft.com/en-us/credentials/certifications/azure-ai-fundamentals as of 2026-02-27. Fees and requirements are subject to change — confirm current details at learn.microsoft.com before your exam date.