AI-901 Cheat Sheet
Azure AI Fundamentals — Quick reference for the exam
CLI Commands
15az cognitiveservices account create -n <nombre> -g <rg> --kind CognitiveServices --sku S0 -l eastusTOPCreate a multi-service Azure AI resource
az cognitiveservices account listList Foundry Tools (Azure AI Services) resources
az cognitiveservices account keys list -n <nombre> -g <rg>TOPGet the resource access keys
az cognitiveservices account show -n <nombre> -g <rg>Show AI resource details
Comparisons
5| Category | Service | Main function |
|---|---|---|
| Vision | Azure AI Vision | Image analysis, OCR, object and face detection |
| Speech | Azure AI Speech | Text-to-Speech (TTS), Speech-to-Text (STT), speech translation |
| Language | Azure AI Language | NLP: sentiment, entities, summarization, classification |
| Translation | Azure AI Translator | Neural machine translation (100+ languages) |
| Search | Azure AI Search | Vector and semantic search, RAG over documents |
| GenAI | Azure OpenAI Service | GPT-5 family, gpt-image, audio models, embeddings |
| Safety | Azure AI Content Safety | Content moderation: violence, hate, sexual content |
| Custom | Azure Machine Learning | Train, deploy, and manage your own ML models |
Quick Reference
30Responsible AI — 6 principles
Fairness — no unjust bias in automated decisions
Reliability & Safety — correct behavior under unexpected conditions
Privacy & Security — protecting personal data and privacy
Inclusiveness — benefiting and empowering everyone
Transparency — understandable and explainable systems
Accountability — humans responsible for the AI systems they build
Azure OpenAI models
Multimodal model (text + vision). The most capable for chat and reasoning
Faster, cheaper version of GPT-4o
Image generation from text
High-quality speech-to-text (STT) transcription
Generates embeddings for semantic search and RAG
Reasoning models for complex step-by-step tasks
ML evaluation metrics
% of correct predictions out of the total
Of the predicted positives, how many are actually positive
Of the actual positives, how many we correctly detect
Harmonic mean of Precision and Recall. Balance between the two
Area under the ROC curve — measures ability to classify correctly
Root Mean Squared Error — error for regression problems
Azure AI Studio and Foundry
Unified platform for building generative AI apps
Hub = shared resources across projects. Project = workspace
Library of models (OpenAI, Meta Llama, Mistral, etc.)
Visual prompt flows; retires in April 2027 (replacement: Microsoft Agent Framework)
Automatic filters for harmful content in Azure OpenAI
Groundedness, Relevance, Coherence, Fluency for evaluating responses
AI services — quick reference
Analyzes images: objects, faces, OCR, spatial analysis
STT + TTS + Speaker Recognition + Speech Translation
Sentiment, entities, classification, QA, summarization
Text translation across 100+ languages via API
Vector search + RAG over your own documents
Build multi-channel chatbots (Teams, Web, etc.)
Ready to practice?
Put what you learned to the test with original AI-901 exam questions.
Start AI-901 mock examUnlimited mock exams with the Pro plan