AI-900 — Microsoft Azure AI Fundamentals Cheat Sheet

Last revised: September 16, 2026

Cheat sheet: AI-900 reference for Microsoft Azure AI Fundamentals: AI workloads, Azure AI services, machine learning, vision, language, and generative AI.

Use the tables for a quick pre-exam check. Expand a topic’s notes for explanations, examples, and additional distinctions.

Scope and study context

AI-900 is not primarily a coding exam. Expect scenario-style questions such as:

Which Azure service fits a workload?
Is the problem classification, regression, clustering, NLP, vision, or generative AI?
Which responsible AI principle is involved?
Which metric or ML concept best matches the scenario?
How do Azure AI services, Azure Machine Learning, Azure AI Search, and Azure OpenAI Service differ?

The exam is fundamentals-level. Expect questions that test whether you can recognize AI workload types, choose appropriate Azure AI services, understand basic machine learning concepts, and apply responsible AI principles—not whether you can build complex production systems from memory.

This page supports IT Mastery review with original practice questions. It is not affiliated with Microsoft.

High-Yield Azure AI Service Selection

Scenario	Best-fit Azure capability	Key exam cues	Common trap
Predict a category, value, or cluster from data	Azure Machine Learning	Train, evaluate, deploy ML models; AutoML; designer; notebooks	Do not choose Azure AI Vision or Language unless the input is image/text-specific
Build an ML model without writing much code	Azure Machine Learning automated ML / designer	Low-code model training, pipelines, drag-and-drop workflow	AutoML chooses models; it is not the same as generative AI
Analyze images	Azure AI Vision	Tags, captions, object detection, OCR, image analysis	OCR extracts text; it does not understand document structure by itself
Train custom image classification or object detection	Custom vision capabilities	Your own labeled images; custom tags/classes	General image analysis uses prebuilt models
Extract fields from invoices, receipts, IDs, forms	Azure AI Document Intelligence	Key-value pairs, tables, structured document extraction	Do not use simple OCR if the question asks for structured fields
Search large document collections with AI enrichment	Azure AI Search	Indexes, indexers, skillsets, knowledge mining, semantic search	Search is retrieval; it is not primarily model training
Detect sentiment, entities, key phrases, language	Azure AI Language	Text analytics, NER, sentiment, PII, summarization	Translator is specifically for translation
Build intent-based conversational understanding	Azure AI Language - Conversational Language Understanding	Intents, utterances, entities	A chatbot channel is not the same as language understanding
Build FAQ-style answers from a knowledge base	Azure AI Language - question answering	Questions and answers, knowledge base, FAQ	Not the same as open-ended generative chat
Translate text between languages	Azure AI Translator	Text translation, language pairs	Speech translation uses speech services
Convert speech to text or text to speech	Azure AI Speech	Transcription, synthesis, speech translation	Speech-to-text is not OCR
Generate text, summarize, chat, reason over prompts	Azure OpenAI Service / Azure AI Foundry	Large language models, prompts, completions, chat, embeddings	Generative AI can hallucinate; grounding may be needed
Detect harmful AI content	Azure AI Content Safety	Hate, sexual, violence, self-harm, harmful content moderation	Content filtering is not the same as model accuracy
Build a bot that connects to channels	Azure Bot Service	Web chat, Teams, channels, bot conversations	Bot Service hosts orchestration; language models handle understanding

Notes and examples

Azure AI service selection workflow

    flowchart TD
	    A[Read the scenario] --> B{What data type?}
	    B --> C[Tabular data]
	    B --> D[Images or video]
	    B --> E[Documents/forms]
	    B --> F[Text]
	    B --> G[Speech]
	    B --> H[Need generated content]
	    C --> C1{Known label?}
	    C1 -->|Numeric target| C2[Regression in Azure Machine Learning]
	    C1 -->|Category target| C3[Classification in Azure Machine Learning]
	    C1 -->|No labels| C4[Clustering in Azure Machine Learning]
	    D --> D1[Azure AI Vision / Custom Vision]
	    E --> E1[Azure AI Document Intelligence]
	    F --> F1{Translate, analyze, or search?}
	    F1 -->|Analyze language| F2[Azure AI Language]
	    F1 -->|Translate text| F3[Azure AI Translator]
	    F1 -->|Search content| F4[Azure AI Search]
	    G --> G1[Azure AI Speech]
	    H --> H1[Generative AI / Azure OpenAI capabilities]

AI Workload Types

Workload	What it does	Example exam scenario
Machine learning	Learns patterns from data to make predictions	Predict customer churn or house prices
Computer vision	Interprets images and video	Detect products in shelf images
Natural language processing	Understands or generates human language	Extract entities from support tickets
Document intelligence	Extracts structured data from documents	Pull invoice totals and vendor names
Knowledge mining	Extracts searchable insights from content	Enrich PDFs and make them searchable
Generative AI	Creates new content from prompts	Draft answers, summarize reports, generate chat responses
Conversational AI	Enables user interactions through natural language	Customer support chatbot
Anomaly detection	Finds unusual patterns	Identify abnormal sensor readings
Speech AI	Processes spoken language	Transcribe meeting audio

Responsible AI Principles

Microsoft emphasizes responsible AI concepts throughout Azure AI workloads. Know the principle and the scenario cue.

Principle	Meaning	Exam cue	Practical controls
Fairness	AI systems should avoid unfair bias or discrimination	Model performs worse for one demographic group	Representative data, bias testing, human review
Reliability and safety	AI systems should work reliably and safely under expected conditions	Incorrect prediction could cause harm	Testing, monitoring, fallback paths, safe deployment
Privacy and security	Protect data and systems	Personal data in prompts, training data, or logs	Access control, encryption, data minimization
Inclusiveness	AI should work for people with diverse abilities and needs	Application excludes users with disabilities or language needs	Accessibility, multilingual support, inclusive design
Transparency	Users and stakeholders should understand AI behavior and limitations	Need to explain why or how AI is used	Explainability, disclosures, model documentation
Accountability	People and organizations remain responsible for AI outcomes	Who approves model use or handles errors?	Governance, auditability, human oversight

Notes and examples

Responsible AI Traps

If the question says…	Think…
“The model works well overall but poorly for one group”	Fairness
“Users should know they are interacting with AI”	Transparency
“Sensitive customer data is used in prompts”	Privacy and security
“A human must approve high-impact recommendations”	Accountability
“The system must avoid dangerous behavior under edge cases”	Reliability and safety
“The app should support people with different abilities”	Inclusiveness

Responsible AI principles

Microsoft emphasizes responsible AI. For AI-900, know the principle and recognize it from a scenario.

Principle	Quick meaning	Scenario clue
Fairness	AI systems should treat people equitably and avoid unfair bias	A loan model performs worse for a demographic group
Reliability and safety	AI should work dependably and handle unexpected conditions safely	A healthcare model needs validation before use
Privacy and security	AI should protect data and resist misuse	Personal data must be secured or minimized
Inclusiveness	AI should empower and support diverse users	An app must be usable by people with disabilities
Transparency	Users should understand system behavior and limitations	Users need to know why a decision was made
Accountability	People and organizations remain responsible for AI systems	A company needs oversight and governance

Responsible AI exam cues

If the question mentions bias, unequal outcomes, or demographic performance, think fairness.
If it mentions explainability, disclosure, or understanding model behavior, think transparency.
If it mentions human oversight, governance, ownership, or responsibility, think accountability.
If it mentions sensitive data, identity, access, encryption, or privacy, think privacy and security.
If it mentions accessibility or broad user participation, think inclusiveness.
If it mentions testing, robustness, fail-safe behavior, or harm prevention, think reliability and safety.

Core AI and ML Terms

Term	Exam-ready meaning
Artificial intelligence	Broad field of systems that perform tasks associated with human intelligence
Machine learning	AI technique where models learn patterns from data
Deep learning	ML using neural networks with many layers; often used for vision, speech, and language
Model	Trained artifact that maps inputs to predictions or outputs
Feature	Input variable used by a model
Label	Known answer used during supervised training
Training data	Data used to fit the model
Validation data	Data used during model selection/tuning
Test data	Held-out data used to estimate final model performance
Inference	Using a trained model to make predictions
Algorithm	Method used to train a model
Overfitting	Model memorizes training data and performs poorly on new data
Underfitting	Model is too simple and performs poorly even on training data
Bias in data	Skewed or unrepresentative data that may lead to unfair or inaccurate outcomes
Feature engineering	Selecting or transforming inputs to improve model performance
Hyperparameters	Training settings chosen before or during training, not learned directly from data

Machine Learning Task Decision Table

Task	Prediction/output	Supervision type	Example
Binary classification	One of two classes	Supervised	Fraud or not fraud
Multiclass classification	One of more than two classes	Supervised	Classify support ticket category
Multilabel classification	Multiple labels can apply	Supervised	Tag an image as outdoor, vehicle, daytime
Regression	Numeric value	Supervised	Predict price, revenue, temperature
Clustering	Groups with similar characteristics	Unsupervised	Segment customers into groups
Anomaly detection	Normal vs unusual pattern	Usually unsupervised or semi-supervised	Detect unusual machine telemetry
Forecasting	Future numeric values over time	Supervised time-series	Predict next month’s demand
Ranking/recommendation	Ordered results or suggested items	Often supervised or hybrid	Recommend products

Notes and examples

Quick Distinctions

Distinction	Remember
Classification vs regression	Classification predicts categories; regression predicts numbers
Binary vs multiclass	Binary has two classes; multiclass has three or more
Clustering vs classification	Clustering has no known labels during training
Training vs inference	Training creates the model; inference uses it
Validation vs test data	Validation helps tune; test estimates final performance
Overfitting vs underfitting	Overfitting is too tailored to training data; underfitting is too simple

Model Evaluation Metrics

Classification Confusion Matrix Terms

Term	Meaning
True positive	Predicted positive and actually positive
True negative	Predicted negative and actually negative
False positive	Predicted positive but actually negative
False negative	Predicted negative but actually positive

Notes and examples

\[ \text{Accuracy} = \frac{TP + TN}{TP + TN + FP + FN} \]\[ \text{Precision} = \frac{TP}{TP + FP} \]\[ \text{Recall} = \frac{TP}{TP + FN} \]\[ \text{F1 Score} = 2 \times \frac{\text{Precision} \times \text{Recall}}{\text{Precision} + \text{Recall}} \]

Metric	Use when…	Watch for
Accuracy	Classes are balanced and all errors have similar cost	Misleading with imbalanced data
Precision	False positives are costly	“When the model says yes, how often is it right?”
Recall	False negatives are costly	“How many real positives did the model find?”
F1 score	Need balance between precision and recall	Useful for imbalanced classification
ROC/AUC	Need overall separation ability across thresholds	Higher generally means better separation
MAE	Regression error in original units	Easier to interpret
RMSE	Penalizes large regression errors more	Sensitive to outliers
R-squared	Proportion of variance explained	Can be misleading if used alone

Metric Decision Cues

Scenario	Prefer
Cancer screening: missing a positive case is very bad	High recall
Spam filter: wrongly blocking valid email is very bad	High precision
Fraud detection with rare fraud cases	Precision, recall, F1; not accuracy alone
Predicting sales amount	Regression metrics such as MAE/RMSE
Comparing classification thresholds	ROC/AUC, precision-recall tradeoff

Azure Machine Learning Cheat Sheet

Concept	What it is	Exam cue
Workspace	Top-level Azure Machine Learning resource	Organizes experiments, jobs, models, compute, data
Data asset	Registered dataset or data reference	Reusable training data
Datastore	Connection to storage	Where data is stored
Compute instance	Development workstation in the cloud	Notebooks, interactive development
Compute cluster	Scalable training compute	Runs jobs at scale
Experiment/job	Training or evaluation run	Track metrics and outputs
Environment	Runtime dependencies	Python packages, Docker image, reproducibility
Model	Registered trained model	Deploy for inference
Endpoint	Hosted model access point	Real-time or batch predictions
Automated ML	Tries algorithms/preprocessing automatically	Low-code model creation
Designer	Visual drag-and-drop ML pipelines	No-code/low-code workflow
Responsible AI dashboard	Model insights and fairness/explainability tools	Evaluate model behavior

Notes and examples

ML Lifecycle

    flowchart LR
	    A[Define problem] --> B[Prepare data]
	    B --> C[Train model]
	    C --> D[Evaluate metrics]
	    D --> E{Good enough?}
	    E -- No --> B
	    E -- Yes --> F[Deploy endpoint]
	    F --> G[Monitor performance]
	    G --> H[Retrain when needed]

Azure ML Decision Points

Need	Choose
No-code visual ML pipeline	Designer
Automatically try multiple algorithms	Automated ML
Full control with code	Notebooks / SDK / CLI
Reproducible training environment	Environment
Scale-out training jobs	Compute cluster
Interactive development	Compute instance
Track training metrics	Experiment/job history
Serve predictions to apps	Managed endpoint

Basic ML vocabulary

Term	Meaning
Dataset	Collection of data used for analysis or training
Feature	Input variable used by a model
Label	Target value the model learns to predict
Training	Process of fitting a model to data
Validation	Checking model performance on data not used directly for training
Inference	Using a trained model to make predictions
Model	Learned pattern or function used for prediction
Algorithm	Method used to train a model
Evaluation metric	Measurement of model performance

Supervised vs unsupervised learning

Type	Labels available?	Typical goal	Examples
Supervised learning	Yes	Predict known target values	Classification, regression
Unsupervised learning	No	Discover structure or groups	Clustering
Reinforcement learning	Feedback/rewards	Learn actions through reward signals	Game playing, control systems

AI-900 most often tests classification, regression, and clustering.

Regression, classification, clustering

Task	Predicts	Output example	Scenario clue
Regression	Numeric value	Price, revenue, temperature, delivery time	“How much?” or “How many?”
Binary classification	One of two categories	Fraud/not fraud, pass/fail	Two possible outcomes
Multiclass classification	One of many categories	Product category, ticket type	More than two known classes
Clustering	Group membership without pre-labeled classes	Customer segments	Discover natural groups

Model evaluation basics

Concept	Use
Train/test split	Train on one portion, evaluate on separate data
Accuracy	Proportion of correct predictions; useful but can mislead with imbalanced classes
Precision	Of predicted positives, how many were actually positive
Recall	Of actual positives, how many were found
Confusion matrix	Shows true positives, false positives, true negatives, false negatives
Overfitting	Model performs well on training data but poorly on new data
Underfitting	Model is too simple and performs poorly overall

For classification, the confusion matrix is a frequent conceptual anchor:

Prediction / Actual	Actual positive	Actual negative
Predicted positive	True positive	False positive
Predicted negative	False negative	True negative

Precision vs recall quick rule

If the business priority is…	Favor
Avoiding false positives	Higher precision
Avoiding false negatives	Higher recall

Examples:

Spam filtering: too many legitimate emails marked as spam is a false positive problem.
Disease screening: missing a true disease case is a false negative problem.

Azure Machine Learning

Use Azure Machine Learning when the scenario requires building, training, comparing, managing, or deploying custom machine learning models.

Azure Machine Learning capabilities

Capability	What it does
Automated ML	Tries algorithms and settings to find a good model for a task
Designer	Visual drag-and-drop interface for building ML pipelines
Notebooks	Code-based experimentation, commonly with Python
Compute	Resources used for training or inference
Data assets	Registered datasets used in experiments and pipelines
Experiments/jobs	Track training runs and results
Model registry	Store and manage trained models
Endpoints	Deploy models for real-time or batch inference

Automated ML vs designer vs notebooks

Option	Best fit
Automated ML	You want Azure to test multiple model approaches quickly
Designer	You want a low-code visual pipeline
Notebooks	You want full code control and flexibility

Machine learning traps

Trap	Correction
Choose regression for “predict risk level: low, medium, high”	That is classification because the output is categorical
Choose clustering when labels already exist	If the target label exists, it is supervised learning
Use Azure AI Vision for tabular sales forecasting	Use Azure Machine Learning regression
Assume high training accuracy means success	Check performance on validation/test data to avoid overfitting
Treat automated ML as a prebuilt cognitive service	It is part of Azure Machine Learning for training custom models

Computer Vision Reference

Capability	What it does	Example
Image classification	Assigns one or more labels to an image	“This image contains a dog”
Object detection	Locates objects with bounding boxes	Find vehicles in traffic images
Image tagging	Adds descriptive tags	Outdoor, building, person
Image captioning	Generates a natural language description	“A person riding a bicycle”
OCR / Read	Extracts printed or handwritten text	Read text from a sign or scanned page
Face detection	Detects faces and attributes depending on configuration	Locate faces in an image
Spatial analysis	Understands people movement/presence in spaces	Occupancy or distancing scenarios
Custom vision	Trains a custom classifier or detector	Identify company-specific product defects

Notes and examples

Vision Service Selection

Requirement	Choose
General image analysis using prebuilt models	Azure AI Vision
Extract text from images	Azure AI Vision OCR / Read
Extract fields from forms and documents	Azure AI Document Intelligence
Custom image labels or object detection	Custom vision capabilities
Face-related detection or recognition scenarios	Azure AI Face capabilities
Search image-enriched documents	Azure AI Search with AI enrichment

Vision Traps

Trap	Correct idea
OCR vs Document Intelligence	OCR extracts text; Document Intelligence extracts structured fields
Classification vs object detection	Classification labels the whole image; detection locates objects
Prebuilt vs custom vision	Prebuilt works for common objects; custom uses your labeled data
Image analysis vs generative AI	Vision analyzes images; generative AI creates content or responses

Computer vision review

Computer vision workloads analyze images and visual content.

Key computer vision tasks

Task	Meaning	Example
Image classification	Assigns a class to an image	“This image is a bicycle”
Object detection	Finds and locates objects in an image	Detect boxes around cars
OCR	Reads printed or handwritten text from images	Extract text from a scanned sign
Image analysis	Tags, captions, objects, visual features	Describe image contents
Face detection	Detects human faces and attributes depending on service capability and configuration	Count faces in an image
Spatial analysis concept	Understands people’s movement or presence in spaces	Occupancy or distancing scenarios

Azure services for vision scenarios

Scenario	Best service area
Generate tags or captions for images	Azure AI Vision
Read text from an image	Azure AI Vision OCR
Extract fields from invoices, receipts, forms, IDs	Azure AI Document Intelligence
Train a custom model to classify product images	Custom Vision / custom image classification
Train a custom model to locate defects in images	Custom Vision / custom object detection
Analyze video content	Azure AI Video Indexer may appear in broader Azure AI contexts

OCR vs Document Intelligence

Need	Use
“Read the text in this image”	OCR
“Extract invoice number, vendor, date, line items, and total”	Document Intelligence
“Classify this image as cat/dog/defect/no defect”	Image classification
“Find each hard hat in a photo and draw bounding boxes”	Object detection

Document Intelligence

Capability	What it extracts	Example
Prebuilt document models	Common structured fields	Invoices, receipts, IDs, tax forms
Custom document models	Fields specific to your forms	Internal order forms
Layout extraction	Text, tables, selection marks, structure	Convert scanned form layout to data
Key-value extraction	Named fields and values	Invoice number, total, due date
Table extraction	Rows and columns	Line items on an invoice

If the question asks for…	Choose
“Read text from an image”	OCR
“Extract invoice total, vendor, and line items”	Document Intelligence
“Search thousands of enriched PDFs”	Azure AI Search plus enrichment
“Classify custom product images”	Custom vision capabilities

Natural Language Processing Reference

Capability	Azure service/capability	What it does
Sentiment analysis	Azure AI Language	Positive, negative, neutral sentiment
Opinion mining	Azure AI Language	Sentiment about specific aspects
Key phrase extraction	Azure AI Language	Important phrases in text
Named entity recognition	Azure AI Language	People, locations, organizations, dates, quantities
PII detection	Azure AI Language	Detects personally identifiable information
Language detection	Azure AI Language	Identifies the language of text
Summarization	Azure AI Language or generative AI, depending on scenario	Condenses text
Custom text classification	Azure AI Language	Assigns custom categories
Conversational language understanding	Azure AI Language	Maps utterances to intents/entities
Question answering	Azure AI Language	Answers from a defined knowledge base
Text translation	Azure AI Translator	Translates text between languages

Notes and examples

NLP Decision Cues

Scenario	Choose
“Is this review positive or negative?”	Sentiment analysis
“Find all company and person names”	Named entity recognition
“Detect credit card numbers or emails”	PII detection
“Find the most important words or phrases”	Key phrase extraction
“Determine whether text is English, French, or Spanish”	Language detection
“Route user request to BookFlight intent”	Conversational Language Understanding
“Answer FAQs from a knowledge base”	Question answering
“Translate a document from German to English”	Azure AI Translator
“Generate a new paragraph from a prompt”	Azure OpenAI Service

Natural language processing review

Natural language processing, or NLP, deals with text and language.

Common NLP tasks

Task	Meaning	Example
Language detection	Identifies language	Detect that a review is in Spanish
Sentiment analysis	Determines positive, neutral, or negative tone	“The service was terrible” = negative
Opinion mining	Identifies opinions about specific aspects	Food positive, service negative
Key phrase extraction	Finds main concepts	“battery life,” “screen quality”
Named entity recognition	Finds entities such as people, places, organizations, dates	“Microsoft,” “Seattle,” “June 18”
PII detection	Identifies sensitive personal information	Phone numbers, email addresses
Question answering	Answers natural language questions from a knowledge base	FAQ bot
Conversational language understanding	Interprets user intent and entities in conversations	“Book a flight to Paris”
Summarization	Produces concise summaries of longer text	Summarize a support case

Azure services for language scenarios

Scenario	Think of
Detect sentiment in customer reviews	Azure AI Language
Extract names, dates, places, or organizations	Azure AI Language
Detect or redact personal data in text	Azure AI Language
Build a FAQ-style question answering experience	Azure AI Language question answering
Interpret user intent in a bot	Conversational language understanding
Translate text between languages	Azure AI Translator
Convert speech audio to text	Azure AI Speech
Convert text to spoken audio	Azure AI Speech
Search large document collections	Azure AI Search

Intent vs entity

Concept	Meaning	Example
Intent	What the user wants to do	“BookFlight”
Entity	Specific details needed to complete the intent	destination = Paris, date = Friday

Candidate trap: In “Book a table for two at 7 PM,” the intent is the action, while “two” and “7 PM” are entities.

Speech and Conversational AI

Requirement	Choose	Notes
Convert spoken audio to text	Azure AI Speech - speech to text	Transcription
Convert text to natural-sounding audio	Azure AI Speech - text to speech	Voice synthesis
Translate spoken language	Azure AI Speech translation	Speech input to translated output
Translate written text	Azure AI Translator	Text-only translation
Build bot channel integration	Azure Bot Service	Connect to Teams, web chat, and other channels
Understand user intent in a bot	Conversational Language Understanding	Intents and entities
Answer predefined user questions	Question answering	FAQ/knowledge-base style

Notes and examples

Bot Architecture at a Glance

Layer	Role
Bot application	Orchestrates conversation flow
Channel	Where users interact, such as web chat or Teams
Language understanding	Detects intent and entities
Knowledge base / data	Provides factual answers
Generative model	Creates flexible natural language responses
Human handoff	Escalates when automation is insufficient

Azure AI Search and Knowledge Mining

Azure AI Search is used to make information searchable. It can combine search indexing with AI enrichment.

Concept	Meaning
Data source	Where content comes from, such as storage or a database
Index	Searchable structure containing fields
Indexer	Crawls data and populates an index
Skillset	AI enrichment steps, such as OCR or entity extraction
Enrichment	Adds AI-generated metadata to content
Knowledge store	Stores enriched outputs for downstream use
Semantic search/ranking	Improves relevance using semantic understanding
Vector search	Retrieves content by similarity using embeddings

Notes and examples

Search vs Other AI Services

Requirement	Best match
“Make a large document repository searchable”	Azure AI Search
“Extract entities during indexing”	Azure AI Search skillset with Azure AI Language
“Read text from scanned documents before indexing”	OCR skill in enrichment pipeline
“Extract invoice fields into structured records”	Document Intelligence
“Generate an answer grounded in search results”	Azure AI Search plus Azure OpenAI Service

Azure AI Search and knowledge mining

Azure AI Search is used to index, enrich, and query content. It commonly appears in scenarios involving document collections, enterprise search, or knowledge mining.

Concept	Meaning
Index	Searchable representation of content
Indexer	Automates ingestion from supported data sources
Skillset	Enrichment pipeline that can apply AI skills
Search query	User request against the index
Cognitive enrichment	Extracts text, entities, key phrases, image text, or other information to improve search

Search vs generative answer

Need	Think of
Retrieve relevant documents from a large collection	Azure AI Search
Generate a conversational answer using retrieved content	Generative AI with grounding/retrieval
Extract entities during indexing	Azure AI Search enrichment with AI skills
Build a traditional keyword/vector search experience	Azure AI Search

Generative AI and Azure OpenAI Service

Term	Exam-ready meaning
Generative AI	AI that creates new content such as text, images, code, or summaries
Foundation model	Large pretrained model adaptable to many tasks
Large language model	Model specialized in language understanding and generation
Prompt	User or system input that guides model output
Completion	Model-generated response
Token	Unit of text processed by a model
Context	Information supplied to the model for a request
Grounding	Providing relevant source data to reduce unsupported answers
RAG	Retrieval-augmented generation; retrieve relevant data, then generate
Embedding	Numeric representation of text or content for similarity search
Fine-tuning	Further training a model on task-specific examples
Hallucination	Plausible-sounding but unsupported or incorrect output
Content filter	Control that detects or blocks harmful content
Temperature	Setting that influences randomness/creativity
System message	Instruction that defines assistant behavior or constraints

Notes and examples

Generative AI Service Selection

Scenario	Choose
Build chat or text generation with Microsoft-hosted OpenAI models	Azure OpenAI Service
Explore, build, and manage AI apps and model deployments	Azure AI Foundry
Add enterprise document grounding to chat	Azure AI Search plus Azure OpenAI Service
Compare or select models for AI apps	Azure AI Foundry model catalog
Moderate harmful user or model content	Azure AI Content Safety
Detect sentiment or entities using prebuilt NLP	Azure AI Language
Translate text	Azure AI Translator

Retrieval-Augmented Generation Pattern

    flowchart LR
	    U[User question] --> A[App orchestration]
	    A --> S[Retrieve relevant chunks<br/>Azure AI Search]
	    S --> A
	    A --> P[Prompt with question + retrieved context]
	    P --> M[Azure OpenAI model]
	    M --> R[Answer with grounded context]

Generative AI Traps

Trap	Correct idea
“The model knows all company data automatically”	Private data usually must be supplied through grounding, retrieval, or integration
“Generative AI is always deterministic”	Outputs can vary depending on model settings and prompt
“Fine-tuning is the only way to use private data”	RAG is often used to ground responses without retraining
“A fluent answer is necessarily correct”	Generated content can be unsupported or incorrect
“Content filtering guarantees perfect safety”	Filtering helps reduce risk but does not remove the need for governance
“Embeddings generate final answers”	Embeddings support similarity search; generation uses a generative model

Generative AI fundamentals

Generative AI uses models that can create content such as text, code, summaries, and responses.

Core generative AI terms

Term	Meaning
Large language model	Model trained on large amounts of text to predict and generate language
Prompt	User or system input that guides model output
Completion / response	Model-generated output
Token	Unit of text processed by a model
Grounding	Providing relevant context so output is based on trusted data
Retrieval augmented generation	Retrieving relevant information and providing it to the model as context
System message	Instruction that shapes assistant behavior
User message	The user’s request
Temperature	Setting that influences randomness/creativity
Hallucination	Plausible-sounding but incorrect or unsupported output

Generative AI scenario cues

Scenario clue	Likely concept
“Answer questions using company documents”	Grounding / retrieval augmented generation
“Reduce unsupported responses”	Provide context, use grounding, evaluate outputs
“Make responses more deterministic”	Lower randomness/temperature conceptually
“Create a chatbot that drafts responses”	Generative AI / large language model
“Need content safety controls”	Responsible AI and safety filtering concepts
“Need private organizational data in answers”	Use retrieval/grounding; do not assume the base model already knows it

Prompting basics

Prompting approach	Use
Clear instruction	Tells the model exactly what to do
Context	Supplies facts the model should use
Examples	Shows desired format or style
Constraints	Sets boundaries such as length, tone, or allowed sources
Role instruction	Gives the model a persona or task frame

Poor prompt: “Summarize this.”

Better prompt: “Summarize the following support case in three bullet points: issue, customer impact, and recommended next action. Use only the provided text.”

Azure AI Resource, Security, and Access Basics

Concept	What to know for AI-900
Azure AI services resource	Azure resource that provides access to AI APIs
Single-service resource	Resource scoped to one service, such as Speech or Language
Multi-service resource	One resource that can access multiple Azure AI services
Endpoint	URL used by applications to call the service
Key	Secret credential used to authenticate API calls
Microsoft Entra ID	Identity platform used for role-based access and managed identities
RBAC	Grants users/services permissions to Azure resources
Managed identity	Lets Azure resources authenticate without storing secrets in code
Key Vault	Secure storage for secrets, keys, and certificates
Private networking	Can restrict access paths for sensitive workloads
Monitoring	Track availability, errors, latency, and usage patterns

Notes and examples

Security Decision Cues

Requirement	Prefer
Avoid hard-coded API keys in application code	Managed identity or secure secret storage
Store API keys securely	Azure Key Vault
Grant least-privilege access to Azure resources	RBAC
Track service errors and performance	Azure monitoring/logging tools
Protect sensitive prompts and outputs	Privacy controls, access control, data minimization
Govern generated content risks	Content filtering, human review, Responsible AI practices

Common AI-900 Scenario Cues

Question cue	Likely answer
“Predict whether a loan will default”	Binary classification
“Predict next month’s revenue”	Regression or forecasting
“Group customers by purchasing behavior”	Clustering
“Detect unusual machine behavior”	Anomaly detection
“Identify objects and locations in an image”	Object detection
“Read handwritten text from a scanned page”	OCR / Read
“Extract fields from invoices”	Azure AI Document Intelligence
“Find names and addresses in text”	Named entity recognition / PII detection
“Determine whether feedback is positive”	Sentiment analysis
“Translate support articles”	Azure AI Translator
“Transcribe call center audio”	Azure AI Speech
“Create a chatbot for Teams”	Azure Bot Service plus language/generative capabilities
“Answer from company documents with citations”	RAG with Azure AI Search and Azure OpenAI Service
“Automatically try multiple ML algorithms”	Azure Machine Learning automated ML
“Build ML pipeline visually”	Azure Machine Learning designer
“Need explanation of model behavior”	Responsible AI / model interpretability tools
“Need to reduce biased outcomes”	Fairness
“Need user disclosure that AI is used”	Transparency
“Need protect personal data”	Privacy and security

Final Review Checklist

Before practice questions, make sure you can:

Distinguish AI, ML, deep learning, and generative AI.
Identify classification, regression, clustering, anomaly detection, and forecasting scenarios.
Choose between Azure Machine Learning, Azure AI services, Azure AI Search, and Azure OpenAI Service.
Match vision tasks to image analysis, OCR, object detection, custom vision, and Document Intelligence.
Match language tasks to sentiment, NER, PII detection, translation, CLU, and question answering.
Explain the six Microsoft Responsible AI principles.
Interpret accuracy, precision, recall, F1, MAE, RMSE, and R-squared at a basic level.
Recognize when RAG, grounding, embeddings, content filtering, and prompt design apply.
Avoid service traps: OCR vs Document Intelligence, Translator vs Speech, Search vs model training, classification vs regression.

Next step: use targeted AI-900 practice questions to drill service selection, workload identification, responsible AI scenarios, and metric interpretation under exam-style wording.

Notes and examples

Final quick checklist

Before moving to question-bank practice, confirm that you can:

Identify the difference between regression, classification, and clustering.
Recognize features, labels, training, inference, and evaluation.
Match major AI workloads to the right Azure AI service.
Distinguish OCR from Document Intelligence.
Distinguish Azure AI Language, Speech, Translator, and Search.
Explain the six responsible AI principles in scenario form.
Recognize basic generative AI concepts: prompts, grounding, retrieval, hallucination, and safety.
Avoid choosing Azure Machine Learning when a prebuilt Azure AI service is the better fit.

Next step: use IT Mastery practice with original practice questions, topic drills, mock exams, and detailed explanations to turn this Cheat Sheet into exam-ready recognition.

High-yield AI-900 map

Area	What to know quickly	Common exam angle
AI workloads	Prediction, classification, anomaly detection, computer vision, NLP, document intelligence, generative AI	Identify the workload from a business scenario
Responsible AI	Fairness, reliability and safety, privacy and security, inclusiveness, transparency, accountability	Choose the principle being violated or improved
Machine learning	Features, labels, training, validation, evaluation, regression, classification, clustering	Match the ML type to the problem
Azure Machine Learning	Automated ML, designer, compute, data, model training, deployment	Know when to use Azure ML instead of prebuilt AI services
Computer vision	Image analysis, OCR, object detection, image classification, face detection, spatial analysis concepts	Choose Azure AI Vision, Custom Vision, or Document Intelligence
Natural language processing	Sentiment, key phrases, entities, language detection, summarization concepts, question answering, conversational language understanding	Match text scenario to Azure AI Language, Speech, Translator, or Azure AI Search
Speech and translation	Speech-to-text, text-to-speech, speech translation, text translation	Distinguish Speech from Translator
Generative AI	Prompts, completions, grounding, copilots, large language models, responsible use	Understand use cases and risk controls

Core decision rule: service before feature

Many AI-900 questions are service-selection questions in disguise. Read the scenario, identify the data type, then choose the service.

If the input is…	And the goal is…	Think first of…
Tabular data	Predict a numeric value	Regression in Azure Machine Learning
Tabular data	Predict a category	Classification in Azure Machine Learning
Tabular data	Group similar records without known labels	Clustering in Azure Machine Learning
Images	Analyze objects, tags, captions, OCR	Azure AI Vision
Images	Train a custom image classifier or detector	Custom Vision / custom vision capabilities
Scanned forms, receipts, invoices, IDs	Extract structured fields from documents	Azure AI Document Intelligence
Text	Sentiment, key phrases, entities, language, PII	Azure AI Language
Text	Search across indexed enterprise content	Azure AI Search
Speech audio	Convert speech to text or text to speech	Azure AI Speech
Text in one language	Translate to another language	Azure AI Translator
Chat, content generation, summarization, code/text generation	Use large language models	Azure OpenAI / Azure AI generative AI capabilities

AI workloads and terminology

AI workload types

Workload	Meaning	Example
Machine learning	Learns patterns from data to make predictions or decisions	Predict customer churn
Computer vision	Interprets visual content	Detect products in shelf images
Natural language processing	Understands or generates human language	Classify support ticket sentiment
Document intelligence	Extracts structured data from forms and documents	Pull invoice number and total from PDFs
Knowledge mining / search	Extracts and indexes information for discovery	Search across manuals and PDFs
Generative AI	Creates text, images, code, summaries, or responses	Draft a response using a company knowledge base

Notes and examples

Common candidate traps

Trap	Correct thinking
“AI” always means machine learning from scratch	Many Azure AI services are prebuilt and require little or no model training
OCR and document extraction are the same thing	OCR reads text; Document Intelligence extracts structured fields from documents
Translation belongs to Language	Text translation is Azure AI Translator; speech translation uses Azure AI Speech capabilities
Search is the same as question answering	Azure AI Search indexes and retrieves content; question answering provides natural language answers from knowledge sources
Generative AI always has current or private knowledge	Large language models need grounding, retrieval, or provided context for specific/private facts

Speech and translation

Speech tasks

Task	Description
Speech-to-text	Transcribes spoken audio into text
Text-to-speech	Produces spoken audio from text
Speech translation	Translates spoken input into another language
Speaker-related features	May involve recognizing or distinguishing speakers depending on scenario and service capability

Translator vs Speech

Scenario	Service choice
Translate a written document from French to English	Azure AI Translator
Transcribe a meeting recording	Azure AI Speech
Read website text aloud	Azure AI Speech text-to-speech
Translate a spoken phrase into another spoken language	Azure AI Speech translation capabilities

High-yield comparison table

Do not confuse	Difference
Classification vs regression	Classification predicts categories; regression predicts numbers
Classification vs clustering	Classification uses known labels; clustering discovers groups
OCR vs Document Intelligence	OCR extracts text; Document Intelligence extracts structured fields
Azure AI Language vs Translator	Language analyzes text; Translator translates text
Speech vs Translator	Speech handles audio; Translator handles text translation
Azure AI Search vs Azure AI Language	Search retrieves/indexes content; Language analyzes text
Azure Machine Learning vs Azure AI services	Azure ML builds custom models; Azure AI services provide prebuilt/customizable AI APIs
Question answering vs conversational language understanding	Question answering responds from knowledge; CLU detects intents/entities
Object detection vs image classification	Object detection locates objects; classification labels the whole image
Grounding vs training	Grounding supplies context at use time; training changes or creates a model

Scenario drills to review before practice

Use these quick examples to test recognition.

Scenario	Best answer pattern
Predict next month’s sales amount from historical sales data	Regression
Predict whether a transaction is fraudulent	Binary classification
Group customers by purchasing behavior without existing segment labels	Clustering
Extract totals and vendor names from invoices	Azure AI Document Intelligence
Read license plate text from an image	OCR / Azure AI Vision
Detect whether product photos contain a damaged item and locate the damage	Object detection/custom vision
Determine if customer comments are positive or negative	Sentiment analysis / Azure AI Language
Extract company names and dates from news articles	Named entity recognition / Azure AI Language
Translate website text into German	Azure AI Translator
Transcribe call-center recordings	Azure AI Speech
Build a searchable index of PDFs and images	Azure AI Search with enrichment concepts
Build a chatbot that answers using internal policy documents	Generative AI with grounding/retrieval
Make AI output easier for users to understand and challenge	Transparency and accountability
Reduce demographic bias in model outcomes	Fairness

Common AI-900 mistakes

Reading mistakes

Choosing the service from a keyword instead of the actual workload.
Missing whether the output is numeric or categorical.
Ignoring whether labels are available.
Treating scanned forms as generic images instead of structured documents.
Treating audio translation as text translation.
Assuming all chatbots are generative AI; some use intent recognition or question answering.

Concept mistakes

Thinking a feature is the same as a label.
Assuming unsupervised learning predicts known target values.
Confusing false positives with false negatives.
Believing accuracy is always the best metric.
Assuming a large language model is always correct.
Thinking responsible AI is only about privacy.

Azure service mistakes

Selecting Azure Machine Learning when a prebuilt Azure AI service directly solves the scenario.
Selecting Azure AI Vision for invoice field extraction instead of Document Intelligence.
Selecting Azure AI Language for speech transcription instead of Speech.
Selecting Translator for sentiment analysis instead of Language.
Selecting Search for sentiment analysis instead of Language.

Fast review: responsible AI mini-scenarios

Scenario	Principle
A hiring model disadvantages applicants from a certain group	Fairness
An AI system must be tested before being used in safety-critical workflows	Reliability and safety
Customer personal data must be protected and access controlled	Privacy and security
A service should work for users with visual, hearing, or motor impairments	Inclusiveness
Users need to know they are interacting with AI and understand limitations	Transparency
A company assigns people to monitor and approve AI use	Accountability

Notes and examples

Fast review: ML mini-scenarios

Scenario	ML type
Predict home price	Regression
Predict loan approval: approve or reject	Binary classification
Predict support ticket category: billing, technical, account, other	Multiclass classification
Group customers into unknown segments	Clustering
Forecast demand as a number	Regression
Identify whether an email is spam	Binary classification

Fast review: Azure AI services

Service area	Remember it for
Azure Machine Learning	Custom ML model lifecycle
Azure AI Vision	Image analysis and OCR
Azure AI Document Intelligence	Structured extraction from forms and documents
Azure AI Language	Text analytics, entities, sentiment, PII, question answering, conversational language understanding
Azure AI Speech	Speech-to-text, text-to-speech, speech translation
Azure AI Translator	Text translation
Azure AI Search	Indexing, retrieval, knowledge mining
Azure OpenAI / generative AI capabilities	Large language model apps, prompts, completions, copilots, grounded generation

Practice strategy for AI-900

For the final stretch, practice by topic rather than only taking full-length mock exams.

Start with service-selection drills. These build fast recognition for Azure AI scenarios.
Drill ML task types. Be automatic on regression, classification, and clustering.
Review responsible AI scenarios. These are often straightforward if you know the principle names.
Practice computer vision and NLP comparisons. Many misses come from choosing the nearby wrong service.
Use detailed explanations. After each question, explain why the correct answer is right and why each distractor is wrong.

Put the review into practice

Study Plan

AI-900 — Microsoft Azure AI Fundamentals Cheat Sheet

High-Yield Azure AI Service Selection

Azure AI service selection workflow

AI Workload Types

Responsible AI Principles

Responsible AI Traps

Responsible AI principles

Responsible AI exam cues

Core AI and ML Terms

Machine Learning Task Decision Table

Quick Distinctions

Model Evaluation Metrics

Classification Confusion Matrix Terms

Metric Decision Cues

Azure Machine Learning Cheat Sheet

ML Lifecycle

Azure ML Decision Points

Basic ML vocabulary

Supervised vs unsupervised learning

Regression, classification, clustering

Model evaluation basics

Precision vs recall quick rule

Azure Machine Learning

Azure Machine Learning capabilities

Automated ML vs designer vs notebooks

Machine learning traps

Computer Vision Reference

Vision Service Selection

Vision Traps

Computer vision review

Key computer vision tasks

Azure services for vision scenarios

OCR vs Document Intelligence

Document Intelligence

Natural Language Processing Reference

NLP Decision Cues

Natural language processing review

Common NLP tasks

Azure services for language scenarios

Intent vs entity

Speech and Conversational AI

Bot Architecture at a Glance

Azure AI Search and Knowledge Mining

Search vs Other AI Services

Azure AI Search and knowledge mining

Search vs generative answer

Generative AI and Azure OpenAI Service

Generative AI Service Selection

Retrieval-Augmented Generation Pattern

Generative AI Traps

Generative AI fundamentals

Core generative AI terms

Generative AI scenario cues

Prompting basics

Azure AI Resource, Security, and Access Basics

Security Decision Cues

Common AI-900 Scenario Cues

Final Review Checklist

Final quick checklist

High-yield AI-900 map

Core decision rule: service before feature

AI workloads and terminology

AI workload types

Common candidate traps

Speech and translation

Speech tasks

Translator vs Speech

High-yield comparison table

Scenario drills to review before practice

Common AI-900 mistakes

Reading mistakes

Concept mistakes

Azure service mistakes

Fast review: responsible AI mini-scenarios

Fast review: ML mini-scenarios

Fast review: Azure AI services

Practice strategy for AI-900

Put the review into practice

Browse Certification Practice Tests