Skip to main content

Challenge 04: Azure AI Services Overview

Estimated Time

25-35 min | Cost: Free | Domain: AI Workloads & Responsible AI (15-20%)

Exam skills covered

  • Map use cases to Azure AI services (Vision, Language, Speech, Decision, OpenAI)
  • Describe the difference between multi-service and single-service resources
  • Identify Azure AI services pricing tiers and endpoints
  • Understand keys, endpoints, and authentication for Azure AI services

Overview

Azure AI services are cloud-based APIs that let developers add AI capabilities to applications without building machine learning models from scratch. They are pre-trained, ready to use, and accessed via REST APIs or SDKs. You don't need to be a data scientist — just call the API and get intelligent results back.

Think of Azure AI services like a restaurant kitchen. You don't need to know how to cook (build ML models) — you just order from the menu (call the API) and get a finished dish (AI-powered result). The kitchen (Microsoft's infrastructure) handles all the complexity behind the scenes.

Azure offers two ways to provision these services: a multi-service resource (one resource for Vision, Language, Speech, etc. with a single key/endpoint) or single-service resources (separate resources for each service). The multi-service approach is simpler for getting started, while single-service resources allow granular billing and access control.

Explore

Task 1: Understand the Azure AI services taxonomy

Service categoryServices includedExample use cases
Azure AI VisionImage Analysis, Custom Vision, FaceDescribe images, detect objects, read text in images
Azure AI LanguageText Analytics, QnA, CLU, TranslatorSentiment analysis, entity recognition, translation
Azure AI SpeechSpeech-to-Text, Text-to-Speech, TranslationTranscription, voice assistants, real-time translation
Azure AI Document IntelligencePrebuilt models, Custom modelsInvoice processing, receipt scanning, ID extraction
Azure OpenAI ServiceGPT-4, DALL-E, Whisper, EmbeddingsContent generation, summarization, code assistance
Azure AI SearchFull-text search, Vector search, AI enrichmentKnowledge mining, RAG patterns, enterprise search

Task 2: Create a multi-service resource (Free tier)

  1. Open portal.azure.com
  2. Click + Create a resource
  3. Search for "Azure AI services"
  4. Select Azure AI services (the multi-service option)
  5. Configure:
    • Subscription: Your subscription
    • Resource group: Create new → rg-ai900-lab
    • Region: Choose one near you
    • Name: ai900-demo-[yourinitials]
    • Pricing tier: Free F0
  6. Review the Responsible AI notice — you must acknowledge it
  7. Click Review + createCreate

Task 3: Explore keys and endpoint

After deployment:

  1. Go to your new Azure AI services resource
  2. Click Keys and Endpoint in the left menu
  3. Notice:
    • Key 1 and Key 2 — used for authentication (two keys for rotation without downtime)
    • Endpoint — the URL your applications call (e.g., https://eastus.api.cognitive.microsoft.com/)
    • Location/Region — where your resource is hosted
  4. These three pieces of information (key + endpoint + region) are what applications need to use the service

Task 4: Multi-service vs single-service resources

FeatureMulti-service resourceSingle-service resource
BillingSingle bill for all servicesSeparate bill per service
KeysOne key accesses all servicesUnique key per service
EndpointSingle endpointSeparate endpoints
Access controlSame permissions for allGranular RBAC per service
Best forGetting started, prototypingProduction with strict access control
Azure CLI Alternative
# Create a multi-service Azure AI resource (Free tier)
az cognitiveservices account create \
--name ai900-demo \
--resource-group rg-ai900-lab \
--kind AIServices \
--sku F0 \
--location eastus

# List keys
az cognitiveservices account keys list \
--name ai900-demo \
--resource-group rg-ai900-lab

# Show endpoint
az cognitiveservices account show \
--name ai900-demo \
--resource-group rg-ai900-lab \
--query "properties.endpoint"

Key Concepts

ConceptDefinition
Azure AI servicesPre-built AI models accessible via REST APIs — no ML expertise needed
Multi-service resourceSingle Azure resource that provides access to multiple AI services with one key/endpoint
Single-service resourceDedicated Azure resource for one specific AI service (e.g., Vision only)
API keySecret string used to authenticate requests to Azure AI services
EndpointURL that applications call to access the AI service
Free tier (F0)No-cost pricing tier with limited transactions per month — ideal for learning
Standard tier (S0)Pay-per-use pricing with higher limits for production workloads
RegionAzure datacenter location where the AI service is hosted and processes data

Common Misconceptions

MisconceptionReality
"You need machine learning expertise to use Azure AI services"Azure AI services are pre-trained APIs. You call them with data and get results — no ML knowledge needed
"Multi-service and single-service resources have different capabilities"The AI capabilities are identical. The difference is only in billing, access control, and resource management
"Azure OpenAI is the same as OpenAI's public API"Azure OpenAI runs OpenAI models on Azure infrastructure with enterprise security, compliance, and regional data residency. It requires separate approval to access
"The free tier is limited to a short trial period"The F0 (Free) tier is not time-limited. It has transaction limits per month but doesn't expire
"You need two keys because one might stop working"Two keys exist to enable key rotation without downtime. While you regenerate Key 1, apps use Key 2, and vice versa

Knowledge Check

1. A developer wants a single Azure resource that provides access to Vision, Language, and Speech AI services with one set of credentials. What should they create?

2. What are the minimum pieces of information an application needs to call an Azure AI service?

3. Why does Azure provide TWO keys for each AI services resource?

4. A company needs to process receipts and extract purchase amounts, dates, and merchant names automatically. Which Azure AI service should they use?

5. What is the primary advantage of using single-service resources instead of a multi-service resource?

Learn More