Select the Appropriate Microsoft Foundry Service

The words you must know before anything else

These terms appear in almost every exam question and every code example across all 6 parts. Learn them here once and you'll recognise them everywhere.

What is Microsoft Foundry and why does it matter?

Before 2024, Azure had dozens of separate AI services each with their own portal, their own SDKs, and their own way of doing things. Microsoft Foundry is the answer to that complexity — a single platform that unifies all of them.

How Foundry is organised

Foundry has a hierarchy you need to know for the exam:

Every Azure AI service you need to know

The exam requires you to choose the right service for a given scenario. Here is every service, what it does in one sentence, and its old name (the exam may use either name — this trips up many candidates).

The decision framework for exam questions

Most exam scenario questions follow a pattern: a company has a problem, you pick the service. Use this framework to work through any scenario systematically, starting with the type of input.

Quick scenario practice

Read each scenario and think about which service you'd choose before revealing the answer.

Free tier vs Standard tier — F0 and S0

When you create an Azure AI resource, you choose a pricing tier. The exam tests whether you know when to use each tier and what the limitations are.

Two Python concepts before any code

You don't need to be a Python developer to pass the AI-102 exam. The exam tests whether you can read Python code and understand what it does. Every code block in this course will have every line explained. But first, two concepts that appear in every example:

Concept 1 — Importing a library

A library (also called a package or module) is a collection of pre-written Python code you can use in your program. Azure provides libraries for each AI service. Before you can use a library, you import it — like taking a tool out of a toolbox before using it.

# The "from" keyword means: "from this library..."
# The "import" keyword means: "...bring me this specific tool"
from azure.ai.textanalytics import TextAnalyticsClient

# Now we can use TextAnalyticsClient in our code
# Think of it like: "I've taken the TextAnalyticsClient tool out of the azure.ai.textanalytics toolbox"

Concept 2 — Calling a function

A function is a named block of code that does a specific task. You call it by writing its name followed by brackets. You can pass information into brackets (called arguments). The function does its work and often returns a result.

# This creates a "client" object — a connection to your Azure AI service
# TextAnalyticsClient is the function name
# endpoint= and credential= are named arguments (information you're passing in)
client = TextAnalyticsClient(endpoint=endpoint, credential=credential)

# This calls the "analyze_sentiment" function on our client
# We pass in a list of texts to analyse
# The result is stored in the variable "result"
result = client.analyze_sentiment(documents=["I love this product!"])

How to install the SDK and make your first call

This walks through a complete, working Python example. Even if you don't run it now, reading through it will make every other code example in the course feel familiar.

Step 1 — Install the Azure AI Language SDK

An SDK must be installed before you can import it. You install Python packages using a tool called pip — run this in your terminal or command prompt:

# pip is Python's package installer — it downloads the library from the internet
# azure-ai-textanalytics is the SDK for Azure AI Language service
pip install azure-ai-textanalytics

Step 2 — The complete Python script

The example below calls the Azure AI Language service to detect the language of a piece of text. Read every comment carefully.

# ── Import the tools we need from the Azure SDK ──────────────

# TextAnalyticsClient is the main object we use to call the Language service
from azure.ai.textanalytics import TextAnalyticsClient

# AzureKeyCredential wraps our API key so the SDK knows how to send it securely
from azure.core.credentials import AzureKeyCredential


# ── Your service details ──────────────────────────────────────

# Your endpoint URL — copy this from Azure portal → your resource → Keys and Endpoint
# Replace the placeholder with your actual endpoint
endpoint = "https://YOUR-RESOURCE-NAME.cognitiveservices.azure.com/"

# Your API key — also from Keys and Endpoint in the portal
# In production code you would load this from an environment variable, NOT hard-code it
key = "YOUR-API-KEY-HERE"


# ── Create the credential object ─────────────────────────────

# AzureKeyCredential wraps your key into a format the SDK understands
# Think of it as putting your key card into a card holder before swiping
credential = AzureKeyCredential(key)


# ── Create the client ─────────────────────────────────────────

# TextAnalyticsClient is your connection to the Language service
# You pass your endpoint and credential so it knows where to connect and how to prove your identity
client = TextAnalyticsClient(endpoint=endpoint, credential=credential)


# ── Call the service ──────────────────────────────────────────

# detect_language() takes a list of text documents
# Even if you only have one text, you still wrap it in a list: ["your text here"]
result = client.detect_language(documents=["Bonjour, comment ça va?"])


# ── Read the response ─────────────────────────────────────────

# result is a list — one item per document we sent
# result[0] is the response for our first (and only) document
doc = result[0]

# .primary_language is a property on the response object
# .name is the full language name (e.g. "French")
# .confidence_score is how sure the service is, between 0.0 and 1.0
print(f"Detected language: {doc.primary_language.name}")
print(f"Confidence: {doc.primary_language.confidence_score}")

# Expected output:
# Detected language: French
# Confidence: 1.0

What the exam specifically tests in this topic

Test your understanding

5 questions. Select an answer to see the explanation immediately.