Session 1: The AI Landscape & Setup

Start from zero: what AI, machine learning, deep learning, generative AI and agentic AI really mean, where they are used, what the AI-901 exam looks like, and how to set up Azure and Python so every lab in this course runs.

Course map. Module 1 (Sessions 1–2) builds concepts and responsible AI. Module 2 (Sessions 3–4) maps AI workloads. Modules 3–6 (Sessions 5–10) build real apps in Microsoft Foundry. Modules 7–8 (Sessions 11–14) are revision, practice and three mock exams. Every module ends with a hands-on lab.

AI, machine learning and deep learning

Think of three circles, one inside the other.

A model is the learned result: a function that takes an input (an image, a sentence, a row of data) and produces an output (a label, a number, new text). Training creates the model; inference is using it.

Classic ML tasks you should recognise:

• Regression: predict a number (house price, monthly sales).
• Classification: predict a category (fraud / not fraud, spam / not spam).
• Clustering: group similar items without labels (customer segments).

Generative AI and agentic AI: what changed

Generative AI models create new content (text, images, code, audio) instead of only labelling existing content. They are built on very large neural networks called foundation models; the text ones are large language models (LLMs). You talk to them with a prompt.

Agentic AI goes one step further. An agent is a model plus instructions, tools and knowledge, so it can plan, call tools (search a document, run code, call an API) and complete a multi-step task, not just reply once.

Why it matters: one general model can now do summarisation, translation, extraction and Q&A that used to need a separately trained model for each task.

Real-world use cases

• BFSI: KYC document extraction, fraud alerts explained in plain language, customer-support assistants, call-centre speech analytics.
• Healthcare: summarising clinical notes, transcribing doctor–patient conversations, triage chatbots (with strict human oversight).
• Retail: product descriptions, visual search, review sentiment, demand forecasting.
• Education: personalised tutors, auto-generated quizzes, speech-to-text for accessibility.

The AI-901 exam at a glance

🎯 Exam tip. Domain 2 is the bigger half and it is hands-on: expect questions about which SDK call, which portal page, or which Foundry tool to use. Doing the labs is the best preparation.

Azure basics you need on day one

• Microsoft Entra ID tenant: your organisation’s identity directory (users, groups, sign-in).
• Subscription: the billing container. Everything you create is charged to a subscription.
• Resource group: a folder for related resources. Delete the resource group and everything inside goes with it, which is the easiest way to clean up after a lab.
• Region: the data-centre location (for example Sweden Central, East US 2, Central India). Not every model is available in every region, so check the model catalog before choosing.
• Resource: one service instance. For this course the key one is the Microsoft Foundry resource, which gives you a project to deploy models and build agents, plus access to Foundry Tools (Azure Language, Azure Speech, Azure Content Understanding and more).

Pricing awareness. Most AI services are pay-as-you-go: you pay per token (text models), per minute of audio, per image or per page. A new Azure free account includes starting credit for the first 30 days, and Azure for Students gives credit without a credit card. Set a budget alert in Cost Management, and delete lab resource groups when you finish.

Naming you will see. Microsoft Foundry was previously called Azure AI Foundry, and before that Azure AI Studio. Foundry Tools is the new name for what used to be called Azure AI services (and before that, Cognitive Services). The exam uses the new names.

Set up your Python environment

You need Python 3.10 or later, VS Code, and the Azure CLI (for keyless sign-in).

# 1) create and activate a virtual environment
python -m venv .venv
# Windows:
.venv\Scripts\activate
# macOS / Linux:
source .venv/bin/activate

# 2) install the packages used across the course
pip install azure-ai-projects azure-identity openai python-dotenv

# 3) sign in so DefaultAzureCredential can use your account
az login

Keep secrets out of your code with a .env file in the project folder:

# .env  (never commit this file: add it to .gitignore)
FOUNDRY_PROJECT_ENDPOINT=https://<your-resource>.services.ai.azure.com/api/projects/<your-project>
FOUNDRY_MODEL_NAME=<your-deployment-name>

import os
from dotenv import load_dotenv

load_dotenv()                                   # reads .env into environment variables
endpoint = os.environ["FOUNDRY_PROJECT_ENDPOINT"]
print("Endpoint loaded:", endpoint[:40] + "...")

API keys vs keyless authentication

Check yourself