AI-103 Develop AI Apps and Agents on Azure Study Cram

John Savill provides a study-cram style walkthrough for the Microsoft AI-103 exam (Develop AI Apps and Agents on Azure), focusing on the core Azure AI Foundry building blocks and how they map to real implementation tasks.

Overview

This video is structured as an end-to-end review of the AI-103 objectives, moving from foundational AI terminology and responsible AI concepts into Azure AI Foundry resources, model choices, deployment patterns, and agent development.

Topics covered

Microsoft learning resources

• Points to official Microsoft Learn resources for the AI-103 certification.

Core AI terms

• Reviews foundational terminology used throughout the exam and Azure AI services.

Responsible AI

• Covers responsible AI concepts and considerations relevant to building solutions on Azure.

Azure AI Foundry resource and project

• Introduces the Azure AI Foundry resource and project concepts.

Models

• Reviews model concepts and how models are selected/used in Foundry.

Model deployment options

• Discusses different ways to deploy models for use in applications.

Using playgrounds to experiment

• Covers using playground experiences to test and iterate.

Model routing

• Reviews model routing concepts and how routing can be tested.

Evaluations

• Covers evaluation concepts for assessing model/app behavior.

Using APIs and SDKs

• Discusses interacting with services via APIs and SDKs.

Responses API and Chat Completions

• Covers the Responses API.
• Covers Chat Completions.
• Includes a comparison segment.

OpenAI vs Foundry SDK

• Compares using OpenAI-style SDK approaches versus Foundry SDK approaches.

Tools (including OpenAI responses tools)

• Reviews tool concepts and how tools are used with responses.

Foundry services and Foundry tools

• Reviews the services and tooling available in Azure AI Foundry.

Skills, toolbox, and knowledge

• Covers skills.
• Covers the toolbox concept.
• Covers knowledge concepts.

Optimizing the solution

• Reviews optimization considerations.

Guardrails

• Covers guardrails for safer/more controlled solutions.

Bringing it together

• Summarizes how the pieces fit into a complete solution.

Microsoft Agent Framework and agents

• Introduces Microsoft Agent Framework.
• Covers agent concepts.

Workflows and agent harness

• Discusses workflows.
• Covers an agent harness concept.

Azure AI Foundry Agent Service

• Covers Foundry Agent Service.
• Discusses prompt agents and hosted agents.
• Covers publishing agents.

Additional AI services

• Language
• Speech
• Translation
• Vision

Using generative AI

• Generating images
• Generating videos

Viewing metrics

• Covers metrics/monitoring at a high level.

Azure Content Understanding

• Mentions Azure Content Understanding as part of the broader solution space.

Key links from the description

• Whiteboard: https://github.com/johnthebrit/CertificationMaterials/raw/main/whiteboards/AI-103-Whiteboard.png
• Repo examples: https://github.com/johnthebrit/RandomStuff/tree/master/AI
• Microsoft Learn certification page: https://learn.microsoft.com/credentials/certifications/azure-ai-apps-and-agents-developer-associate/?practice-assessment-type=certification
• Model router docs: https://learn.microsoft.com/azure/foundry/openai/how-to/model-router?tabs=foundry-responses#test-model-router-with-foundry-responses-and-chat-completions
• Foundry tools/toolbox docs: https://learn.microsoft.com/azure/foundry/agents/how-to/tools/toolbox?pivots=python#feature-support