AB-100: Agentic AI Business Solutions Architect

O'Reilly Live Learning Course | 4 Hours | Microsoft AB-100 Exam Prep + Real-World Agentic AI

Design, deploy, and govern multi-agent AI business solutions across the Microsoft stack — and walk in prepared to pass Microsoft's AB-100 certification exam on the first attempt.

This repo has two sides:

1. Live training scaffolding — hour-by-hour course material in src/ and the plan in docs/.
2. AB-100 Cert Buddy — a GitHub Copilot agent in .github/ that generates exam-realistic practice questions, short architect-grade labs, and personalized study plans, all grounded in Microsoft Learn via MCP. Open the repo in VS Code and invoke @ab100-cert-buddy-agent in Copilot Chat. See the Cert Buddy quick start below.

Course overview

Prerequisites

• To be finalized — see docs/course-plan.md

Repository structure

ab100/
├── README.md                 # This file
├── CLAUDE.md                 # Claude Code instructions
├── CODE_OF_CONDUCT.md        # Community guidelines
├── contributing.md           # Contribution guide
├── SECURITY.md               # Security policy
├── LICENSE                   # MIT License
├── .markdownlint.json        # Markdown linting config
├── .editorconfig             # Editor formatting consistency
│
├── .github/                  # Cert Buddy agent + skills + prompts
│   ├── agents/
│   │   └── ab100-cert-buddy-agent.agent.md
│   ├── skills/
│   │   ├── ab100-item-creator/SKILL.md
│   │   ├── ab100-lab-creator/SKILL.md
│   │   └── ab100-study-planner/SKILL.md
│   ├── prompts/
│   │   ├── ab100-practice-question.prompt.md
│   │   ├── ab100-practice-lab.prompt.md
│   │   └── ab100-study-planner.prompt.md
│   ├── copilot-instructions.md
│   └── workflows/validate.yml
│
├── .vscode/                  # MCP server + recommended extensions
│   ├── mcp.json
│   └── extensions.json
│
├── docs/                     # Course plan + reference notes
│   └── course-plan.md
│
├── references/               # Cert Buddy grounding data
│   ├── ab100-objectives.md   # AB-100 skills measured (synced from Learn)
│   ├── fictional-companies.md
│   └── style-guide.md
│
├── images/                   # Course images and assets
├── scripts/                  # Setup helpers (if needed)
└── src/                      # Per-hour teaching materials
    ├── hour1-plan/           # Plan AI Solutions
    ├── hour2-design/         # Design AI Solutions
    ├── hour3-deploy/         # Deploy AI Solutions
    └── hour4-security/       # Security & Exam Mastery

Quick start

1. Clone the repository

   git clone https://github.com/timothywarner-org/ab100.git
   

2. Hour 1 — Plan AI Solutions: src/hour1-plan/

3. Hour 2 — Design AI Solutions: src/hour2-design/

4. Hour 3 — Deploy AI Solutions: src/hour3-deploy/

5. Hour 4 — Security & Exam Mastery: src/hour4-security/

Hour summaries

Hour 1: Plan AI Solutions

Theme: TBD before delivery

Maps to AB-100 Domain 1 (25–30%). See src/hour1-plan/.

Hour 2: Design AI Solutions

Theme: TBD before delivery

Maps to AB-100 Domain 2 (25–30%). See src/hour2-design/.

Hour 3: Deploy AI Solutions

Theme: TBD before delivery

Maps to AB-100 Domain 3 implementation & operations (~20–25%). See src/hour3-deploy/.

Hour 4: Security & Exam Mastery

Theme: TBD before delivery

Maps to AB-100 Domain 3 security & governance (~20–25%), plus exam readiness. See src/hour4-security/.

Course schedule

Full plan in docs/course-plan.md.

Instructor

Tim Warner — Microsoft MVP (Azure AI and Cloud/Datacenter Management), Microsoft Certified Trainer

• LinkedIn
• Website
• O'Reilly Author Page

AB-100 Cert Buddy

An AI-powered study companion for the Microsoft AB-100: Agentic AI Business Solutions Architect certification, built entirely on GitHub Copilot agents. No application code — the buddy lives in .github/ as agent definitions, skill specs, prompt templates, and an MCP server configuration that turns GitHub Copilot Chat into an interactive exam coach.

The agent generates exam-realistic practice questions with two-phase interactive evaluation, short hands-on labs (15–25 minutes) across agent authoring, design walkthroughs, ALM, and governance, and personalized study plans based on your self-assessed confidence across the three AB-100 domains. All content is original and grounded in Microsoft Learn via MCP.

Cert Buddy prerequisites

No API keys, Node.js, or Python are required. The Microsoft Learn MCP server is a free hosted service with no sign-up.

Cert Buddy quick start

1. Open the folder in VS Code:

   code ab100
   

2. The Microsoft Learn MCP server is defined in .vscode/mcp.json and auto-configures when the workspace loads.

3. Open GitHub Copilot Chat and invoke the agent:

   @ab100-cert-buddy-agent Generate 5 practice questions on Copilot Studio extensibility with MCP
   

   Or use the slash commands. Type / in Copilot Chat:

   /ab100-practice-question
   /ab100-practice-lab
   /ab100-study-planner
   

Cert Buddy skills

Cert Buddy MCP server

Cert Buddy key rules

• Grounded in Microsoft Learn. Every question and lab is grounded in official Learn docs via MCP before any other source.
• Current terminology only. Legacy names like "Azure AI Studio" and "Power Virtual Agents" are silently replaced with "Microsoft Foundry" and "Microsoft Copilot Studio." See the full rename table in .github/copilot-instructions.md.
• Original content only. The agent never recreates, paraphrases, or references real exam questions, braindumps, or leaked content.
• No contractions. No trick wording. Distractors must be real.
• Answer randomization. Correct answer position is distributed across A/B/C/D.
• Company randomization. Scenarios draw from references/fictional-companies.md — not always Contoso.
• Labs include cleanup. Every lab ends with exact rollback or deletion steps.
• Microsoft style. All content follows the Microsoft Writing Style Guide (references/style-guide.md).

Disclaimer

This is an unofficial study guide. Always refer to the official Microsoft exam page for the most current information.

License

MIT License. See LICENSE for details.

Questions? Open an issue or reach out via the course chat during live sessions.