Agent Journey: Milestone 1

What You'll Learn

The AI Driven Development Methodology is structured, repeatable, and aligned with how professional teams work. Instead of reactive prompting, you'll define intent (PRD), structure work (PLAN), and assign tasks (TASK prompts) to GitHub Copilot agents.

The Four Phases

1. PRD (Product Requirements Document) — Define what you want to build. Scope, constraints, success criteria. This is what the AI agent will reason about.
2. PLAN (Implementation Plan) — Break the PRD into structured steps. What builds first? What depends on what? This is your architectural map.
3. TASK (Agent Task Prompt) — Give the agent one precise instruction per task. Include context, constraints, and verification rules. This is how you maintain control.
4. IMPLEMENTATION (Build, Verify and Decide) — Execute the bounded TASK, inspect the change, run the relevant tests and present evidence for human review. Before anything merges, check scope, conventions, security and edge cases using the included checklist.

The human stays the architect. The AI becomes the builder. Never merge code without human review.

This Starter Kit now includes all four phases — PRD, PLAN, and TASK templates plus an IMPLEMENTATION-checklist.md for the human review gate. See how the review works on the IMPLEMENTATION phase page.

How to Start

1. Download the Starter Kit (four resources, 100% free)
2. Copy the templates into your project repository root
3. Fill in the PRD for your current feature or project
4. Ask Copilot to propose a PLAN — supply the approved PRD, then review and approve the approach before delegating work
5. Use TASK prompts to execute each step — one task, one approval, at a time
6. Review before merging — work through IMPLEMENTATION-checklist.md: tests pass, conventions followed, security and edge cases checked

What's Inside the Kit

Constructed Example: an API Feature

You want to build a feature: "Add user authentication to my API."

1. Fill in PRD-template.md: Define success (JWT tokens, Redis cache, 2FA optional, tests required)
2. Send the PRD to Copilot and ask: "Generate an implementation plan based on this PRD"
3. Structure the output using PLAN-template.md — maybe 5 tasks: setup dependencies, create token service, write middleware, implement refresh, add tests
4. For each task, use TASK-prompt-template.md to give Copilot one focused instruction with full context
5. Before merging each task, work through IMPLEMENTATION-checklist.md — tests pass, conventions followed, security checked

Why This Works

• Consistent output — AI agents work best with structured, explicit intent
• Architectural control — you stay in charge; AI is a tool, not the decision-maker
• Repeatable workflow — same structure works for microservices, React frontends, data pipelines, anything
• Evidence before acceptance — checkpoints help expose issues; they do not guarantee production readiness or authorize release
• Team-friendly — your teammates can follow the same workflow and handoff

Next Steps

This starter kit is free and ready to use immediately. No sign-up, no course required. Just download, copy the templates, and start.

For the vendor-neutral definition and responsibilities, read the AI Driven Development Methodology reference. For tool-specific certification preparation, explore:

• GitHub Copilot GH-300 Study Guide — Free written domain guides, official source links and a separate route to the Udemy preparation course

Questions? Email info@aiarchitectmastery.com