Skip to main content

1.0 Program Overview

Welcome to the Liatrio Forge: Immersive AI Mastery Program — an immersive, fast-paced experience with minimal lectures and maximum hands-on building.

What This Program Is​

This is language immersion for AI-native engineering. Just like learning a language by living in a country where it's spoken, you'll learn AI-native development by doing everything with AI. The goal is to experience firsthand what AI is good at, what it struggles with, and how to orchestrate it effectively.

Work at your own pace. The content is organized into modules with clear exit criteria. Move through them at the speed that works for you.

If You're Stuck

If you're not making progress or feel like you're spinning, reach out for help. Getting stuck is normal — staying stuck is not.

What is AI-Native Engineering?​

AI-native engineering is a fundamental shift in how software gets built. In this program, you'll experience what it's like to:

  • Delegate 99%+ of coding to AI agents while you provide specifications, guidance, and critical review
  • Use Spec-Driven Development (SDD) to maintain clarity, prevent scope creep, and produce evidence for every feature
  • Ship real artifacts — working software, documentation, pipelines — not toy examples

By the end, you'll confidently treat AI not just as a tool, but as the primary driver in product development.

Program Structure​

The program progresses through four modules, each building on the last:

ModuleFocusExit Criteria
1AI-Driven Development FoundationsFixed all open issues in Pet Clinic with 99% AI-written code
2AI-Augmented Features & Systems ThinkingIntegrated a major AI feature across components/services
3AI-Accelerated DevOps & Platform EngineeringBuilt a working pipeline or IaC deployment slice
4Autonomous Capstone & Client EnablementDelivered an end-to-end solution autonomously and explained it to others

Demos and Sharing​

You will post a demo twice weekly to #liatrio-forge-demos.

Record a short (5-10 minute) demo that walks through:

  • What you built or fixed
  • How you used AI to solve the problem
  • What tools and techniques you used

Post the recording to #liatrio-forge-demos. Learning from each other's approaches is a key part of the program, so watch a few of your wave's posts when you get the chance.

Proof Artifacts​

As you work, you naturally generate proof artifacts that demonstrate your work:

  • Spec document (markdown) — The specification you wrote
  • Task breakdown — How you decomposed the work
  • PR link(s) — Your implementation, reviewed by AI
  • Test output — Proof that acceptance criteria passed
  • Demo recording — Your walkthrough of the solution

These live in your repo as you go — specs and tasks under docs/specs/, PRs and tests in GitHub. There's no formal submission or drop box; keep them up to date and available in case a facilitator asks to see them. The one exception is the demo recording, which you actively post to #liatrio-forge-demos.

Prerequisites​

Jumpstart Webinar

Before starting this program, watch the AI-Native Jumpstart | Building with Spec-Driven Development webinar. This background gives you the shared language for:

  • The context window and why it matters
  • Context rot and how to combat it
  • The basics of Spec-Driven Development
  • The "Golden Path" concept for AI-native workflows

Use the liatrio.ai webinar page as the Jumpstart reference for Forge. You do not need a separate Jumpstart document or tool list to complete this program.

Forge officially supports two ways to run Claude Code: the terminal CLI, or the Claude Code Desktop app (the Code tab in the Claude Desktop app). Follow this module's setup pages as the authoritative install path for whichever you choose — both give you the same underlying Claude Code engine, slash commands, and configuration. Do not install additional IDE agents or competing AI coding tools unless a Forge lesson explicitly asks for them.

Tip: You can also complete Section 1.1 Environment Setup while you watch — pause the webinar as Damien walks through each step and follow along on your own machine.

tip

Keep the SDD playbook handy. The Spec-Driven Workflow playbook is the direct reference for the workflow you will use throughout Forge. Start there when you need the bigger picture, and check Common Questions before you get stuck on setup, workflow mechanics, or what an SDD step is asking you to do.

Technical Requirements​

  • Familiarity with Git and version control
  • Basic command-line proficiency
  • Experience with at least one programming language
  • Claude Code access via a Claude subscription, unless directed to use an Anthropic API key — start with 1.05 Setting Up Claude Max
  • If you plan to use the Claude Code Desktop app instead of the terminal CLI, a Pro, Max, Team, or Enterprise Claude subscription (or API access)

Reading Markdown Artifacts​

Almost everything this program produces is plain Markdown written to disk — spec.md files, task lists with checkboxes, validation reports, and the AGENTS.md context file. You will read these far more often than you write them, and a rendered view (real tables, checked boxes, collapsed headings) is much faster to scan than raw text in a terminal pager.

Pick whichever you already have:

OptionHow to preview
VS CodeOpen the file and press Cmd+Shift+V (macOS) or Ctrl+Shift+V (Windows/Linux)
JetBrains IDEsOpen the file and switch to the Preview tab in the editor gutter
GitHubPush the branch and view the file in the browser — GitHub renders Markdown natively
A viewer is not an agent

The rule above is about AI agents — don't run a second coding assistant alongside Claude Code. A Markdown previewer only reads files, so use one freely. Keep Claude Code (terminal or Desktop app) as your only coding agent.

What You'll Work With​

Throughout this program, you'll work with Emerald Grove Pet Clinic — a full-featured veterinary management application. This is a production-style codebase designed to mimic the complex, brownfield environments we see at real customers:

  • Real architecture patterns
  • Comprehensive test suites
  • CI/CD pipelines
  • AI-native documentation (AGENTS.md, specs, standards)

You'll fix issues, add features, and extend functionality — all with AI doing the coding.

Tools You'll Master​

  • Claude Code — Anthropic's AI-native development agent, used via the terminal CLI or the Claude Code Desktop app
  • Spec-Driven Development (SDD) Skill — Structured workflow for reliable AI delivery
  • MCP Servers — Tools that give Claude access to external systems
  • Hooks — Lifecycle scripts that trigger automatically on agent events (e.g., before a tool runs), enforcing rules deterministically instead of relying on prompts alone
  • AI Code Review — Automated PR review and feedback

Expectations​

  1. AI writes the code. You direct, review, and validate — but you're not typing code by hand.
  2. Demo your work. Twice weekly, post a recording of what you built and how you built it to #liatrio-forge-demos.
  3. Ask for help. If you're stuck for more than a few hours, reach out. Don't spin.
  4. Help each other. If you finish quickly, check in with others who might be struggling.

Collaboration​

This is individual work — you're responsible for your own deliverables. But you're not alone:

  • Ask questions in #liatrio-forge
  • Share tips and techniques that worked for you
  • Pair with others if you're stuck on similar problems
Capstone pitch review

Before you start building in the Capstone Module, you'll post a short pitch in #liatrio-forge so facilitators can review scope before you dive in. 5.0 Capstone covers what to include.

Let's Begin​

The next section walks you through setting up your development environment. By the end of this introduction, you'll have:

  1. Claude Code installed and configured
  2. The SDD workflow installed as an agent skill
  3. MCP servers and skills ready to extend Claude's capabilities
  4. The example repository forked and explored

Let's get started.

Knowledge Check​

Test your understanding of the Forge program before moving on.

Question 1 of 3

Program Overview Knowledge Check

Q1. What methodology does the Forge program use to ensure AI delivers reliable, scoped features?