A practical guide to using AI tools like someone who actually builds things — not someone who just talks about them.
Who This Is For
You can code. You can navigate a terminal. You’ve used an LLM at least enough to know that sometimes it’s magic and sometimes it’s confidently wrong. But you’re not running enterprise infrastructure or shipping SaaS to thousands of users. You build tools for yourself, automate your own workflows, and want AI to make that faster and smarter.
This course assumes basic programming literacy and some LLM experience. It does not assume you know anything about agent architectures, workflow automation platforms, or data engineering. Yet.
What You’ll Get Out of This
- Prompting that works reliably, not prompting that works when you’re lucky
- A mental model of the AI tool landscape — what exists, when each tool earns its price tag, and when to stay put
- A professional development workflow with specs, context management, and subagents — instead of just vibing and hoping
- Working knowledge of autonomous agents and how to set one up without waking up to a surprise bill
- An honest assessment of which business tasks AI handles well today and which ones it botches
- The data engineering mindset — think like an architect, let the LLM be the contractor
- The connective tissue — MCP, local LLMs, workflow automation, and how they wire into a personal AI stack
- At least one concrete project started or completed
Course Structure
Nine modules plus a project lab. Each one is self-contained — read them in order for the full arc, or skip to whatever’s most relevant. Every module is designed to be digestible in a single 30–60 minute sitting.
| # | Module | What It Covers |
|---|---|---|
| 01 | Prompting That Actually Works | Role framing, plan-first workflows, critical review, pushing back |
| 02 | The Vibe Coding Landscape | Why platforms exist, when to use each, when to stay put |
| 03 | Spec-Driven Development | CLAUDE.md, specs, context engineering — the professional workflow |
| 04 | Multi-Agent Workflows | Subagents, dispatch patterns, cost management |
| 05 | OpenClaw Agent Runtime | Autonomous agents: setup, skills, security |
| 06 | AI for Business Tasks | Bookkeeping, legal, e-commerce — what’s real vs. hype |
| 07 | Data Engineering Mindset | Think like an architect, let LLMs be the contractor |
| 08 | Connective Tissue | MCP, local LLMs, n8n — wiring it all together |
| 09 | Project Lab | Five concrete projects with full breakdowns |
How to Use This Course
Go in order if you want the full experience. The modules build on each other — prompting foundations feed into spec-driven development, which feeds into multi-agent workflows, and so on. Later modules occasionally reference earlier ones.
Skip around if you already know your stuff. Comfortable with prompting? Start at Module 03. Already using subagents? Jump to Module 05. The modules stand alone well enough that you won’t be lost.
Don’t try to marathon it. Each module is a 30–60 minute read. Let things settle between sessions. The techniques stick better when you try them on a real project between modules.
How Resources Are Handled
Each module teaches the material directly. You shouldn’t need to leave the guide to understand anything.
External links appear only when a tool requires you to visit its site to use it, a resource goes significantly deeper than the course needs to, or a video walkthrough genuinely adds something text can’t convey. These are marked as Further Reading or External Tool so you know the difference.
A Note on Shelf Life
AI tooling moves fast. Specific tool names, pricing, and features will drift. The frameworks — how to evaluate tools, how to structure your work with AI, how to think about data — are more durable. When you revisit this course, expect the tools to have changed but the thinking patterns to have compounded.
Start here: Module 01 — Prompting That Actually Works