AI Use Cases for Normal People - Tutorial Video
ai
tutorial
video
In this 2-hour hands-on tutorial, I break down how AI coding agents actually work — starting with Claude Cowork, how coding agents like this compare with ChatGPT and then building up to Claude Code, Skills, and real-world demos you can use today.
No coding background required. If you’ve used ChatGPT, you’ll understand this.
The slides and materials used in the video are here.
What You’ll Learn
- The difference between ChatGPT and coding agents like Claude Code
- What “bash” is and why it gives AI superpowers
- How Claude Code lives on your computer and reads your files
- What Skills are and how to use them
- Real demos: Excel data, dashboards, video editing, ML pipelines
- How to talk to Claude using plain English (no coding needed)
- Token costs, model choices (Opus vs Sonnet vs Haiku), and how to save money
- How to run AI models locally on your own Mac
Timestamps
- 00:00:00 Intro — ChatGPT vs AI Coding Agents
- 00:01:13 What are agent frameworks? (LangChain, LangGraph)
- 00:02:14 Claude Code — what it is and when it came out
- 00:02:48 Demo: Using Claude Code in the terminal
- 00:03:17 What is bash? (plain English explanation)
- 00:05:18 The model vs. the harness — how Claude Code is built
- 00:06:50 The big difference: Claude Code lives on YOUR computer
- 00:09:50 Using AI to query databases (BigQuery example)
- 00:11:35 Demo: Claude Code edits videos (brisket video story)
- 00:12:36 Safety warning: prompt injection and what to watch out for
- 00:14:17 What are Skills? (giving Claude an instruction manual)
- 00:17:23 Demo: My ML pipeline skills (find data → train → deploy)
- 00:20:27 A skill that creates other skills
- 00:22:15 3 methods: base prompting vs. skills vs. full workflows
- 00:25:25 Claude for Excel — no coding needed
- 00:26:14 Demo: Processing Shannon’s Excel data live
- 00:38:08 Demo: Building an HTML dashboard from real data
- 00:41:04 Claude explores the data and writes the merge script
- 00:55:30 Demo: SQL skill for overtime hours calculations
- 00:57:07 Where skills live — just plain text files!
- 01:03:07 CLAUDE.md — giving Claude your project instructions
- 01:05:24 Project instructions vs. user instructions (scoped settings)
- 01:09:56 How this compares to ChatGPT’s memory and instructions
- 01:13:27 How long Claude Code has been out + what to learn next
- 01:16:04 Claude as a conversational coding agent — just ask in English
- 01:18:22 Live demo with Shannon’s data
- 01:19:00 Running multiple Claude instances at the same time
- 01:20:25 Demo: Front-end design skill — building UIs automatically
- 01:23:37 Choosing the right Claude model (Opus, Sonnet, Haiku)
- 01:24:06 Token pricing explained — what does it actually cost?
- 01:29:10 How transformers and tokens work (simple explanation)
- 01:33:02 Doing the math: what did today’s session actually cost?
- 01:38:06 AI infrastructure — why companies are building power plants
- 01:40:05 The Claude Code source code leak story
- 01:42:26 Running Python and other languages inside Claude Code
- 01:45:21 Claude’s tools: read, edit, bash, grep
- 01:47:58 Model cards — how to evaluate and grade AI models
- 01:49:20 Chinese open source models and the competition
- 01:49:51 Running AI models locally on your own computer
- 01:50:42 Using your Mac’s memory as a GPU
- 01:55:22 Demo: Building a data pipeline visual with one prompt
- 01:56:03 Plan mode vs. other approaches