Tutorial
AI Tutorial
A practical tutorial for using AI and large language models in your personal and work life. Covers LLM fundamentals, prompting techniques, tools, use cases, limitations, and practical workflows.
Chapters
About this tutorial
A practical guide to using large language models in your personal and work life.
Who This Is For
- Anyone new to AI tools who wants to use ChatGPT, Claude, or Gemini effectively
- Working professionals who want to fold AI into daily tasks (email, research, code, planning)
- Builders who want to graduate from chat UIs to APIs and prompt design
Contents
| Chapter | Topic | Description |
|---|---|---|
| 01 | Fundamentals | What LLMs are, how they work, key concepts and terminology |
| 02 | Prompting Basics | Core principles of effective prompting |
| 03 | Advanced Prompting | Techniques for complex tasks and better results |
| 04 | Tools and Platforms | ChatGPT, Claude, Gemini, APIs, and the ecosystem |
| 05 | Personal Use | AI for productivity, learning, creativity, and daily life |
| 06 | Business Use | AI for work, automation, and team workflows |
| 07 | Limitations and Risks | Hallucinations, security risks, and how to use AI responsibly |
| 08 | Future Trends | What is coming and how to stay current |
| 09 | Practical Examples | Real-world templates and workflows |
| 10 | Quick Reference | Cheat sheet and best practices |
How to Use This Tutorial
- Read sequentially if AI is new to you. Chapters 01 and 02 give the vocabulary you need before the rest makes sense.
- Type out the prompts. Copy them into a real chat with a real model. Reading them is not the same as running them.
- Build something after each chapter. A prompt library, an automated workflow, a small script. Theory without practice fades fast.
Quick Reference
Pick a Tool
ChatGPT general use, plugins, code execution, image generation
Claude long documents, careful analysis, writing
Gemini huge context windows, Google Workspace users
Perplexity real-time search with citations
A Decent Prompt
You are [role with relevant expertise].
I need you to [specific task].
Here is the context:
[the facts the model needs]
Output as [format].
Constraints: [length, tone, what to avoid].
Common Patterns
Step-by-step "Walk through this step by step."
Few-shot Give 2-3 examples of input -> output, then ask for a new one.
Role "You are a [role]. Answer as one would."
Negative "Do not [specific failure mode]."
Format-first Show the table or JSON shape, then ask the model to fill it.
Learning Path Suggestions
New to AI (one week)
- Read chapters 01 and 02
- Sign up for one of ChatGPT, Claude, or Gemini
- Use it for three real tasks (an email, a summary, a tricky decision)
- Read chapter 04 and pick a primary tool
Working professional (two weeks)
- Skim 01, read 02 and 03 carefully
- Pick three workflows from 05 or 06 and try them on real work
- Read 07 before relying on AI for anything important
- Build a prompt library of 10 prompts you actually re-use
Builder or developer (three weeks)
- Read 01-04
- Open the API docs for one provider and run the quickstart
- Read 09 for templates and 03 for the harder techniques
- Build a small RAG or agent project
Core Philosophy
- AI is a tool. It is not magic and it is not an oracle. Treat outputs as drafts.
- Prompts in, results out. The quality of what you write determines the quality of what you get back.
- Iterate fast. First-pass prompts rarely work. Adjust and try again.
- Context is most of the work. Models with the relevant facts produce relevant answers. Models without them confidently make things up.
- Stay skeptical. Verify anything that matters. Especially numbers, citations, and code.
- Keep up. The field moves quickly. Block 30 minutes a week for it.
A Note on Dates
This tutorial is written in 2026 and reflects the model lineup, prices, and capabilities current at that time. Specific model names and token prices age fastest. Treat them as snapshots, not eternal truths. Concepts like prompting structure, hallucinations, and context windows will stay relevant much longer than any specific model name.