The AI fluency to use Claude confidently and accurately in daily work — how AI models process information, how to write prompts that work on the first try, and which model to reach for.
By the end, you will
Lessons in this module
What AI actually does, the terms you'll see daily, and how the agentic loop works.
HOW you ask vs. WHAT you give the AI to work with, five patterns, and the complete prompt formula.
Four core techniques, the COSTARS framework, interview-style prompting, and validation habits.
Haiku, Sonnet, and Opus — a task-based selection guide and real cost comparisons.
Key Concepts
Core terms introduced in this module. The lesson where each term is first defined is listed so you can return to the source explanation.
| Term | Defined In | Plain English Meaning |
|---|---|---|
| Prompt | Lesson 1.1 | The instruction or question you type to the AI. |
| Context Window | Lesson 1.1 | The AI's working memory. How much text it can hold and process at once. |
| Token | Lesson 1.1 | A chunk of text roughly equal to three-quarters of a word. Models measure capacity in tokens. |
| Agent | Lesson 1.1 | A pre-configured AI assistant with specific instructions and, sometimes, live connections to other apps. |
| MCP Server | Lesson 1.1 | A live connection between your AI assistant and another app or system, such as your email or a project tracker. |
| Knowledge Base | Lesson 1.1 | A collection of documents you provide to the AI. It searches them to answer your questions. |
| Hallucination | Lesson 1.2 | When the AI generates false information presented as fact — happens when it lacks real data and predicts instead. |
| LLM | Lesson 1.1 | Large Language Model. The technical name for AI models that generate text. |
| Agentic Loop | Lesson 1.1 | The cycle where an agent reasons, selects tools, executes actions, and repeats until the task is complete. |
| Prompt Engineering | Lesson 1.2 | Structuring HOW you ask — the wording, specificity, and format of your instruction. |
| Context Engineering | Lesson 1.2 | Controlling WHAT information you give the AI to work with before it generates a response. |
| Zero-Shot Prompting | Lessons 1.2, 1.3 | Asking once with clear instructions and no examples provided. Works for well-defined tasks. |
| Few-Shot Prompting | Lessons 1.2, 1.3 | Providing examples of the desired output before asking the AI to produce new output. Improves consistency. |
| Role Prompting | Lessons 1.2, 1.3 | Assigning the AI an identity (e.g. "You are an HR specialist") to shape its tone, vocabulary, and priorities. |
| Chain-of-Thought | Lessons 1.2, 1.3 | Asking the AI to reason step by step before giving a final answer. Reduces errors on complex tasks. |
| Interview-Style Prompting | Lesson 1.3 | Asking the AI to interview you to gather context before producing output. Useful for complex or unclear tasks. |
| Meta Prompting | Lesson 1.3 | Asking the AI to write a prompt for you. Useful when you know the goal but not how to phrase the instruction. |
| COSTARS | Lesson 1.3 | A framework for building complex prompts: Context, Objective, Style, Tone, Audience, Response, Scope. |
| Claude (Haiku/Sonnet/Opus) | Lesson 1.4 | Anthropic's model family, and your default for almost everything. Haiku is fast and cheap. Sonnet is the balanced default. Opus handles the most complex reasoning. |
| Gemini | Lesson 1.4 | Google's model family. The one niche exception worth knowing: Gemini handles video and audio natively, which no Claude model does. |
Support Resources
| Resource | Link | Purpose |
|---|---|---|
| Claude | claude.ai | Sign up and use Claude in your browser |
| Claude Docs | code.claude.com/docs | Official documentation |
| Claude Support | support.claude.com | Help center and account support |
Up next
Module 2: Setting Up Your Computer (optional)