Day 01 Foundations

Claude Basics — Your First Prompt That Gets Results

Learn how Claude differs from ChatGPT, the anatomy of a good prompt (role + context + task + format), and get real results from your first conversation.

~1 hour Hands-on Precision AI Academy

Today's Objective

Here's why most people get bad results: they give Claude a task without giving it enough to work with.

01

How Claude Is Different From ChatGPT

Claude and ChatGPT are both AI assistants built on what's called a large language model — a type of AI trained on vast amounts of text to understand and generate language. They're built by different companies (Anthropic vs. OpenAI) with different design priorities. Understanding those differences shapes when to use each one.

CapabilityClaudeChatGPT
Context window (how much text it can read at once) 200K tokens (~150K words — roughly a full novel) 128K tokens (GPT-4o)
Document analysis Excellent — full contracts, reports, code Good
Honesty / refusals More likely to say "I don't know" Sometimes more confident than it should be
Writing quality More nuanced, better tone control Good, slightly more generic
Web browsing Limited (claude.ai has some search) Yes, with GPT-4o
Image generation No Yes, with DALL-E
Plugins / GPTs No plugin store GPT Store
💡
When to pick Claude: Long documents, nuanced analysis, careful writing, anything where you need an honest answer more than a confident one. If you're uploading a 100-page contract and need a summary, Claude wins. If you need to generate an image or browse the web, use ChatGPT or Gemini.

Claude is built by Anthropic, a company whose stated focus is "AI safety." In practice, that means Claude is designed to be honest rather than just confident. It's more likely to say "I'm not certain about this" rather than inventing a plausible-sounding answer. It's also more likely to push back if a task seems off. For analysis work where accuracy matters more than speed, that's exactly what you want.

02

The Anatomy of a Good Prompt

Here's why most people get bad results: they give Claude a task without giving it enough to work with. A vague prompt gets a vague response. Claude isn't a mind reader.

A good prompt has four parts. You don't always need all four, but knowing them makes you a better prompter.

The 4-part prompt framework
Role Who should Claude be for this task? "You are an experienced contract lawyer." "You are a plain-English technical writer." This sets the expertise and tone.
Context What does Claude need to know? Background info, constraints, who the audience is, what tool/format you're working in.
Task What exactly do you want done? Be specific. "Summarize" is vague. "Summarize in 3 bullet points, each under 20 words" is precise.
Format How should the output look? Bullet list? Table? Paragraph? JSON? Numbered steps? If you don't specify, Claude guesses — and often guesses wrong.

Bad prompt vs. good prompt

Here's the same request written two ways. Notice the difference in what you'd actually get back.

Bad Prompt
Bad Prompt
Make this email better:

Hi John, I wanted to follow up on our meeting last week.
Let me know if you have any questions. Thanks
Good Prompt
Good Prompt
You are a senior business communications consultant.

Context: This is a follow-up email to a prospect after
a discovery call. John is a VP of Operations at a
mid-size logistics company. The meeting went well —
he seemed interested but asked for a pricing proposal
that I'm sending in a separate attachment.

Task: Rewrite the email below to be warm but
professional, include a clear next step, and make it
feel like I'm confident and organized.

Format: Keep it under 5 sentences. No bullet points.

Email to rewrite:
"Hi John, I wanted to follow up on our meeting last week.
Let me know if you have any questions. Thanks"
ℹ️
You don't need to write an essay every time. Once you internalize the four parts, prompts just come out better. For a quick task: one sentence is fine. For complex analysis: give Claude everything it needs.
03

Your First Real Prompt

Let's put this into practice right now. Go to claude.ai — free account is fine for this exercise.

Before we get to the exercise, here are a few things worth knowing about how to interact with Claude:

04

Common Beginner Mistakes

These are the things that trip people up in the first week:

Supporting References & Reading

Go deeper with these external resources.

Docs
Claude Basics: Your First Conversation That Actually Produces Results Official documentation for claude.
GitHub
Claude Basics: Your First Conversation That Actually Produces Results Open source examples and projects for Claude Basics: Your First Conversation That Actually Produces Results
MDN
MDN Web Docs Comprehensive web technology reference

Day 1 Checkpoint

Before moving on, confirm understanding of these key concepts:

Continue To Day 2
Day 2 of the Mastering Claude course