7 Days Claude Bootcamp: The Prompt Engineering Blueprint

If you have been using Claude the same way you use ChatGPT, you have been leaving your best results on the table. This lecture is where that changes.

Claude and ChatGPT were not built the same way. They were trained with different objectives, different reward structures, and fundamentally different approaches to how they interpret and respond to prompts. That difference matters — because a prompt optimized for one model will underperform on the other, often significantly.

In this lecture, you will learn exactly how Claude's architecture differs from GPT-based models at the reasoning level — not the technical level. No jargon, no PhD required. Just a clear, practical mental model that will immediately change how you write every prompt from this point forward.

You will discover why Claude attempts to understand the intent behind your request rather than just the literal instruction — and why that single difference makes context and purpose more important than clever wording. You will learn why Claude's consistent character and willingness to push back on unclear prompts is a feature, not a limitation — and how to use it as a real-time quality signal for your prompts.

By the end of this lecture, you will have a new mental model for working with Claude — one built on how Claude actually thinks, not how you assumed it did.

What you will learn in this lecture:

• Why Claude and ChatGPT require fundamentally different prompting approaches

• How Claude's training process shapes the way it interprets your instructions

• Why Claude uses intent and context, not just literal commands, to generate responses

• How to recognize when Claude is signaling that your prompt needs more information

• The foundational mindset shift that makes every other technique in this course work

Day 1 — SEO Course Descriptions: All Lectures and Assignment

Lecture 01: The Claude vs. ChatGPT Mental Model Shift

Lecture Description:

If you have been using Claude the same way you use ChatGPT, you have been leaving your best results on the table. This lecture is where that changes.

Claude and ChatGPT were not built the same way. They were trained with different objectives, different reward structures, and fundamentally different approaches to how they interpret and respond to prompts. That difference matters — because a prompt optimized for one model will underperform on the other, often significantly.

In this lecture, you will learn exactly how Claude's architecture differs from GPT-based models at the reasoning level — not the technical level. No jargon, no PhD required. Just a clear, practical mental model that will immediately change how you write every prompt from this point forward.

You will discover why Claude attempts to understand the intent behind your request rather than just the literal instruction — and why that single difference makes context and purpose more important than clever wording. You will learn why Claude's consistent character and willingness to push back on unclear prompts is a feature, not a limitation — and how to use it as a real-time quality signal for your prompts.

By the end of this lecture, you will have a new mental model for working with Claude — one built on how Claude actually thinks, not how you assumed it did.

What you will learn in this lecture:

• Why Claude and ChatGPT require fundamentally different prompting approaches

• How Claude's training process shapes the way it interprets your instructions

• Why Claude uses intent and context, not just literal commands, to generate responses

• How to recognize when Claude is signaling that your prompt needs more information

• The foundational mindset shift that makes every other technique in this course work

Keywords naturally integrated: Claude vs ChatGPT, Claude prompting, how Claude works, Claude mental model, prompt engineering for Claude, ChatGPT vs Claude differences, Claude AI training

Lecture 02: Constitutional AI in Plain English — And Why It Changes Your Prompts

Lecture Description:

Every response Claude gives is shaped by something most users have never heard of — Constitutional AI. Understanding it does not require a background in machine learning. It requires about 20 minutes and this lecture.

Constitutional AI is Anthropic's method for training Claude to evaluate its own outputs against a written set of principles — covering helpfulness, honesty, and avoiding harm. Unlike models trained purely on human feedback, Claude was built to reason about whether its response is good, not just whether a human approved it. That distinction changes everything about how Claude processes your prompts.

In this lecture, you will learn how Constitutional AI works in plain, practical terms — and more importantly, how it changes what Claude does with your prompts before it ever writes a word of response. You will understand the three core pillars Claude balances in every reply, why those pillars sometimes create tension, and how to write prompts that resolve that tension so Claude can help you at full capacity.

You will also learn the single most important reframe in this entire course: Claude is not an execution engine. It is a reasoning system with internalized values — and when you write prompts that account for those values, you get dramatically better results than when you fight against them.

This lecture replaces frustration with understanding. Once you know why Claude responds the way it does, you will never write a vague or ambiguous prompt again.

What you will learn in this lecture:

• What Constitutional AI is and how it was used to train Claude's behavior

• The three pillars Claude balances in every response — and why they sometimes conflict

• How Claude evaluates your prompt before generating a response

• Why providing context and intent unlocks better results than clever prompt tricks

• How to frame prompts that work with Claude's values instead of triggering resistance

You have probably heard the term "context window" before. Most explanations stop at "it is how much Claude can remember." That explanation is incomplete — and the gap between what most people understand and what is actually happening is costing them output quality every single day.

This lecture gives you the real picture. Claude's context window is not a memory bank — it is the entire world Claude exists in during your conversation. Everything Claude knows about your task, your history in this session, your instructions, and your examples lives inside that window. And how Claude prioritizes information within that window directly determines which parts of your prompt get followed and which parts get quietly ignored.

In this lecture, you will learn how tokens work and why they matter for prompt design — without any technical complexity. You will discover the most important positioning principle in prompt engineering: where you place your instructions inside a prompt matters as much as what those instructions say. You will learn about the "lost in the middle" effect — the documented pattern where instructions buried in long prompts receive significantly less attention than those placed at the beginning or end.

You will also learn how Claude resolves conflicts when different parts of your prompt appear to contradict each other — and how to structure every future prompt so your most critical instructions always get the weight they deserve.

This lecture turns the context window from a vague concept into a practical design tool.

What you will learn in this lecture:

• What the context window actually is and how Claude uses it during a conversation

• How tokens work and what the 200,000 token window means in practical terms

• Why instruction placement inside your prompt affects how Claude prioritizes your request

• The "lost in the middle" effect and how to avoid it in every prompt you write

• How Claude handles conflicting instructions — and how to design prompts that eliminate that conflict

This is where Day 1 stops being theory and starts being yours.

The Prompt Audit assignment is designed to take everything you learned in Day 1's three lectures — the Claude vs. ChatGPT mental model shift, Constitutional AI, and context window architecture — and apply it immediately to prompts you are already using in your real work.

You are not going to write hypothetical prompts for imaginary scenarios. You are going to take three actual prompts you have used before — on Claude or any other AI tool — and rebuild them from the ground up using the mental model and principles you learned today. Then you are going to run both versions, compare the outputs side by side, and document exactly what changed.

This exercise does three things at once. First, it locks in your Day 1 learning faster than any amount of re-reading or note-taking ever could. Second, it gives you your first three upgraded prompt entries for your personal prompt library — the core deliverable of this entire bootcamp. Third, it shows you in concrete, measurable terms how much of Claude's capability your old prompts were leaving untouched.

Most students who complete this assignment report that seeing the before-and-after output comparison is the moment the entire course clicks into place. The difference between a prompt written without the Day 1 mental model and one written with it is not subtle — it is visible, immediate, and often surprising.

By the time you finish this assignment, you will not be able to write a prompt the old way again. That is the goal.

Assignment steps:

• Select three real prompts you have used previously on any AI platform

• Rewrite each prompt applying the RCTFC awareness, context-first thinking, and intent framing from Day 1's lectures

• Run both the original and rewritten version in Claude and save both outputs

• For each prompt pair, write two to three sentences documenting what specifically changed in the output and why

• Add all three rewritten prompts to your prompt library with a version tag of v1.0 and a date

What you will produce:

• Three upgraded, documented prompts ready to add to your personal prompt library

• A before-and-after output comparison for each prompt that demonstrates the Day 1 principles in action

• A written reflection on the most impactful single change you made across the three rewrites

Why this assignment matters: Every assignment in this bootcamp is designed to build your personal prompt library — a reusable, evolving collection of tested prompts for your specific work. The Prompt Audit is the foundation. The prompts you refine today are the baseline every future day's work will improve upon. Do not skip it.

Estimated completion time: 20–25 minutes