Does my own framework hold up across AI platforms? I decided to find out — and what I discovered reinforces exactly why the Framework works.
The Honest Question
When I developed the Perfect Prompt Framework™, ChatGPT was the dominant AI tool most entrepreneurs were using. The framework was built around how that tool thinks. So when Claude emerged as a serious alternative — built by a different company, trained on different principles, designed with a different philosophy — I had to ask myself an honest question:
Does my framework still work? Or did I accidentally build something that only functions on one platform?
The answer surprised me. Not because the framework failed. Because it revealed something deeper about why it works in the first place.
A Quick Refresher: The Perfect Prompt Framework™
For those newer to my training, here are the four steps:
- Tell the AI what type of expert it should act as
- Give the AI background relevant to the task
- Ask your question
- End with: “Ask me any questions you have.”
Simple. Repeatable. Effective. The question is whether those four steps translate to Claude — which is built around something called Constitutional AI, a framework that trains Claude to reason about values and context rather than simply execute rules.
Step-by-Step: How the Framework Performs on Claude
Step 1: Tell the AI What Type of Expert It Should Act As
This step was built for ChatGPT. With ChatGPT, assigning a role is like giving an instruction: “You are a copywriter” functions as a command, and ChatGPT executes within that definition.
With Claude, something more interesting happens. Claude doesn’t just adopt the label — it reasons from the identity. That means the more specific and purposeful your role description, the more Claude leans into it.
“Act as a copywriter” gets you competent output. “Act as a direct-response copywriter who specializes in helping non-technical entrepreneurs communicate their value clearly” gets you something that feels intentional and targeted.
The nuance for Claude: Be specific about the purpose behind the role, not just the title. Claude uses the role to understand what you’re trying to accomplish — not just how to dress up the output.
Step 2: Give the AI Background Relevant to the Task
This step was always good advice. With Claude, it becomes genuinely important.
Claude is built to understand the spirit of what you’re asking — the intent behind the request, not just the request itself. That means the background you provide doesn’t just add context. It shapes how Claude reasons about your goal.
There’s one addition I’d make specifically for Claude: include the why. Not just what the task is, but what outcome you’re working toward.
Compare these two versions of background context:
Version A:
My audience is small business owners between 40-60 years old who are new to AI.
Version B:
My audience is small business owners between 40-60 years old who are new to AI. I want them to feel empowered, not overwhelmed. The goal of this piece is to build their confidence so they take one small action today.
With ChatGPT, both work reasonably well. With Claude, Version B produces output that genuinely reflects the emotional intent you described.
The nuance for Claude: Add a sentence about the purpose or intended outcome. Claude uses intent to guide judgment — and judgment is where Claude excels.
Step 3: Ask Your Question
No modification needed here. A clear, specific question produces better results on every AI platform. That’s true for Claude, ChatGPT, Gemini — all of them.
If anything, Claude rewards clarity even more than ChatGPT does. Because Claude is reasoning about your goal rather than just parsing your command, a vague question gives it less to work with. Be specific about what “good” looks like.
Step 4: “Ask Me Any Questions You Have”
This step was designed to break ChatGPT out of execution mode — to invite it to clarify before producing output that might miss the mark.
With Claude, you may find this step is less necessary. Claude’s default tendency is to surface ambiguity and ask clarifying questions when context is incomplete. It often does this on its own before producing output.
That said, I still recommend keeping it. It signals that you want a collaborative exchange, not just a one-shot response. And it keeps you in the habit of framing AI interactions as conversations rather than commands — which is the right mental model for getting the most out of both tools.
The nuance for Claude: Claude may preempt this step. Don’t be surprised when it asks you questions before you ask it to. That’s a feature.
The Deeper Insight: Why the Framework Translates
Here’s what this exercise revealed.
The Perfect Prompt Framework™ wasn’t built around a specific AI tool. It was built around how humans communicate when they want excellent work from another person.
When you hire a talented contractor, you don’t hand them a list of restrictions. You tell them who they are in this context, give them the background they need, ask your question clearly, and invite them to clarify anything before they start.
That’s the framework. And it turns out, that’s also exactly how Claude is designed to be communicated with.
ChatGPT responds well to clear commands because it’s built to follow instructions with precision. Claude responds well to role, context, and intent — because it’s built to reason from them. The Perfect Prompt Framework™ naturally does both.
Quick Reference: ChatGPT vs. Claude Prompt Styles
| Prompt Element | ChatGPT Prompt Style | Claude Prompt Style |
| Opening | Direct command | Role assignment (“You are a…”) |
| Constraints | Rules (“Never,” “Always”) | Context (“Keep it brief”) |
| Tone guidance | Specific rules | Descriptive framing |
| Format | Explicit instructions | Embedded in context |
| Best for | Structured, repeatable tasks | Nuanced, judgment-based tasks |
What This Means for You Practically
You do not need two different frameworks for two different AI tools. You need one good framework — and the awareness of a few small adjustments.
When you use the Perfect Prompt Framework™ on Claude:
- Make your expert role specific to the purpose, not just the job title
- Add the intended outcome to your background context — the “why” behind the task
- Ask your question with the same clarity you already bring
- Keep the “ask me questions” prompt — it reinforces the collaborative dynamic Claude thrives in
That’s it. Four steps. Two minor nuances. One framework that works across platforms.
Want to Build This Skill?
Inside AI Insiders, I walk entrepreneurs through the practical application of the Perfect Prompt Framework™ across the tools you’re actually using in your business. Not theory. Real prompts, real workflows, real results.
If you’re ready to stop guessing and start getting consistent, high-quality output from AI — this is where that happens.
Join AI Insiders at WhiteBeardStrategies.com →
Jonathan Mast is an internationally recognized AI coach and founder of White Beard Strategies LLC. He helps entrepreneurs and organizations leverage AI to save time, make money, and deliver more value.
