How a Skeptical Salesman Changed My Approach to AI

"Personal relationships are really important to me. I think AI is going to make everything feel impersonal."

The salesman's words hung in the air during our client meeting. As someone who helps businesses integrate AI into their workflows, I've heard this concern before. But this time, it sparked something different.

What if we could use AI to strengthen relationships instead of weakening them?

The Follow-up Email Problem

Every salesperson knows the power of a thoughtful follow-up email. The kind that references specific conversation points, acknowledges personal details, and moves the relationship forward. The kind that often doesn't get written because it takes too much time.

That's when it hit me: What if we could drop any meeting transcript into ChatGPT and get back a perfectly written, personalised follow-up email?

The Raw Material Revolution

Most people approaching this problem would obsess over writing the perfect prompt. I knew that would fail. Why? Because AI has been trained on humanity's collective output—including all the impersonal marketing drivel we've created over the years.

The secret isn't in the prompt. It's in the raw material.

From Skepticism to System

I decided to prove it. After my next meeting, I wrote a perfect follow-up email the old-fashioned way. Then I gathered the meeting transcript from fireflies.ai (not an affiliate, I just use it for meeting transcription) and did something different.

Instead of trying to craft the perfect prompt, I asked Claude to study the relationship between these two documents—to understand how my mind transformed one into the other.

The Pre-prompt Framework Emerges

This approach revealed a powerful progression:

1. Pre-prompt: You teach AI to understand your thought process

2. Prompt: AI generates its own system of instructions based on its analysis

3. Prompt Blueprint: You transform AI's output into a reusable template

Think of it like creating a bespoke suit pattern rather than a single suit. The pattern captures your style while allowing for endless variations.

Building the Blueprint

The magic happens in three simple steps:

First, you show AI two documents: your raw meeting transcript and your perfectly crafted follow-up email. You ask it to study how one gives rise to the other—like teaching it to think the way you think.

Next, let the AI create its own system prompt based on what it learned. This prompt will contain your specific details and style choices, capturing your unique approach. But it will also likely include a set of variables that the AI will try to extract from every transcript.

Finally, you take that prompt and replace the specific details with placeholders. Now you have a blueprint—a template that your colleagues can use by filling in their own details while maintaining your proven method.

Testing the Theory

To underline this wasn't a one-off success, I applied the same approach to something completely different: generating unique yet valid CrossFit workouts. Using exercise physiology data and CrossFit methodology as input, I created WODGPT—a system that generates workouts that make even seasoned CrossFitters question their life choices. (Try it here for yourself: WODGPT)

The Return to Relationships

Remember that skeptical salesperson? His concern helped reveal something crucial: generic AI outputs can indeed damage relationships. But when you feed AI rich, detailed input data and teach it how to think through a thoughtful pre-prompt, you create something powerful—a system that maintains the human touch while scaling your best practices.

That's the real breakthrough. We're not just writing prompts; we're teaching AI to understand how humans transform information into meaningful communication and let the AI create a prompt based on that understanding. It’s very different from just prompting.

Your Turn

Stop crafting one-off prompts. Start building systems that capture your expertise and scale your best practices. Whether you're writing follow-up emails, creating content, or solving complex problems, the principles remain the same:

1. Start with rich, detailed input

2. Create one perfect output example

3. Build a system to bridge the gap

If you found this valuable, send it to a colleague who might find it useful too.

Transform your relationship to AI with 10xbetter.ai