From The Desk of Matt Zimmerman

---

Dear AI Prompt Creator,

You know those levers I spoke about? Yes they do exist and I use them to guide the AI to write high-quality responses. I call these levers my “prompting principles.”

I discovered a few of them during 600 hours of prompt engineering for a company that makes search engine optimization software.

The rest of the principles I learned while coding my own AI writing software. It’s been about 2,000 hours of programming and prompt engineering, but the end result has helped tens of thousands of users create millions of blogs and high quality content. (Shameless plug: it’s called ZimmWriter.)

Now here’s the crazy part.

I haven’t seen many of these prompting principles shared by anyone else. Maybe it’s because these principles require thousands of hours of prompt engineering to discover. But whatever the reason, I’m going to share them with you in my training.

The training is called: “Get Prompted.”

In the training, I’ll share the prompting principles I’ve discovered to make the AI obey your prompt and deliver high-quality responses.

Here’s a handful of things you’ll learn in the training:

• ​A science-backed principle (inspired by Sun Tzu) to indicate when “chatting” with the AI is a bad idea.
• ​​​The #1 WORST mistake you can make when creating a prompt in ChatGPT. (Even if you do everything else “right”, the AI finds it hard to obey your prompt if you make this one mistake that few people know about.)
• ​​​Fastest known way to create a complex prompt that the AI will follow. (Even when it’s full of complicated instructions).
• ​​​The honest, no “bull crap” truth about the different AI models and how they’re programmed to obey (or disobey) your prompts.
• ​​​The ugly little secret about what saying “It” causes the AI to do. (This is a tactic I often keep in mind when creating my prompts if I want a high-quality response.)
• ​​​An almost unknown way to “peek” inside the AI’s brain, pulling back the curtain to show how it’s interpreting part of your prompt. (I’ve never seen anyone do this before, and it’s surprising what you can learn about how the AI understands what you’re telling it.)
• ​​​The dead giveaway which “cages” the AI in a box and prevents it from carrying out your wishes.
• ​​​A weird (but effective) way to ask a hundred people a question (without actually asking them) to gauge whether the AI understands a word in your prompt.
• ​​​The major flaw in telling your spouse they look “good” and how the AI doesn’t like that word either.

Listen, AI is just like us in some respects.

No, it doesn’t have feelings or emotions. (You never need to tell it please and thank you. But make sure to tell your spouse those niceties.) However, there are certain types of words that you should avoid (or at least be careful of using) around the AI.

It’s incredible how this works–when you’re prompting the AI and you use these trigger words, it starts acting differently from how you’d expect.

I explain all of this in the training.

But we’re just getting warmed up.

Here are even MORE of the secret tips in this training…

• The fallacy of playing “make believe” with the AI and how it can harm the AI’s ability to give you the output you want.
• ​​The vital importance of “brainpower” on whether the AI can follow a rule classified as “quantifiable”.
• ​​How the “Rule of One” works in helping you achieve consistent, repeatable output. (But don’t use it too much otherwise Sauron might seek you out.)
• ​​The correct way to “validate” a prompt to check whether the AI can give you the output you want.
• ​​A five-second exercise to determine (on a rule-by-rule basis) whether the AI is obeying your prompt.
• ​​Very simple tactic to “stack” prompts (kind of like the game of Jenga) and how it works to help the AI follow even your most complex prompts.
• ​​Sneaky ways to classically condition the AI (like Pavlov’s scientific anticipatory salivation tests on dogs) to break its will and force it to carry out certain “tricky” prompts.
• ​​Cheap tricks to make the AI obey you when it doesn’t want to.
• ​​One dirty little linguistic tactic that lawyers use to wiggle out of contracts, which also happens to apply when prompting the AI. (I found this out in my second year of law school!)

I was never an “ace” student in law school, but I did pick up a few hacks on how to structure a deal, read a contract, and wiggle out of it when the need arises. (Thanks professor Eisler.)

But when programming ZimmWriter, I discovered that the AI succumbs to the same “hacks” when analyzing words. And sometimes…

The Words You Use In Your Prompt
Are Interpreted By The AI Differently
Than What You Wanted!

So I’ll show you how to sniff this out and nip it in the bud.

• ​​​A better way to use AI to get what you want without worrying about finding an old “chat” you had with it.​​
• ​​​Why building a prompt in ChatGPT has one fatal flaw. (But there is an EASY trick to overcome this problem and still create prompts in ChatGPT).
• ​​​An “almost magic” way to use repositioning in your prompt to make the AI obey and provide higher-quality responses.
• ​​​The secret, almost automatic way, to ensure that the AI understands a certain rule you’re telling it to follow.
• ​​​How the AI can get unhealthy (like Morgan Spurlock in Super Size Me) from your prompts and what to do about it.
• ​​​Key strategies for determining which words in your prompt are causing the AI to respond a particular way.
• ​​​How to “sucker punch” the AI and beat it over the head so it complies with what you want it to do.
• ​​​​​A rare known fact: the AI is terrible at randomization and loves patterns. (But this can cause havoc on certain kinds of prompts.)
• ​​​Key strategies for using a “delimiter” with tricky prompts.
• ​​​The real reasons ChatGPT doesn’t obey you.
• ​​​A special way I found to add an almost humanized voice to the AI’s responses through adding “flavor.”

Okay, so that’s a TASTE of what’s in the Get Prompted training.

The entire training is 15 lessons. Here’s the general outline:

• Lesson 1 – Know Your Goal
• Lesson 2 – AI Models
• Lesson 3 – Chat Bias
• Lesson 4 – Cause & Effect Ambiguity
• Lesson 5 – Subjective vs. Objective Words
• Lesson 6 – Too Many Instructions
• Lesson 7 – Prompt Formatting
• Lesson 8 – Few Shot Prompting
• Lesson 9 – Brainpower
• Lesson 10 – Pattern Bias
• Lesson 11 – Last But Not Least
• Lesson 12 – The Rule of One
• Lesson 13 – The Don’t Say “It” Rule
• Lesson 14 – Playing Make Believe
• Lesson 15 – Add Some Flavor

It clocks in at about 179 minutes andalso includes 3 bonus lessons (mentioned below) for an additional 45 minutes.

It’s designed to save you THOUSANDS OF HOURS of your valuable time from having to slog through advanced prompting. I show you exactly what I’ve learned from my 2,600 hours of prompt engineering experience so that you can immediately start making the AI obey your prompts and give high-quality responses.

Anyway, here’s the deal.

I charge $500 an hour as my AI prompt engineering consultation fee. That makes this 179 minute course valued at nearly $1,500.00.

But if you buy it today, the training is only $97.

Now I have to warn you before you get the training, it’s not for lazy people.

If you’re lazy then this isn’t for you.

That’s not to say this training is hard. It’s actually very easy to implement. As long as you’ve used ChatGPT before, then you can easily understand and implement the techniques in my training.

But you have to follow through. You have to actually go through the training and implement the steps I suggest in your prompts.

If you can do that, then you’ll do much better at getting the AI to obey your prompts and deliver high-quality responses.

Reviews of Get Prompted