A free guide by New Money School
Pro Tips
If you still write prompts that tell AI exactly how to think, you may be making it less intelligent. The smarter models get, the more that old habit gets in the way. Here is the switch.
The prompting advice everyone handed you two years ago is quietly going stale. Follow it today and you might be making your AI less intelligent, not more.
Here is the whole guide in one line. Stop telling AI exactly how to think, and start telling it what you actually want. That is the switch. Writing out every single step used to help a lot. Now, with the smart models we have, it mostly gets in the way. Let me show you why, and exactly how to fix it.
Part 1Then vs Now
A year or two ago, models were not very smart. So you had to walk them through everything. You would write something like: first analyze this, then summarize it, then make it shorter, then rewrite it in this tone. You were the brain, and the AI was just the hands. Back then, that worked. If you left a step out, the model would wander off and lose the thread.
Today's models are on a different level. GPT-5.6 and Claude's Fable 5 are dramatically better at reasoning. They can plan, check their own work, and find the smart path on their own. So when you spell out every tiny step, you are boxing in something that could have found a better route than yours. You are handcuffing the exact reasoning you are paying for.
How vs What
A how prompt sounds like this: write a headline, then three bullet points, then a short paragraph, then a button label for my landing page. A what prompt sounds like this: I need a landing page that gets busy founders to book a demo, here is my product and my audience, build me the page. The second one gives the model room to do its best thinking, and it almost always comes back stronger.
Part 2Process vs Outcome
Here is the clean way to hold it in your head. Prompt engineering describes the process. Intent engineering describes the outcome. The old way: here is exactly how I want you to solve this. The new way: here is the result I want, you figure out the best way to get there.
That is the entire shift. You go from being the micromanager who dictates every keystroke to being the person who sets a clear goal and trusts a capable team to hit it. And the smarter AI gets, the more that difference pays off. On a weak model, dictating steps was a safety net. On a strong model, it is a cap on how good the answer is allowed to be.
Part 3Say This, Not That
Let me make this practical. Here are a few quick rewrites. Notice how each one drops the steps and names the result instead.
- Instead of "first research my competitors, then list their prices, then compare them in a table," say "I want a clear picture of how my pricing stacks up against my top competitors. Show me what matters most."
- Instead of "write an intro, then five tips, then a conclusion," say "Write a blog post that makes a nervous beginner feel confident about starting. You decide the shape."
- Instead of "summarize this, then pull the action items, then format them," say "Turn this messy meeting transcript into something my team can actually act on tomorrow morning."
- Instead of "use a formal tone, short sentences, and no big words," say "Write this so a smart twelve year old gets it on the first read."
Part 4The Nuance
Now the honest caveat, because this is where people get it wrong. Outcome based does not mean vague. "Make me a website" is not intent engineering. It is just a lazy prompt, and you will get a lazy result right back.
You still give rich context. You still give real constraints, like a word count, a budget, a brand voice, or a hard rule the model cannot break. You still hand over an example or two of what good looks like. And you still give a clear definition of done, so the AI knows when it has actually finished. The only thing you drop is the step by step method.
Think of it like briefing a brilliant freelancer. You give them the goal, the context, the guardrails, and a picture of what success looks like. You do not stand over their shoulder telling them which key to press next. You brief them well, then you get out of the way.
Part 5The Converter Prompt
Want a shortcut? I built a prompt that does this switch for you. Paste in any of your old how prompts, and it hands you back a cleaner outcome based version. This is the same converter I gave away with the video, so grab it and keep it.
CopyIntent Engineering Converter
You are an expert prompt engineer. I am going to paste one of my prompts below. Your job is to rewrite it as an outcome based prompt, so a smart model can do its best work instead of just following my steps.
Do this in order:
1. Read my prompt and find the true goal. What result am I actually after? Say it back to me in one plain sentence. 2. Strip out any step by step instructions that only dictate HOW to do the work. Trust the model to choose the method. 3. Keep every real constraint, fact, example, and piece of context. Do not lose these. They matter. 4. Add a clear definition of done, so it is obvious when the result is good enough to ship. 5. Rewrite the whole thing as a single prompt that describes the outcome I want, the context you need, and what a great answer looks like.
Return two things. First, the rewritten prompt in a clean block I can copy straight out. Second, one short line telling me what you changed and why it will get a better result.
Here is my prompt: [PASTE YOUR PROMPT HERE]
Your Homework
Go audit your five favorite prompts. Every time you catch yourself telling AI exactly how to solve something, stop and ask: could I just describe the result I want instead? Do that for a week, and you will feel your AI get smarter, because you finally stopped getting in its way.