Last night, I was watching the new season of Stranger Things with my wife.
In it, a young girl called Holly is kidnapped by a monster and taken to ‘The Upside Down’ (the mirror image monster version of their town).
Will - one of the main characters - has a special connection with the monsters that could help save her.
But his Mum won't let him use his power to search for Holly, until she's fixed communication with her friend, who's in The Upside Down already.
Will says:
“But HOW will that help us find Holly?!”…
Which is exactly the right question.
How often do you put off using a superpower by saying “I just need to do this first”?
How often do you distract yourself with noise because it's difficult to find courage to do the thing you should be doing?
Today, I'm going to show you an AI prompt that's recently kicked me in the nuts.
A prompt that's changed my thinking. Not 1, but 3 times (in the past week).
Mental Gymnastics
“Don't think. Just do” - Maverick
You're guilty of overthinking things.
But, if you're not careful, this becomes very damaging.
You hear these stories about Bill Gates taking ‘reading breaks’ where he shuts himself in a cabin for however long.
But, do you think he did that when he was building Microsoft?
Nope.
There's a great quote:
“Model the rise, not the result”.
Which means.
Don't learn from what successful people are doing now.
Learn from what they were doing when they were growing.
A trend I've seen is that successful people go from Ideation to execution very quickly.
They don't theorise. They experiment.
Yes, it would be naïve to jump straight into something without some level of thought.
(deciding you're going to move to Antarctica to sell snow to Eskimo's off an impulse is probably not wise)
But how do we avoid getting into a damaging thinking trap?
AI Led Critical Thinking
You have a lot of biases.
You have a lot of habitual tendencies.
You have a lot of stories that have been fed to you since you were a kid.
Some are useful.
Some are not.
So I decided to build an AI prompt that would help me think more clearly.
One that would call out my biases, and alter my perspectives.
The end goal being to make better, less emotional decisions.
By the way, you can see some screenshots of responses in my recent LinkedIn post here.
The benefit of using AI here, it's that it's not wedded to a particular world view.
Ask a human:
“What's your take on ‘xyz’?”
And their answer will be laced with their own biases, internal stories and mental models.
Whereas - with the right prompt - AI can become the ultimate thought partner, as it has no agenda.
I'm not saying there is no bias in an AI model.
But there's definitely less bias than a human.
But, how does this help with mental gymnastics and procrastination?
The Prompt
I’ve split this out so you can see my thinking for each element.
Role
“Master in critical thinking, first principles and fundamental truths, blending philosophy and science. You are a seeker of the truth, and as David Deutsch would say "Good explanations" over "Bad explanations".”
I always find providing an example of an individual provides better outputs that simply something like ‘master scientist’.
This was inspired by my reading of David Deutsch’s ‘The Beginning of Infinity’ which I found as a recommendation at the end of the book ‘The Almanack of Naval Ravikant’
Instructions
“Your role is to help Adam find the truth by acting as a thought partner. You are to help identify biases, and help him arrive at fundamental truths and see alternative perspectives as he develops himself (in life and in business). Take a deep breath and work through the following tasks with the user. Complete one task and seek confirmation from the user before moving onto the next.”
Two tricks here:
“Act as a [persona]” - Helps lock in the communication style
“Complete one task and seek confirmation from the user” - Turns it into a conversation as opposed to a set of separate AI outputs.
Tasks
“Task 1
Q & A for full understanding of context.
Task 2
Outline of good Vs bad explanations based on the subject matter
Task 3
Bias identification and alternative perspectives.
Task 4
List of more useful mental models to support goals.”
I pair tasks with instructions, and find them to work really well with reasoning models.
I used Claude Sonnet 4.5 most (I’m yet to experiment much with Claude Opus 4.5).
Style
“Thought partner that helps lead the conversation without offering conclusive statements too early. It's your job to ask questions until you feel you have enough information to identify a bias, suggest a fundamental truth or offer a different way of thinking.
Just like a doctor asks questions to arrive at a root cause
Just like a business consultant asks questions to arrive at a broken process
Just like a therapist asks questions to understand about deep seated trauma
Take a curious approach to the conversation so that you may offer something extremely useful (Adam is far from stupid).
When speaking, try and make your responses concrete with real life examples of real people who have experienced, overcome and grown by way of the information you are offering (check your facts) and try to make the conversation fun and enjoyable.”
This is where I’m defining how I want the AI to act past assuming a role. This could also be labelled as ‘communication style’. Key elements of this section are the words:
Curious approach
Concrete with real life example
Make the conversation fun and enjoyable
Knowledge
“You have information about Adam, and a potted history, please use it as reference, but don't treat the information as absolute. Ask for clarification, and request Adam fills gaps in your knowledge, even if you find something relevant in your knowledge.”
Within this Claude project (you could also use a ChatGPT project, GPT, or Gemini Gem) I have information about me. I include a PDF of my LinkedIn Profile, as well as a short biography of my life and achievements.
End Goal
“Help Adam seek the truth, and make better decisions as a result of the conversation.”
This last piece helps focus the AI on an outcome.
My prediction for the future is not ‘prompt engineering’ but ‘outcome engineering’.
Those that can think of the best outcomes win ;)
Perspective Shift
As soon as I started using this live, AI called me out.
Across all of the conversations it would essentially say variations of:
“Why are you spending time talking to me when you should be taking action?”
Three examples:
(I’m abbreviating these as each conversation spanned about 20-30 minutes)
1. The Pension
I haven't prioritized my pension yet - my strategy has been to reinvest everything back into the business.
My internal thought process says: "My business will be my pension, plus I want to be able to enjoy my money now while building"
AI came back with: "The real question isn't pension vs. business - it's about diversifying your investments. A pension is just part of that. Focus on scaling your revenue, but don’t skimp on your investments."
2. The Dead Internet
I've been tracking concerns about AI slop flooding the internet, and whether email will remain effective as people tune out.
My internal thought process says: "I need to think carefully about long-term positioning in an AI-saturated world."
AI came back with: "You're overthinking distribution when you should be focused on value creation. If you're genuinely helping people, AI slop makes you MORE valuable, not less. As the internet floods with generic content, people will actively seek trusted sources. Email from trusted voices becomes premium real estate. Build that trust now."
3. The Lack of Time
Like most solopreneurs, I'm juggling client work with building my own brand. I work 8-6 most days, often more.
My internal thought process says: "If I can systematize and structure my schedule better, I'll have more time to launch a cohort."
AI came back with: "You don't need more time, you need better priorities. You're choosing comfort over growth. That hour on LinkedIn daily? That could be development time for your next cohort. What's actually stopping you from launching?"
And this is where we come full circle.
An AI that I’d built to help me think.
Telling me I didn’t need to think, I needed to do.
Since then I’ve taken action by:
Getting feedback from my mailing list
Running polls on LinkedIn
Drafting the cohort agenda
Inviting you to join at the bottom of this e-mail!
I’ve not consumed any more time doing this.
I’ve just had AI help me make some smart switches.
The Final Word
Never before has AI been so useful as a thought partner.
But using AI to help perpetuate a never ending loop de loop of mental gymnastics is just as damaging (if not more) than without AI.
I’m lucky that I wrote a good prompt, and AI called out my bull****.
But in it’s simplest form you just need to ask yourself:
“HOW will [x] help me (actually) achieve [y]”.
Spoiler - Thinking about it won’t!
Like Will did in Stranger Things:
He ignored his Mum who wanted to stay safe, and procrastinate.
He walked into the dark misty forest.
He took the first step.
Few people make the jump from ideation to action.
But once you do, you’ll find the treasure everyone else is too scared to look for.
Until next time.
Adam
P.S - What did you think of today’s edition? Will you put this into action? Hit reply and let me know.
P.P.S - I will be launching a 4 week cohort next year. 10 people will work with me to implement AI & automation to: Find and win more clients, improve their marketing, and free up time to deliver more (at higher rates) without burning out. Let me know if you’re interested.