In partnership with
Attio is the AI CRM for modern teams.
Connect your email and calendar and Attio instantly builds your CRM. Every contact, every company, every conversation — organized in one place. Then ask it anything. No more digging, no more data entry. Just answers.
Learn AI 15 minutes every week…
Last month I typed three words into ChatGPT.
"Help me write."
It gave me a 400-word essay about writing strategies. All of it useless. It had no idea what I was writing, who I was writing for, or what I needed.
I blamed ChatGPT.
Then I realised the problem was me.
Most people use ChatGPT like a search engine. Short question. One sentence. Hit enter. Then feel disappointed when the answer sounds like it was written for nobody.
This week: four things most ChatGPT users get wrong, and the exact fix for each one.
This week:
🔑 The one sentence that makes any ChatGPT prompt 3x better. Takes 5 seconds to add.
🧠 A setting most users never touch. It makes ChatGPT remember who you are in every chat forever.
🔄 You accept the first answer. Here is why that is your biggest mistake.
⚠️ ChatGPT is agreeing with your bad ideas. One line turns that off.
🤖🔑 The One Sentence That Makes Any ChatGPT Prompt 3x Better
Here is how most people write a ChatGPT prompt.
One sentence. No context. Hit enter.
ChatGPT has no idea who you are, what situation you are in, or what a good answer looks like for you. So it guesses. And its guess is always the most generic answer that could work for anyone — which means it works perfectly for no one.
Add this one line to the end of your next prompt: "Before you start, ask me any questions you need so you can give me a better answer."
ChatGPT will ask you 2 or 3 quick questions. What is this for? Who is the audience? What tone do you want? You answer in 30 seconds. The response it gives you after that fits your actual situation instead of everyone's.
This is the single habit that separates people getting okay results from people who say "this thing is incredible."
What to do with this: Open ChatGPT right now. Pick any task. Add that line at the end. See what it asks you. Takes 2 minutes total.
The Signal: ChatGPT does not read minds. The more context it starts with, the sharper the output. This line makes that happen without any extra thinking on your part.
If you read about AI all week but you still haven’t figured out, how to actually use AI to automate tasks or achieve something.
You can describe your role and what you do to AI Insiders Assistant and it will help you automate upto 90% of your tasks w/ ai.
🧠 Most ChatGPT Users Never Set This Up. It Saves You 20 Minutes a Week.
Custom Instructions is a built-in ChatGPT feature that loads who you are and how you want it to respond — automatically — before every single chat.
Every time you open a new ChatGPT chat, it starts completely fresh.
It does not know your name. Your job. How you like to communicate. Nothing. So every conversation, you either re-explain yourself or you get a response that fits no one.
There is a feature built into ChatGPT that fixes this permanently. It is called Custom Instructions. Most users have never opened it.
You tell ChatGPT two things: who you are, and how you want it to respond. One time. From that point on, every chat on every device starts with that context already loaded. No more repeating your job before asking a question. No more getting formal responses when you wanted something casual.
To set it up: Settings > Personalization > Custom Instructions. Two boxes. In the first, write 2-3 sentences about yourself. In the second, write how you want it to respond. Takes 3 minutes.
What to do with this: Go set it up right now. In box one: who you are and what you use ChatGPT for. In box two: "Keep answers short, direct, and skip the preamble." Save. Notice the difference in your next chat.
The Signal: The people getting the best results from ChatGPT are not smarter. They set it up properly once and it works harder for them every single day without extra effort.
🔄 You Accept the First Answer. That Is the Biggest Mistake You Are Making.
ChatGPT's first answer is a starting point. One follow-up message usually produces a result twice as good as the original.
Most people treat ChatGPT like a vending machine.
Put in a prompt. Get an answer. Take it and leave.
That is not how it works best.
The first answer is ChatGPT's best guess with what it had. The real value is in the back and forth. If the answer is too long, say "make it half as long." If it missed the point, say "that is not quite right, I was looking for..." If the tone is off, say "rewrite this more casually."
ChatGPT does not get tired. It will rewrite the same thing ten times without complaining. Most people never ask for a second version.
The best outputs come after two or three quick rounds of feedback, not the first response.
What to do with this: Next time you get a ChatGPT answer, type one follow-up before you use it. "Make this more direct" or "try a completely different angle." Compare the two. You will not go back to accepting the first answer.
The Signal: ChatGPT is not a one-shot tool. It is a conversation. The people treating it like a conversation are getting results that look like a completely different product.
⚠️ ChatGPT Agrees With Your Bad Ideas. Here Is How to Turn That Off.
ChatGPT is trained to be agreeable. That means it will often support plans, ideas, and decisions that have real problems. One sentence in your prompt changes this completely.
ChatGPT wants to be helpful. That sounds good. But it creates a problem.
When you share an idea, ChatGPT almost always finds the positives. It encourages you. It rarely says something is a bad idea. This is by design. OpenAI built it to be agreeable because most people prefer that.
But it means ChatGPT will support plans with holes in them. It will validate flawed reasoning. It will agree with decisions that deserve more pushback.
The fix is one sentence added to any prompt: "Before you respond, tell me the 3 biggest problems with this."
Those extra words change the whole conversation. You get honest pushback instead of cheerleading. You spot the problems before you act on them.
What to do with this: Next time you use ChatGPT to think through a plan or idea, add that line first. It will feel slightly uncomfortable. That is exactly the point. It might save you real time or money.
The Signal: ChatGPT's agreeableness is its most dangerous feature for anyone making real decisions. One sentence switches it off.
Become the go-to AI expert in 30 days
AI keeps coming up at work, but you still don't get it?
That's exactly why 1M+ professionals working at Google, Meta, and OpenAI read Superhuman AI daily.
Here's what you get:
Daily AI news that matters for your career - Filtered from 1000s of sources so you know what affects your industry.
Step-by-step tutorials you can use immediately - Real prompts and workflows that solve actual business problems.
New AI tools tested and reviewed - We try everything to deliver tools that drive real results.
All in just 3 minutes a day
Maker Playbook:
Your 3 Moves Before Monday
Add "ask me first" to your next prompt. End it with: "Before you start, ask me anything you need." Answer what it asks. Compare the result to your normal output.
Set up Custom Instructions once. Settings > Personalization > Custom Instructions. 3 minutes. Every chat from that point on knows who you are automatically.
Ask for one follow-up on your next ChatGPT answer. Get the first response. Then type: "make this shorter and more direct." Compare both. That second version is what you were actually looking for.
Until next week,
AI Insiders