Most people type a question into ChatGPT and get one answer. Confident, polished, safe.
Then they accept it and move on. They never find out that the same question, with one word added before it, comes back completely different. Sharper. Deeper. Honest. Sometimes unrecognizable from the original.
These are 10 words and phrases you type before your prompt. Each one changes how ChatGPT processes your question. Not what you ask. How it thinks before answering.
Try all 10 on the same question and watch 10 different answers come back.
1/ /human
Put this before any writing request and the output comes back humanized: natural rhythm, no robot phrasing, the kind of text that reads like a person wrote it.
The prompt:
/human
Rewrite this so it reads like a real person wrote it, not an AI. Natural sentence rhythm, no corporate phrasing, no filler words like "delve" or "landscape" or "it's important to note." Vary the sentence lengths. Some short. Some longer where the thought needs room. Cut anything that sounds rehearsed or polished for the sake of polishing. Keep every fact and point, just make it sound like someone talking, not someone performing.
The text: [PASTE YOUR TEXT]This is the difference between output you paste and output you actually send. Run everything through /human as the last step before it leaves your hands.
2/ /red team
Put this at the end of any idea and ChatGPT critiques it so aggressively it exposes every single weakness before the real world does.
The prompt:
Here is my idea: [PASTE YOUR IDEA, PLAN, STRATEGY, OR DRAFT].
/red team
Attack this like your only job is to find what is wrong. Do not be balanced. Do not be encouraging. Find the fatal flaw, the assumption I have not tested, the part that sounds good on paper but falls apart in execution. Name the 3 biggest reasons this fails, who would reject this and why, and the one thing I am clearly too close to see. Then tell me the minimum viable fix for each weakness, so I walk away with something stronger, not just something criticized.Most people ask ChatGPT "is this a good idea?" and get a yes. This forces it to find why your idea is a bad one. The ones that survive /red team are the ones worth building.
3/ /meta
Put this at the start and ChatGPT zooms out and analyzes your question before answering it, which is why the answer comes back ten times better.
The prompt:
/meta
Before you answer my question, analyze it first. Tell me: what am I actually trying to solve here, what assumption am I making that I might not realize, and what would a better version of this question look like. Then answer the better version instead of the one I asked. If my original question was already the right one, say so and answer it directly.
My question: [YOUR QUESTION]Most people ask the wrong question and get the right answer to it. /meta forces ChatGPT to fix the question before it wastes your time answering the wrong one.
4/ /alt 3
Add this to any request and you get three completely different versions of the answer, not one safe take.
The prompt:
[YOUR REQUEST]
/alt 3
Give me three completely different versions of this. Not three rephrasings of the same idea. Three genuinely different approaches, angles, or strategies. Make version 1 the safe, conventional answer. Make version 2 the unexpected angle most people would not think of. Make version 3 the bold one that might be wrong but would be a breakthrough if it is right. Label each one and tell me which one you would bet on and why.One answer is one perspective. Three answers is a decision. The bold version alone is usually worth more than the safe one ChatGPT would have given you by default.
5/ /truth
Put this before your prompt and ChatGPT stops guessing, stops filling gaps, and admits what it does not know instead of hallucinating.
The prompt:
/truth
Answer this with maximum honesty. If you are not sure about something, say "I am not sure" instead of guessing confidently. If the real answer is "it depends," say that and explain on what. If you are making an assumption, flag it. If a claim needs a source and you do not have one, say so instead of inventing one. I would rather have a shorter, honest answer than a longer one that sounds confident but is partly made up.
My question: [YOUR QUESTION]ChatGPT is trained to sound certain. /truth breaks that. You start getting answers with "I am not confident about this part" and "this is my best guess, not a fact." That honesty is worth more than a polished lie.
6/ /eli10
Put this before any topic and ChatGPT explains it like you are 10 years old. Complex stuff finally clicks.
The prompt:
/eli10
Explain [TOPIC] to me like I am a smart 10-year-old. Use simple words, real examples from everyday life, and zero jargon. If you have to use a technical term, define it in plain language right there. Break it into small pieces. After each piece, check if the explanation makes sense before adding the next layer. End with the one thing most adults get wrong about this topic, explained simply.This is the prompt that makes you finally understand the thing you have been nodding along to for years. Mortgages, stocks, legal terms, tax rules, anything. If it clicks for a 10-year-old, it clicks for real.
7/ /predict
Put this before your prompt and ChatGPT predicts the likely outcomes based on everything it knows about your situation.
The prompt:
/predict
Here is my situation: [DESCRIBE YOUR SITUATION, DECISION, OR PLAN IN DETAIL].
Based on this, predict the 3 most likely outcomes if I go ahead. For each one give me: how likely it is (percentage), what leads to that outcome, and the one thing that would push it from likely to almost certain. Then tell me the outcome I am probably not considering at all, the blind spot, and what would trigger it. Be honest about confidence levels. Where you are guessing, say so.You stop hoping and start seeing the map. The blind-spot prediction is usually the one that saves you.
8/ /OODA
Put this at the end of any problem and ChatGPT runs a military decision framework on it: observe, orient, decide, act, and tells you exactly what to do.
The prompt:
Here is my problem: [DESCRIBE THE SITUATION].
/OODA
Run the OODA loop on this:
- Observe: what is actually happening right now, stripped of my assumptions and emotions. Just the facts.
- Orient: what context am I missing, what has changed recently, and how does this compare to similar situations.
- Decide: given the observation and orientation, what are the real options and which one wins.
- Act: the exact first move, today, in one sentence. Not a strategy. The step.
Then tell me what most people do at this stage that looks smart but is actually a mistake.Designed for the moments when you are stuck in your own head. /OODA forces the answer through a filter that was built for high-pressure decisions where overthinking gets you killed.
9/ /x10 think
Add this and ChatGPT thinks ten times deeper before answering. Same question, noticeably smarter output.
The prompt:
/x10 think
Before you answer, think about this 10 times harder than you normally would. Consider edge cases, second-order effects, things most people miss, and the non-obvious implications. Do not give me the surface-level answer. Give me the answer someone would arrive at after sitting with this question for an hour, not 3 seconds. If the deep answer contradicts the obvious answer, tell me both and explain why.
My question: [YOUR QUESTION]The default answer is the fast one. /x10 think gives you the slow one. Use this for anything where being wrong costs more than being late.
10/ /socrates
Put this before your question and instead of answering, ChatGPT asks you questions one at a time until you figure out the answer yourself. The best one for big decisions.
The prompt:
/socrates
Do not answer my question. Instead, ask me one question at a time that forces me to think through the answer myself. Do not move on until I have responded. Start with the most fundamental question, then go deeper based on my answers. If I am wrong, do not correct me directly, ask a question that makes me see the flaw myself. Keep going until I arrive at a clear answer, or until I realize I was asking the wrong question entirely.
My question: [YOUR QUESTION]This is the one that changes how you think, not just what you know. Use it for the decisions you keep going in circles on. The answer was always yours. /socrates just pulls it out.
Start tonight. Take one question you asked ChatGPT this week. Ask it again with /truth in front. Then again with /red team at the end. Then again with /x10 think. Watch three completely different answers come back from the same question.
While everyone else types one question and trusts one answer, you will type one word and get the real one.
🚨 Reply "BIZ" and I will send you my 30-prompt business pack. Not a lecture on AI productivity. 30 prompts that do the boring work and hand you something you can sell, send, or deploy tonight.
Cheers,
Matas Jonaitis
Founder, DailyChatGPT (@dailyprompter)