Most people tell AI what to do. The people getting unreal answers tell it how to think first. Here are all seven modes to steal, one line each, and they work in any chat you already use.
Most people open a chat, type what they want, and take whatever comes back. The answer is fine. Forgettable. The people getting genuinely useful answers do one small thing first: they tell the AI how to think before they tell it what to do.
That is the whole trick. These tools match the energy you give them. Ask a vague question and you get a vague, agreeable answer. But open the chat with a single line that sets a role and a way of thinking, and everything after it changes: sharper, more honest, more original. Below are the seven modes I actually paste in, each with a quick note on when to reach for it and the exact line to copy. Steal every one.
How to use these. Paste the mode line at the very start of a fresh chat, hit enter, then ask your real question underneath. That first line sets the tone for everything that follows. You can also stack two when a problem needs both, for example Red Team plus Devil's Advocate before a big decision. It all works the same in ChatGPT, Claude, or any assistant you already use. No settings, no plugins, just the words.
The One Rule
Put the mode line first, above your actual question. Paste it after a long request and the AI has already committed to how it will answer. Set the thinking, then ask.
When to use it: when you have fallen a little in love with your own idea. Red Team mode stops agreeing with you and starts hunting for the crack that would sink it, so you find the flaw before your customers, your boss, or the market does.
When to use it: when you cannot even name the right question yet. Instead of guessing at what you need, the AI interviews you one question at a time and pulls the real problem out of your head before it answers.
When to use it: when safe and obvious will not cut it. This mode trades the polite, middle-of-the-road answer for something with an actual point of view, built to hold attention instead of blend in.
When to use it: when you are buried in detail and cannot see the shape of the decision. CEO mode ignores the small stuff and forces the conversation up to the few choices that actually move the needle.
When to use it: for anything brand new. It assumes you are smart but starting from zero, then explains with plain language, analogies, and examples instead of burying you in jargon.
When to use it: right before you commit to anything. It argues the opposite of whatever you believe, as hard as it can, so you hear the strongest case against your plan while you can still change course.
When to use it: for your knottiest, most tangled problems. It breaks the problem into parts, names the tradeoffs out loud, and hands you a ranked recommendation instead of a shrug.
Notice what all seven have in common. You are not writing a longer prompt or memorizing some secret phrasing. You are handing the AI a way to think before it answers a single thing. That is the actual skill, and it is the one that keeps paying off. The person who tells the model how to reason gets a thinking partner. The person who only tells it what to produce gets a vending machine.
Save the two or three modes you reach for most, then paste, ask, and watch the quality jump. The next time you face a real decision, try stacking two and see how much sharper the answer comes back.
Stack Them
The biggest jumps come from combining modes. Red Team plus Devil's Advocate before you commit. Teacher plus Socratic when you are learning something hard. Two lines, one chat, a noticeably better answer.