10 ChatGPT Response Modes

These are custom shortcuts you define. They are not hidden ChatGPT features, and typing a label without its instruction will not reliably produce the same result.

What this helps you do

Switch between direct feedback, visual learning, red-teaming, business critique, natural editing, long-term thinking, misconception checks, prioritization, Socratic teaching, and time-tested advice.

Before you start

Start a new chat and decide what evidence ChatGPT may use. Add source files when factual accuracy matters. Do not paste sensitive information.

Terms to know

• Red team: actively looks for flaws and failure paths.
• 80/20: looks for the few actions likely to create most of the useful result.
• Socratic method: teaches by asking questions instead of immediately giving the answer.
• Lindy: favors ideas with a long record of surviving real use.

Step-by-step

1. Paste the full setup block below.
2. Type one mode with your task, such as /REDTEAM Review this launch plan.
3. Give ChatGPT the facts or files it needs.
4. Verify factual claims and keep the mode only if it improves the result.

Copy this

Create these custom response modes for this conversation:

/TRUTHMODE: Be direct. Separate verified facts, reasonable inferences, and unknowns. Disagree when the evidence supports it.

/VISUALIZELEARNING: Explain the idea as a simple labeled diagram or flow. Then give one concrete example and three recall questions.

/REDTEAM: Find the strongest failure paths, hidden assumptions, and missing safeguards in my plan. Rank them by likelihood and impact.

/HORMOZI: Critique the offer using value, proof, speed, effort, audience, pricing, and distribution. Do not impersonate Alex Hormozi or invent his endorsement.

/HUMAN: Rewrite in plain, natural language while preserving every fact. Remove canned framing, repetition, and corporate filler. Do not claim the result can bypass AI detectors.

/FUTUREYOU: Answer from the perspective of the person I want to become, using only the goals, constraints, and values I give you. Label guesses.

/UNLEARN: List common misconceptions about this topic, explain why each fails, and replace it with a better evidence-based model.

/8020: Find the smallest set of actions most likely to create the target result. Show what you are deprioritizing and why.

/SOCRATES: Teach me by asking one question at a time. Wait for my answer, give brief feedback, then ask the next question.

/LINDYMODE: Start with methods that have worked across many years or settings. Separate durable principles from current trends and cite sources when facts matter.

Use a mode only when I type its label. These are custom instructions, not built-in ChatGPT commands.

Things to know

A response mode changes framing, not the model's access or accuracy. “Truth Mode” cannot force truth. “Hormozi” is a business-analysis lens, not the real person. “Lindy” can undervalue newer methods, while /FUTUREYOU can sound confident about details you never supplied.

If something goes wrong

• If a mode makes up facts, require sources and ask it to label unknowns.
• If two modes conflict, use one at a time.
• If /HUMAN changes the meaning, ask for line-level edits and compare against the original.
• If /SOCRATES gives away answers, repeat “wait for my reply before continuing.”

Final check

Run one factual task through /TRUTHMODE. It passes only if the answer clearly separates what is known, inferred, and missing.