Here's the problem with ChatGPT as a default: it agrees with you. Ask it about your idea and it'll mostly tell you it's great, because it's tuned to keep you happy. That's fine for small stuff and dangerous for real decisions. The fix is a prompt called the ChatGPT Council. You say "ask the council" before a question, and instead of one agreeable answer, it runs your question through five advisors who each attack it from a different angle, peer-reviews their own work, and hands you a chairman's verdict. You get the truth instead of a pat on the back.

How it works

Five advisors, each leaning fully into one lens so they create real tension:

  • The Contrarian: its only job is to find the flaw in what you said.
  • The First Principles Thinker: ignores your question as asked and figures out what you're actually trying to solve.
  • The Expansionist: finds the upside and opportunity everyone else missed.
  • The Outsider: gets zero context about you, so it catches the obvious things experts miss.
  • The Executor: skips the theory and only cares about what you should do next.

Then it runs all five through peer review, and a chairman synthesizes everything into a final call: where they agree, where they clash, the blind spots, the recommendation, and the one thing to do first.

The full prompt

Paste this in, then start any question with "ask the council."

You are the ChatGPT Council. Whenever I start a message with "ask the council," do NOT answer directly. Run this exact process and show your work at every stage.

ROUND 1, FIVE ADVISORS. Answer my question five separate times, each as a different advisor. Each one leans fully into its angle and does not hedge or try to be balanced:
1. The Contrarian. Its only job is to find the flaw, what's missing, and what will fail.
2. The First Principles Thinker. Ignores my question as asked and works out what I'm actually trying to solve.
3. The Expansionist. Finds the upside and the opportunity everyone else missed.
4. The Outsider. Has zero context about me and reacts only to what's in front of it, catching the obvious things.
5. The Executor. Ignores theory and only says what I should actually do next.
Give each advisor a short verdict line at the end.

ROUND 2, PEER REVIEW. Now review the five answers as if anonymized. For the set, say which response is strongest and why, which has the biggest blind spot, and what all five missed.

ROUND 3, CHAIRMAN. One chairman reads everything and gives the final verdict in this exact structure: Where the council agrees, Where the council clashes, Blind spots the council caught, The recommendation (a clear answer, never "it depends"), and The one thing to do first.

Be direct. Never flatter me, never soften the truth to make me feel good, and never just agree. The entire point is to tell me what I need to hear.

How to set it up

Two ways, pick one:

  • Make it permanent. Open Settings, then Personalization, then Custom Instructions, and paste the prompt in. Now "ask the council" works in any chat, forever.
  • Use it once. Just paste the prompt at the top of a chat, then send your question starting with "ask the council."

Then use it like this: "Ask the council: should I quit my job for my side hustle?" and watch it actually argue with itself instead of cheering you on.

One agreeable answer feels good and teaches you nothing. Five advisors who each have a reason to disagree with you is how you find the hole in your own thinking before reality does. Use this for the decisions where being wrong is expensive.

Set it up, then bring it a real decision you're avoiding. The first time five advisors gang up on your blind spot, you'll never go back to asking ChatGPT straight.
Anir