# The War Council: a copy-paste template for making better decisions with AI Made by Bluhook (https://bluhook.com). Free to use, copy, and share. You are about to run a "war council": instead of asking one AI for a balanced opinion, you make several AI voices argue from opposing value systems, then you read the disagreement and decide. A single balanced answer averages away the tradeoff that mattered. This surfaces it. It works in any chatbot (ChatGPT, Claude, Gemini). No tools, no setup. --- ## How to run it in 10 minutes 1. Write your decision as ONE question with two named options. Not "what should I do about marketing." Instead: "Lead Q4 with the loyalty-program launch, or the new- customer discount?" A vague question gives you vague answers. 2. Pick 4 to 5 "seats." Each seat is one value system that has a real stake in the call. Use one of the ready-made panels below, or write your own. 3. For EACH seat, open a FRESH chat and paste the SEAT PROMPT below, filled in. Fresh chats matter: they stop the AI from quietly balancing across the seats. 4. Collect the five answers. Paste all of them into one final chat and run the SYNTHESIS PROMPT. 5. Read the synthesis. Then YOU decide. The council advises. It never decides. --- ## The seat prompt (paste once per seat, in its own fresh chat) ``` You are the [SEAT NAME] on a decision-making war council. You care ONLY about: [WHAT THIS SEAT VALUES]. Here is the situation and the two options: [PASTE 3 TO 6 SENTENCES: what the business is, who the customer is, the decision, and the two named options.] Argue ONLY from your seat. Do not be fair. Do not hedge into anyone else's concern. Do not write a balanced take. Make the strongest honest case your value system would make for one option over the other. Return exactly: POSITION: your one-sided argument. VERDICT: Option A or Option B (or "conditional" with the one condition). BIGGEST RISK: the single thing most likely to go wrong if we follow your advice. ``` --- ## The synthesis prompt (paste once, with all five answers) ``` Below are five verdicts from a war council, each arguing from a different value system, on this decision: [RESTATE THE DECISION AND THE TWO OPTIONS]. [PASTE ALL FIVE POSITION / VERDICT / BIGGEST RISK ANSWERS.] Do not average these into a recommendation. Instead give me: 1. TALLY: how many for each option. 2. WHERE THEY AGREE: any point most seats converged on, especially a shared risk named by seats that disagreed on the verdict. 3. THE SHARPEST INSIGHT: the single thing this panel surfaced that one balanced answer would have hidden. 4. THE STRONGEST DISSENT worth keeping. 5. Is the question itself wrong? If a seat implied we are solving the wrong problem, say so plainly. Keep it short. I will make the call myself. ``` --- ## Ready-made panels ### Panel A: the Marketing Council Use for campaign bets, launch positioning, brand-voice shifts, channel and budget allocation, the messaging angle that defines a quarter. - Brand Guardian: cares only about long-term brand equity and voice integrity. - Performance: cares only about the funnel, cost to acquire a customer, and whether the math closes. - Growth and Distribution: cares only about reach and whether the idea can travel. - Audience Skeptic: is the target customer, unimpressed, asking why anyone should care. Runs the swap test: could a competitor say the exact same thing? - Contrarian: attacks the option everyone is drifting toward, and questions whether the whole framing is right. ### Panel B: the Business / Strategy Council Use for pricing, hiring, whether to take on a big client, build vs buy, entering or leaving a market. - The CFO: cares only about cash, margin, and downside. What breaks us if it goes wrong? - The Operator (wartime COO): cares only about whether we can actually execute this with the people and time we have. - The Growth lead: cares only about the upside and the size of the prize. - The Customer: the person who pays, asking whether this makes their life better. - The Contrarian board member: makes the strongest case for NOT doing it, and names the assumption everyone is taking for granted. ### Build your own panel The seats are just value systems with a stake in the call. For a hiring decision: the budget hawk, the team-culture guardian, the speed-to-productivity lens, the skeptical future teammate, the contrarian. For a product cut: the power user, the new user, the support burden, the roadmap cost, the contrarian. Pick the four or five tensions your decision actually pulls on, and make each one argue. --- ## The two rules that keep it useful 1. It is for decisions that are expensive or hard to undo. Do not convene a council to pick a font. Five arguing voices on a tiny call is theater, and it produces worse work, not better. 2. A human always decides. If you let the AI collapse the five seats back into one recommendation, you are back to the mushy middle. The point was to see the disagreement before you choose. Made by Bluhook. We build custom AI systems that do the work, and we run them for businesses. If this was useful, the full write-up and more are at https://bluhook.com/learn.