AI Decision Maker
Describe the decision. Four rival AIs argue every side, weigh what actually matters, and land on a clear recommendation — not a fence-sit.
A weighted pros-and-cons analysis with a clear recommendation.
Signed off by an independent Final QA audit — with a “verify before you rely on this” list, never a black box.
Why this team
A pros-and-cons list fails because every item weighs the same on paper, and the mind that wrote it already has a favourite. The Synthesiser forces genuine weighting instead of counting, the Optimist and the Devil’s Advocate make sure both columns are argued at full strength rather than one side written by a lawyer and the other by a hostage, and the Risk Officer separates recoverable downsides from the one that is not. The Pragmatist hunts the hidden option every binary list conceals — usually “neither, yet” or “both, sequenced”.
Each seat runs on a different frontier model — GPT, Claude, Gemini, Grok — so it’s a genuine cross-model review, not one AI in costumes. How a run works →
What the debate sounds like
An illustrative excerpt — A freelance photographer is deciding whether to sign a year-long exclusive contract with an agency that guarantees steady income but forbids the direct client work she has spent five years building.
The cons column is missing its heaviest item: exclusivity does not pause the direct-client business, it decays it. Clients who are told “ask me again next year” find someone else and mostly do not return. So the real trade is not one year of freedom for one year of stability — it is stability now against a client base that took five years to build and will not take one to rebuild. Weigh that item, not “less variety”, which is what the list currently says.
Argued at equal strength, the pro side is undersold too: guaranteed income for a year is not just comfort — it is the first chance in five years to work without the anxiety tax, and the agency’s clients are a tier she cannot currently reach. A year of that portfolio and those relationships compounds as well. The steady cheque is the least of what is being bought.
Asymmetry check, which is what actually decides close calls: the downside of signing — client decay — is severe but partially recoverable, and its worst case is known. The downside of declining is a bad year, fully recoverable. But note the reversibility trap in the contract itself: as described there is no exit clause, so signing is the one choice that cannot be adjusted mid-course. Irreversibility deserves its own weight, separate from probability.
Questions
How is this different from the pros-and-cons list I already made?
Your list gets used as the input — then the council weighs the items instead of counting them, argues the weak column at full strength, and hunts what the list omits. In the example above, the decisive item (client decay) and the real option (a carve-out) were both absent from the original list. That is typical.
What does a “weighted” verdict actually look like?
Not equal-length columns: the synthesis names the two or three items that actually carry the decision, states the asymmetry — which downside is recoverable and which is not — and commits to a recommendation with a confidence level. Ten trivial pros never outweigh one irreversible con, and the format makes that visible.
When is this brief the wrong tool?
When the decision is really a values conflict wearing a logistics costume — those belong in the Life-Decision Council, where the deeper question gets named. If that is what is happening, this debate tends to detect it: a list where the items refuse to weigh cleanly usually means the real decision is elsewhere.
Your material is used only to run your review — never to train public models. Encrypted in transit and at rest. Security & privacy →
Want full control — pick your own minds, set the depth? Open the full council →