Ask any AI a hard business question and you get the same thing: one fluent, confident answer. It reads beautifully. That is exactly the problem.
Confidence isn't correctness. A single answer hides the things that actually matter when you are about to spend money or make a call you can't easily walk back: the strongest argument against it, the second-order effects nobody mentioned, and whether the model quietly invented something to sound complete.
I kept running into this in my own work. The decisions I cared about (a pricing move, whether to chase a new segment, a change to how the team operates) were the ones where a smooth paragraph from a chatbot felt thin. Not wrong, exactly. Just unaccountable. There was no one in the room to push back.
Good decisions come from a room
And that is the thing about good decisions. They rarely come from one smart person thinking alone. They come from a room. The optimist sees the upside. The skeptic goes hunting for the hole. The operator asks who is actually going to do this and what it costs. Someone questions whether you are even asking the right question. The value is not in any single voice. It is in the friction between them.
So I built that room, for myself, out of AI.
Building a room that disagrees with itself
Instead of one model answering, several distinct advisors each take a different lens on the same question. One looks for leverage and upside. One attacks the assumptions. One traces the knock-on effects over time. One stress-tests whether it can actually be executed. One strips the question back to first principles and asks if the framing itself is wrong. Then they do the thing a single chatbot never does: they critique each other, revise their positions, and a chair pulls the whole debate into one clear recommendation.
In plain terms, it is built to disagree with itself, and to tell you when it isn't sure.
The guardrails matter as much as the debate
A room full of confident voices can still talk itself into a bad idea, so the design leans hard the other way. After the chair writes the recommendation, a separate auditor goes back over it and checks the factual claims against the evidence that was actually gathered, then flags anything it cannot support. The disagreement between the advisors is kept, not smoothed away: you see who pushed back and why, in a plain dissent log, instead of a tidy answer that quietly buries the argument. And when the evidence is thin, it tells you that too, rather than dressing up a guess as a conclusion.
You stay in control of the question
This is the part I care about most. You set the context once, your company, your market, the decision you are actually chewing on, so the answers fit your situation instead of a generic one. Before a deliberation spends anything, there is a short window to stop and reword, so a half-formed question costs you nothing. If the answer doesn't sit right, you push back: challenge it, and the council asks for whatever context it is missing, then runs the whole debate again against your objection. You connect your own key and pay only for what you use, your sessions stay private to you, and the tool never makes the call for you. It hands you the argument. You decide.
Where it earns its keep
I used it on my own calls before I showed anyone. A few patterns held up:
- A marketer can drop in a campaign concept or a new positioning line and hear the skeptic explain why it might fall flat before the budget is committed, not after.
- An operations lead can run a vendor switch or a process change through it and get the second-order effects and the execution friction spelled out, the stuff that usually only surfaces three months in.
- Anyone weighing a real bet (build versus buy, entering a market, a reorg, a steep discount) can hear the hardest version of the counter-argument first, in private, instead of in the meeting where it counts.
That last one is the use I keep coming back to. It is a place to lose the argument safely, before losing it gets expensive.
It doesn't replace your judgment
I am not going to pretend this replaces judgment. It doesn't. It does something more modest and, I think, more useful: it makes the disagreement visible, so the judgment stays yours and it is informed. A confident answer tells you what to think. An argument shows you what you are actually choosing between.
A confident answer tells you what to think. An argument shows you what you are actually choosing between.
Try it on a decision you're sitting on
I called it Executive Council. Mostly I just wanted my own decisions to survive contact with a few voices that were not afraid to tell me I was wrong, and it turned out a lot of us are making those calls alone. If you are sitting on a decision like that, you can try it at executive-council.vercel.app. Bring something that actually matters: the harder and higher-stakes the question, the more the room earns its place.