The Council

Problem

AI sounds confident even when it’s wrong, and it loves to agree with you. If your team makes decisions with those answers, confident nonsense can cost real money.

Why I built it

I needed a check I could inspect before I acted. Asking one AI to grade another, in the same chat, usually produces manners, not a verdict.

What I built

Before I trust an AI answer on anything important, I make two AIs argue it out: one defends the idea, one attacks it, and a scorekeeper that can’t be charmed keeps the record. It’s quality control for AI answers, and it’s one reason my clients can trust what I ship.

How it works

The scorekeeper is the point. Without it, you just have a conversation. With it, you have a review you can look at later. This is a tool I use. It is not a product you can buy, unless that changes.

Why it matters

If your team makes decisions with AI, you need a check that catches confident nonsense before it costs you money. The Council is that check for my own work.