Six questions, and an answer that is allowed to be no. Two of the three possible outcomes do not involve AI at all, which is the point.
What it is doing
This is the order a real discovery conversation runs in, compressed into six questions. It checks the foundations before the opportunity, because a business with an excellent use case and its data spread across four spreadsheets has a data problem wearing an AI costume.
- Foundations first. If the information is on paper or scattered, or the process is not written down, nothing else matters yet. Any one of those is disqualifying on its own.
- Then frequency. What repays automation is how often something happens, not how much it annoys you. Monthly is usually not worth building for.
- Then rules against judgement. A task that is the same every time is a rule, and rules are cheaper, faster and more predictable than a model.
- Then the owner. This decides more than the technology does, and it is the question that most often turns a real use case into a shelved one.
What it deliberately does not do
It does not give you a score. A percentage or a rating out of ten derived from six multiple-choice answers is invented precision, and a number is harder to argue with than a sentence. Instead it shows the specific answers that produced the result, so you can disagree with the reasoning rather than just the verdict.
It also does not collect anything. No email, no sign-up, nothing sent anywhere. Your answers exist in the browser tab and vanish when you close it.
If you land on an outcome that does not sound right for your situation, that is worth a conversation rather than a rethink of the answers. Describe what you are actually dealing with and you will get a straight read on it. How AI work is approached here is set out separately.