How it works
From question to a decision you can defend
Four steps, published rules, and every number traceable to the input behind it.
- 1
Answer plain questions
About your venture, or about each AI initiative: what it claims, what it costs, who owns it.
- 2
Attach your evidence
Customer interviews, pilots, A/B tests, production data. Each type counts for a published amount; vendor claims count least.
- 3
Read the decision
Ranges, not single numbers. What is proven, what is assumed, and the next test that would change the answer.
- 4
Take the paper to the meeting
A formal board paper in Word with your logo, and every figure traceable to the inputs behind it.
Why trust it
Built to be checked
The same answer every time
Fixed, published rules: the same inputs always give the same result.
Evidence, not opinion
An AI chat tends to agree with you. Here every claim is weighed by the strength of the evidence behind it.
Every setting graded
Each setting the engines use is recorded with how strong the case for it is, so you can challenge it.
Go deeper
Read the method, or open the examples
The method
How evidence is weighted, how ranges are set and how decisions are suggested.
OpenExplaining your results
A companion guide for presenting a result to a board, an investor or a committee.
OpenWorked venture example
A finished venture check with its inputs shown, so you can see how the answer was reached.
OpenWorked AI example
A ten-initiative AI portfolio, from claims and evidence to scale, pilot or pause.
OpenSample board paper
The same AI example as an 18-page committee paper in Word.
OpenFor developers and agents
The REST API and MCP server, with schemas and examples.
Open