Trust
Trust & method
How the systems produce results, where the data comes from, and how we correct mistakes in public.
01
Methodology
Each product grounds its output in a named dataset with a snapshot date. Where a number appears on this site, it is rendered at build time with an as-of stamp so it can be cited as a fact, not a claim.
02
Data provenance
The four systems read original data: live registries and sanctions lists, agent-simulation outcomes, real build-cost distributions, and live SaaS posture findings. None of it is a rewrite of someone else's content.
03
Corrections
When we get something wrong, we fix it in public and date the change. A standing corrections log is part of being citable.