How it works
Why the answer is worth trusting
Your policy becomes data, rules decide, and the AI explains. In that order.
The document becomes structured data
A policy is a PDF written however each company likes. We read it into one common shape: coverages, limits, deductibles, exclusions, and the conditions that can void a cover if they are not met. The same shape for every insurer, in Greek and English.
Rules check that data
Anything we show you as a coverage gap was found by a rule that looked at specific fields of your policy — not by a model forming an opinion. Every finding keeps which rule produced it and what values that rule read, so it can be checked and argued with.
The AI puts it into words
Once the rules have decided, the AI explains the finding in plain language. It cannot add a gap the rules did not find, and it cannot judge how serious one is — those fields do not exist in the interface we give it.
“Unknown” does not mean “not covered”
If the document says nothing about a cover, we do not conclude it is missing: silence is not evidence of absence, and a finding built on silence is the expensive kind of wrong. Where a rule checks whether something was recorded at all, the finding says exactly that — “not recorded”, never “not covered”.
What this does not do
The rules cover what someone has written a rule for — not every possible gap in every branch. Where there is no rule, we show you no finding: we would rather leave the screen empty than fill it with something we cannot stand behind.
The analysis is not insurance advice and does not replace the policy document itself or your licensed intermediary.