What exactly does this tool check?
It extracts 4-digit years (candidate founding-year mentions) and capitalized multi-word phrases (candidate location or proper-noun mentions) from each snippet you paste, then flags any that appear in some sources but not others. It also strips common stopwords from each snippet and computes a keyword-overlap percentage between every pair, flagging pairs below roughly 40% overlap.
Is this checking whether my facts are actually correct?
No. This is a heuristic keyword aid, not a factual accuracy checker. It cannot tell you whether '2019' is the correct founding year, only that '2019' appears in one snippet and not another, so you know where to look and align the wording yourself.
Is any of my pasted text sent to a server?
No. All extraction and comparison happens in your browser using regular expressions and word-set comparisons. Nothing you paste is uploaded, logged, or sent anywhere.
Why would low keyword overlap between two sources matter?
If your homepage About section and your Crunchbase profile share very little wording, that can be a sign they were written independently and may drift on other details too, such as category or differentiators. Low overlap is not proof of a factual mismatch, but it is a reasonable signal to review those two sources side by side.