A Return Review assistant that reads the return against its source documents
The most valuable review is the one that catches the thing everyone else read past, so the Return Review assistant does a careful second read of a prepared return against its own source documents. It flags a number on the return that disagrees with the document it should match, a state filing that looks missed given where the income came from, and a meaningful movement from last year's return. Each flag carries a citation to the exact document or line it is based on, so a reviewer can check it in one click. It never changes the return. It surfaces, cites, and hands off to a human who makes the call.
Reads the return against the documents behind it
Source-document and return figures are compared, and mismatches surface.
- →The assistant compares figures on the return against the source documents they should match, and flags where they disagree, so a transposed or missed number does not slip through.
- →A state that looks missed given the income is flagged, so a filing obligation is caught before the return goes out rather than after a notice arrives.
- →A notable change from the prior-year return is surfaced, so a swing worth a second look gets one.
Everything is cited
Each flag points to the document or line it came from.
- →Every flag carries a citation to the exact source document or return line behind it, so a reviewer verifies the finding in one step instead of hunting for it.
- →A cited flag is checkable, so review time goes to judgment rather than to reconstructing where a number came from.
Human-reviewed, never auto-corrected
The assistant surfaces issues; a person decides.
- →Nothing is auto-corrected: the assistant flags what looks off and a human decides what to do with each item.
- →It is a second set of eyes on the numbers, added on top of your existing review, not a replacement for it.