Drop borrower PDFs here
Or click to choose files. Several at once is fine.
Use Gemini to read scans and answer questions — optional
Google AI Studio key — kept in this tab only, never saved to the file
Model
Ask something about everything extracted so far
Scanned pages take a minute. Tax returns, W-2s and paystubs usually have no text
layer, so they get read character by character. The first scanned page downloads a
~15 MB recognition model, then it's cached. Digital PDFs — credit reports, bank
statements, anything printed from a portal — come through instantly.
Copy is at both ends of the box, and Ctrl+Shift+C
copies everything from anywhere on the page.
Check the numbers that matter. Reading a photocopied 1040 is imperfect. Anything
that feeds qualifying income should be eyeballed against the page before you use it.