v1 supports PNC personal-checking CSVs natively and a generic 4-column format (date, description, amount). Parsing is best-effort; if rows don't show up correctly, fall back to a generic CSV with those columns. Amounts will be normalized so positive = money leaving checking.
What happens next: the system parses the CSV, runs your saved rules to auto-categorize what it can, flags repeat-merchant transactions as recurring, then drops you on the review screen where you fix anything wrong and create rules for future imports.