Match transactions to your ledger, flag the gaps.
Reconciliation is the accounting task everyone defers: two lists that should match, don't quite, and finding out why means eyeballing hundreds of rows. The commercial tools that automate it price for mid-market finance teams. This blueprint builds the workspace a small business actually needs, upload two CSVs, let the matching engine do the eyeballing, review only the exceptions.
A reconciliation workspace with two CSV uploads, bank statement and internal ledger, and a matching engine that works in passes: exact matches first, then fuzzy matches (same amount within a few days, similar descriptions) each carrying a confidence score. The review screen collapses what matched, asks you to confirm or reject the suggestions, and isolates what's unmatched on each side. The output is a reconciliation report: beginning and ending balances, matched totals, and every exception with a reason field, exportable to CSV.
The critical design rule is transparency: every suggested match shows why the two rows were paired. You're never asked to trust a black box with your books.
It ships with sample files seeded, 40 transactions where 32 match, 5 mismatch on amount or date, and 3 are missing from one side, so you can watch the engine work before trusting it with real statements.
Which banks does it support? Any bank that exports CSV, which is effectively all of them. The importer adapts to your format with one follow-up message.
Does my financial data leave my control? No. Files are processed in your sandbox; the app and its data live in your repository.
Can it handle multiple accounts? Ask for it, per-account workspaces are a natural follow-up.