← Examples

Interactive prototype · fictional data

Cash Review

Follow the cash.
Keep the questions visible.

Bank-to-ledger reconciliation, with source-linked workpapers and a clear list of items to review. Explore the example, then discuss a scoped reconciliation workflow or an Acquisition Financial Review.

The working example

Juniper Workshop Supply

Fictional company · January 2026 · USD · OPERATING-01

The balance check

A match is not a reconciliation.

The two records start $65 apart. Their closing balances also differ. Matching more transactions does not make either difference disappear.

The statement and ledger each roll forward arithmetically. The opening difference and unpaired items still require supporting evidence; no adjustment is assumed.

Inspect the constructed sources

Bank minus ledger / USD
ControlBankLedgerDifference
Opening balance12,500.0012,435.0065.00
Net movement21,331.0023,937.50−2,606.50
Closing balance33,831.0036,372.50−2,541.50

$65.00 opening difference + (−$2,606.50) movement difference = −$2,541.50 closing difference.

Source register

Inspect what the workpaper uses.

These are deliberately small, constructed sources. Positive amounts increase cash; negative amounts reduce it. The ledger contains only the selected cash account, not both sides of every journal entry.

STMT-01 / constructed statement

Bank activity

12 transactions, opening and closing balances. Includes an account fee and an outgoing wire with no corresponding ledger row.

GL-01 / constructed cash ledger

Cash-account detail

12 cash postings and balances. Includes a check and deposit without a statement counterpart. Their resolution remains open.

What this prototype demonstrates

The working rules show exact signed-amount matching, a configurable date window, reciprocal uniqueness, ambiguous duplicates and source-linked exceptions. Amounts are calculated in integer cents.

It does not resolve split payments, netted settlements, currency conversions or missing records. It uses fixed rules; it is not a demonstration of autonomous AI, extraction accuracy or a completed quality-of-earnings review. Numerical agreement does not establish transaction validity.

Fictional evidence

Record evidence