Delay an expected item
Move a selected future cash movement later and compare the projected line with the base forecast.
13-week forecast
A projection from known cash inputs.
The forecast is calculated in application code rather than generated by a language model. It produces the same result from the same ledger and expected-item inputs.
Starting point
The current organization cash position, derived from the opening balance and confirmed transactions.
Weekly movement
A recent historical run rate provides the base movement. Dated expected items are added once in the week when they fall due.
Outputs
Thirteen weekly balances, the lowest projected point, a possible run-out week, and a confidence label based on available history.
Forecasts are planning estimates, not guarantees. Thin or unusual history can make the projection less representative of future activity.
What-if scenarios
Test an assumption without changing the books.
Add a weekly cost
Apply an extra recurring weekly outflow to understand its effect on the 13-week position.
Scenario results are disposable. Running one does not edit transactions, expected items, or the saved base forecast.
Signals and confidence
Rank attention, not certainty.
Signals combine logged facts and calculated estimates. The displayed confidence value describes the strength of evidence behind a signal; it is not a probability that an event will happen.
| Signal type | Evidence used | Interpretation |
|---|---|---|
| Overdue item | A dated expected item already recorded in the organization | A logged fact, shown at the highest evidence level |
| Runway floor | The deterministic 13-week forecast | Inherits the forecast’s history-quality label |
| Spend anomaly | Recent vendor spending compared with its trailing pattern | Stronger when more comparable history exists and the change is larger |
| Concentration | Ranked spending or inflow groups from confirmed transactions | Shows dependency visible in the current ledger |
Statement classification
Suggestions first, ledger second.
Parse the source file
CSV and Excel rows are mapped from their columns. Standard PDF statements may use Anthropic to extract structured transaction data.
Apply known organization mappings
Previously confirmed corrections can classify familiar descriptions before another AI classification is needed.
Propose categories
Remaining rows can receive cash-flow type, management category, counterparty, and confidence suggestions.
Require review
The import remains pending until a user confirms it. Low-confidence or unfamiliar rows can be corrected before saving.
Accounting view
Cash first, with an accrual view.
The confirmed cash ledger is the default source. Reports can also present an accrual view using expected receivables and payables, while keeping realized and expected amounts distinct.
CashCatalyst supports management cash-flow work and reporting. It does not replace professional accounting,
tax, audit, or legal advice, and its outputs should be reviewed before formal use.