Foresight converts a fixed ledger cutoff into a reconciled 13-week cash position. It shows the weekly path, the evidence behind expected receipts, the basis of scheduled outflows, and the receivables creating exposure.
A governed, deterministic, ledger-derived cash position, not a statistically validated forecast. Evidence, registered rules, accounting identities, and document controls determine whether the line can be shown at all.
The weekly path is the opening question. The movements behind it, and the evidence that placed each one in time, answer the rest, read from the same object, not written beside it.
The trough clears the illustrative floor by $76,900, less than the unscheduled items carried outside the position.
Four scheduled categories, one identity. Payroll alone is $1,538,000, more than the cash the window ends with.
Evidence classes stay separate so an assumption cannot pass as measured customer behavior. Under half of projected money in rests on invoice-linked history, with the balance on observed timing or a registered rule.
Cash Pulse reads where cash stands now. Foresight lays out what the current books imply, week by week. The Scenario Engine models how a contemplated decision could change that path, bounded and deterministic, on the same measured baseline: components the ledger cannot support are explicitly excluded under a registered rule and declared, never silently simulated.
The figures in the panels below are derived from one illustrative position object. Seven identities (weekly arithmetic, the running balance, the headline figures, the disclosed dependencies, the receivables destination, the rendered chase queue, and the cash-out basis) are re-derived before the document renders. A divergence in any of them stops it.
Foresight keeps destination and scheduling basis separate. The receivables identity must foot, and the timing evidence remains visible.
| Counterparty / lot | Open amount | Age | Schedule basis | Registered rule |
|---|---|---|---|---|
| $126,300 | 74d | Measured | R4 · deep backlog | |
| $104,800 | 91d | Observed | R4 · deep backlog | |
| $57,400 | 119d | Declared | R4 · deep backlog | |
| $64,200 | 148d | Declared | R4 · deep backlog | |
| $57,000 | 91d+ | Mixed | See chase object |
Input is validated before the ledger is reconciled. The weekly position is then constructed and checked as a document. A failed control returns a named refusal.
Reject malformed engine-consumed fields before ingest.
One point-in-time Sunday boundary governs the position.
Open AR, open AP, cash accounts, and settlements must tie.
Build 13 weekly movements and the running cash balance.
Headlines, identities, statements, and printed arithmetic are checked.
Return one deterministic document or a named stopped condition.
All registered controls passed. The document and structured diagnostics are available.
The calculation and control system is separated here from enterprise capabilities that remain outside the present release boundary.
Evaluation begins with accounting identities, evidence classifications, refusal behavior, and the release boundary.
The current method is deterministic rather than probabilistic. It uses registered accounting, settlement, ageing, scheduling, and presentation rules. It does not rely on Monte Carlo draws, probability-weighted invoice lines, or an undisclosed confidence score.
A registered fallback may be applied and disclosed where one exists. If the missing or contradictory information would make the position unsupported, the system returns a named refusal instead of a partial document.
Foresight does not modify the source ledger. Its processing boundary is read-only: it calculates and reports without posting, transferring, transacting, or altering the accounting record.
No public predictive-accuracy claim is made here. The completed method studies used synthetic books and are not treated as customer accuracy evidence. Money-out timing remains unbacktested.
A forward cash position should remain legible from its headline balance back to the ledger evidence that produced it.
