This is how a contract reads before it is signed.
The project below is invented and stays that way: no customer’s work, no customer’s numbers. What is not invented is the method applied to it — which clause carries a time bar, how many days the window runs, and what triggers a notice were read out of the product’s own notice register.
The one example project this site names, here and elsewhere. Two views, one contract.
- VALEUR · EPC-2026-007 · REV C
- Petrochemical · turnkey EPC
- Three volumes · 742 pages
- FIDIC-based particular conditions
From raw contract to decision, in four steps.
The five figures on the panel are the output of the four steps below, and the steps stand in the order the contract is read rather than the order in which a product would introduce itself.
-
CONTRACT VALUE
$1.2 billion
ILLUSTRATIVEThe turnkey EPC value. It is the denominator under every ratio on this panel.
-
PAGES ANALYSED
742
THREE VOLUMESParticular conditions, general conditions and appendices. All parsed; none searched.
-
CLAUSES IDENTIFIED
183
WHOLE CONTRACTThe clauses that fall under notice, claim, time and payment.
-
HIGH RISK
31
17%Thirty-one of a hundred and eighty-three. The reader can divide it out too.
-
CALCULATED EXPOSURE
$24.7 million
2.1%Two point one per cent of the value. Not a loss: the total exposure the unclosed clauses carry.
LIMITValeur does not claim to move this figure. It says what it does: it puts the clause, its window and its evidence on the table before signature.
-
01
The contract lands: three volumes, 742 pages.
Parsing is not searching, and the difference starts here. A search finds where a word occurs; parsing knows where a clause begins and ends, which heading it sits under and which appendix it refers to.
The product holds a clause by its own anchor: number, heading and body become one record, and sub-clauses stay attached to the clause above them.
ScopeThe volume count and the page count on this page were written for the scenario. What parsing does was not.
-
02
The notice clauses separate: part of the 183 matches the register.
The product’s notice register carries twenty-three FIDIC clauses. The contract’s clauses are read against those twenty-three, and every match brings its own trigger and its own window with it.
Ten of the twenty-three carry a fixed window, and those windows are 14, 21 and 28 days. The narrowest is 14 days; the remaining thirteen carry no fixed period, and their timing comes out of the event’s own circumstances.
What is realNeither the register nor the windows are invented: they were read by parsing the product’s own source while this page was written. The figure a hundred and eighty-three is invented.
-
03
Risk is ranked: 31 clauses are marked high risk.
The ranking is not an opinion but the join of two fields: the clause’s severity, and whether it carries a time bar. Eleven of the twenty-three records in the register carry one.
Three of those eleven carry a second flag that makes the bar absolute: 8.4, 19.1 and 20.1. When the window closes the entitlement falls away, without the merits being reached.
What is realThe clause numbers, the severity field and the time-bar field come from the product’s source. The figure thirty-one was written for this scenario.
-
04
The financial effect is written: $24.7 million of exposure.
Exposure is the sum of the amounts the unclosed clauses carry — a total, not a forecast. It comes to two point one per cent of the contract value, and the reader can divide that out too.
Every amount stays attached to its own clause; whoever opens the total sees which clause carries how much, and when that clause’s window closes, on the same screen.
ScopeThis amount is not a result and is certainly not a saving. It is an example total for an example contract.
AN ILLUSTRATIVE SCENARIO. Above, the page and clause counts, the risk counts and the money figures belong to a contract that does not exist: the contract in this scenario and its figures are invented. But the window days do not — the twenty-three record notice register and the periods it carries were read by parsing the product’s own source while this page was written. The method is real, the contract is not, and none of this is a customer outcome.
What we invented, and what we read.
An illustrative scenario that cannot show its own line is not an illustration but an advertisement. The line is here, in three rows.
The sentence that carries the line does not sit at the foot of the page but inside the section it belongs to, and a gate keeps it there: take the label out of its section and the build does not run.
What this page gets asked.
Is this a real customer project?
No. Both the contract and its figures were invented for this page. No real customer name, reference or figure is published anywhere on this site.
Then which part is real?
The method. The twenty-three FIDIC clauses in the notice register, the eleven that carry a time bar, the three whose bar is absolute, and the fixed windows of ten clauses — 14, 21 and 28 days — were read by parsing the product’s own source.
Is the exposure figure a promise of savings?
It is not. Exposure is the sum of the amounts the unclosed clauses carry: visibility, not a result. Valeur does not claim to change that figure.
What keeps the label on the page?
A gate. The label sentence has to be printed inside the section it vouches for, has to name every class of figure above it, and has to carry the phrase saying the contract is invented. Drop any one of the three and the build does not run.
Would you like to see this with your own contract?
The contract above is invented. Yours is not. The same steps run over your own particular conditions.