Knowledge / AEC-native data
Construction contracts (NEC, JCT, FIDIC)
The conditions of contract, particulars, amendments and schedules that define obligations, risk allocation, notices and payment between the parties.
The contract is where the project's risk is allocated, and it is the document most people avoid reading until something has gone wrong. Standard forms such as NEC, JCT, FIDIC and the Irish PW-CF suite arrive with the documents that change what the printed conditions mean: Z clauses and Contract Data in NEC, schedules of amendments in JCT, Particular Conditions in FIDIC, the Schedule in the public works forms. The obligations they create run through notices, deadlines and payment cycles that live in other documents. Structured knowledge makes the contract something a commercial team can question directly, with every answer traceable to the clause and the amendment that governs it.
What teams need to ask of it
What does this clause actually require once the amendments and particulars are applied?
Which notices do we have to give, to whom, by when, and what happens if we miss one?
Where has risk been moved from the standard form position, and in whose favour?
Which obligations flow down to the subcontract unchanged, and which have been altered?
What does the contract say about this event: a compensation event, a relevant event, a variation, or nothing?
Which documents take priority when the employer's requirements and the contractor's proposals disagree?
What did we agree to in the schedule of amendments that the site team has never seen?
Why generic AI gets it wrong
Standard forms are widely available, so a general-purpose model has read them. That is the problem. Ask it about a clause and it will explain the unamended standard form, fluently and correctly, while your contract says something different in a Z clause or a schedule of amendments that the model has never seen. The answer is textbook accurate and contractually wrong.
The contract also does not work as a stand-alone text. Obligations depend on defined terms, on the particulars that fill in the blanks, on the order of precedence between documents, and on dates that only exist once the programme is agreed. A model reading the conditions alone cannot tell you whether a notice is late, because it does not know when the event occurred or what the contract data says the period is. It will produce a confident summary of a document that, as amended, does not say what the summary claims.
One question, answered
QUESTION
ANSWER
SOURCES
We onboard a focused slice of your project data, configure the ingestion paths, extract a structured context graph, and build a workflow against the resulting structured knowledge layer.
Related to
The contract sits above the employer's requirements and the contractor's proposals and sets the order between them. It defines how variations and compensation events are raised and valued, and it is the document the tender pack was priced against. Contract obligation extraction and variation impact review both start from it.
Which contract forms does it work with?
Does it replace legal review?
Can it flag notice deadlines from the contract?
Related pages
Employer's requirements
The client's statement of scope, performance and quality requirements under a design and build contract, against which the contractor's proposals are measured.
Change orders, variations and compensation events
Instructed or notified changes to scope, time or cost, and the notices, quotations and assessments that follow under the contract.
Tender packs and ITT documents
The invitation to tender and its enclosures: employer's requirements, drawings, specifications, pricing documents, conditions and returnable schedules.
Contract obligation extraction
Extracting obligations, notice periods, risk allocations and deliverables from the executed contract into a structured register the project team can act on.
Variation and change impact review
Tracing a proposed change through the drawings, specification, programme and contract to identify what it touches and what it triggers.