Yes, as an early-warning reader with your contracts team deciding. AI can read site instructions, meeting minutes, RFI answers and client letters, flag events that may be a change or a delay, track the notice deadline in your contract, and draft the notice and supporting record. Your contracts manager decides whether to notify, and nothing is sent without their signature.
#Key takeaways
- Most lost variations start as an instruction nobody recognised as a change, not as a weak argument.
- AI can watch the documents where changes hide: site instructions, minutes, RFI answers, emails and drawing revisions.
- Every flag should link to the source document and the contract clause that sets the notice requirement.
- Deciding to notify, and the commercial valuation, stay with your contracts manager and quantity surveyor.
- A clean notice register and a weekly review habit are worth setting up before any AI.
#Why do contractors miss variation and claim notices?
Changes rarely arrive labelled as changes. An engineer's reply to an RFI adds a detail, a revised drawing moves a service route, a client representative asks for something in a coordination meeting, a late approval holds up a trade.
Each one may be a variation or a delay event. The site team is busy building, and the contracts team only hears about it later. The NIST AI risk management framework, january 2023 makes the same point for any AI output: a named person stays accountable for the decision, so the AI flags and a person on the contracts team decides.
Timing matters under most standard forms. Under the 2017 FIDIC red book, a notice of claim must be given within a fixed period after the party became aware, or should have become aware, of the event, as explained by ICC, march 2020.
#What can AI watch for in your project documents?
- instructions and RFI answers that add, remove or change scope compared with the contract drawings and specification.
- drawing revisions whose changes affect work already built, procured or programmed.
- meeting minutes and emails recording a request, an approval delay or access restrictions.
- daily reports mentioning standby, disruption or work out of sequence.
- deadlines: the notice period in your particular conditions, counted from the date of the source document.
For each flag, AI drafts a short record: what happened, which documents show it, which clause may apply and when notice would be due. The contracts manager confirms or dismisses it.
#What should a person always decide?
Whether an event is a change at all, whether to notify, the wording of anything sent to the client or engineer, and the valuation. Those decisions carry commercial and relationship judgement that a model does not have. AI's job is to make sure the question reaches the right person while there is still time to act.
This is the same approach as a contract review with a full audit trail: the AI reads and flags, a named person decides, and every step is logged.
#When is a register and a checklist enough?
On a single project with an experienced contracts manager who sits in every meeting, a notice register in Excel, a weekly review of instructions and minutes, and a standard notice template may cover most of the risk. A general AI assistant can help summarise a long letter.
It stops being enough across several projects, when document volume grows, correspondence is bilingual, and the contracts team cannot read everything. Then a workflow that reads every incoming document and keeps the register current earns its place through the notices it catches.
#Where does Stennir fit?
Stennir's AI consulting engagement starts with the AI Opportunity Audit: it maps where changes enter your projects and ranks five workflows, each with a cost and a payback case, including whether notice tracking is the right first build. The report is yours whether or not you build with Stennir.
If you go ahead, AI implementation builds the reading and flagging workflow on your document control system and email, with the contracts manager approving every notice. You pay for finished work. AI training teaches site and commercial staff to record events so they can be found.
We are
Stennir builds a workflow that reads your project documents, flags possible changes and drafts notices for your contracts manager to approve.
We aren't
Stennir is not a claims consultant or a law firm, and AI does not decide entitlement or send notices.
#What should a contracts manager do next?
Take one finished project, list the variations and claims you raised late or not at all, and trace which document first showed each one. That list is your test set. See AI for operations, take the free readiness assessment or book a 30-minute call.
#Common questions
What do you need from us?
The contract and particular conditions for one project, your notice register, a sample of instructions, minutes and correspondence, and the contracts manager who would approve flags. Stennir maps where changes enter your projects and agrees which decisions always stay with your commercial team.
Does it read Arabic correspondence?
Yes. Arabic and English are fully supported, including bilingual letters and minutes. Where a contract says the Arabic text prevails, the flag shows the Arabic wording beside any translation, so your contracts manager reads the original before deciding. Other languages are available on request.
Does AI give legal advice on entitlement?
No. It flags events, links source documents and drafts paperwork from your templates. Entitlement, notice strategy and anything that needs legal judgement stay with your contracts manager and your lawyers. The workflow makes sure they see the event while there is still time to act.
What could go wrong?
An event missed because it sat in a handwritten note, a false flag that wastes a reviewer's time, or a wrong deadline because particular conditions were misread. That is why deadlines are confirmed by a person, every flag links to its source, and nothing is sent without the contracts manager's signature.
