Why we built Telamont: the document problem behind construction disputes

Telamont started from an observation that isn't ours alone. Trace a construction dispute back far enough and you usually arrive at the same place: two documents that stopped agreeing, and no point in the process where anyone was going to catch it.

We come at this from two directions. Charles Goulet is third generation in a construction family, with a grandfather and a father who ran work on Quebec megaprojects, including the James Bay dams. He grew up watching projects end in disputes often enough that it read as normal. William Gagnon came from the AI side, applying it to the way manufacturers run their processes on the plant floor. The team around them carries technical experience from firms like WSP and AtkinsRéalis. Both backgrounds teach the same lesson: the failure is rarely in the document you are reading.

No single document holds the meaning of a project. The meaning sits in the relationships between them, and those relationships shift every week. Contract and general conditions. Specifications by division. Drawings across revisions. Addenda from the tender period. Change orders, RFIs and their answers, site instructions, meeting minutes, progress reports. Subcontracts that inherit obligations from the prime contract, sometimes faithfully.

Each is correct on its own terms. The failures live in between:

• A revision that invalidates an assumption three other documents still rest on
• A change order that moves a scope boundary a subcontract still describes the old way
• A spec requirement a bid never priced

We are not the first to notice

The industry keeps its own record of this. Arcadis publishes an annual tally of the disputes its contract solutions team handled, and the list of causes barely moves between editions:

• Failure to properly administer the contract
• Poorly drafted or unsubstantiated claims
• One party not understanding or not complying with its obligations
• Errors and omissions in contract documents, which came first six times over a nine-year stretch

Those aren't four findings. They're one finding described four ways.

The cost attached is large enough to feel like someone else's problem. The 2025 edition puts the average US dispute at $60.1 million, with North American disputes running about 12.5 months. Those figures come from the disputes a global consultancy gets called into, so they skew toward megaprojects. The mechanism doesn't. A $40 million job has the same document problem at a different scale, with less margin to absorb it.

Why nobody fixed it

Back in 2014, the same firm's head of contract solutions summed up what should be done: better contract administration, more robust documentation, a proactive approach to risk. He was right. It's still the advice. Dispute values have roughly doubled since.

The industry didn't ignore it. It couldn't act on it. Following that advice properly means holding a project's entire documentation in view, continuously, for the full duration of the work, and re-checking it every time something new arrives. On an eighteen-month job that's thousands of pages compared against each other, over and over. No team has ever had the time to do it.

So the industry built its operating model around that gap. Miss some of it, deal with the consequences later, through claims and negotiation and the twelve months of resolution in that data.

That isn't a failure of the model. It is the model.

What changed

Telamont reads a project's documentation in full. Contract and general conditions, specifications, drawings across revisions, addenda, change orders, RFIs, subcontracts. Not one document at a time: all of it together, in one frame of reference, and again every time a new version lands. The cross-checking no team has ever had the time to do runs continuously, from tender to close-out.

Reading was only half of the 2014 advice. The other half is judgment: which document governs when the specification and the drawing disagree, which notice clock starts at discovery rather than at the event, what an exclusion buried in a subcontract actually moves. No general-purpose tool knows any of that. We build it in, clause by clause, with the engineers, architects, construction lawyers and general contractors who do this work, read what the system produces and tell us where it is wrong. Judgment is the product.

A single read at signature, by one person, was always a partial answer to a problem that unfolds across the whole project. Technical, contractual and compliance risks surface while they are still decisions rather than damages, each tied to the clause behind it, in a system you can question and push back on rather than a report that gets filed.

That's what we built. A project should end on the quality of the work, not on a contradiction nobody caught. Finishing on time, with the margin you planned and no claim to defend, is not luck. That is what we intend to make normal.