When four project records disagree about the same pour
Construction disputes are built from sequences. A missed milestone depends on what happened before it, what changed afterward, and which project participants documented the event.
The problem appears when the record is separated by document type:
Schedules show planned sequencing, but not always what happened in the field.
RFIs and submittals show questions and responses, but not always downstream impact.
Daily logs capture field conditions, but may not connect directly to contract obligations.
Change orders describe scope and pricing, but responsibility and timing may remain disputed.
Consider a structural-steel delivery dispute on a large industrial project. Before preparing a defense strategy, counsel needs to know whether the steel issue actually controlled the critical path, whether another event had already created delay, and whether project participants treated the issue as a change, a design problem, or a scheduling problem.
That answer does not live in a single file. It lives in the relationship between files.
A PDF pile is not a project record
A major construction matter may contain thousands of documents generated by different participants for different purposes and at different points in time. The scheduler records one version of progress. The superintendent's daily log records another. An RFI describes a field condition. A later change order assigns scope and price to some part of it. Project correspondence adds context. The pay application describes work as complete. Inspection records or photographs may tell a more complicated story.
Reading each category separately creates summaries. It does not necessarily create understanding. Construction claims are relational. The question is rarely just what a document says. The questions are what happened next, what happened before it, which party knew, which activity was affected, whether the condition changed scope or sequence, whether notice followed, and whether later records confirm or contradict the original account.
At Turbo Law, our Fact Graph goes beyond entity extraction. Identifying the owner, general contractor, subcontractors, architect, project dates, or change-order numbers is necessary, but that is only the beginning. The objective is an intelligent graph that captures an evolving understanding of the matter: planned activity and actual performance, a field condition and the RFI raised about it, the response and the directive that followed, claimed completion and contemporaneous inspection evidence, a payment request and the work supporting or challenging it, allegations in the pleadings and the project records that support, qualify, or contradict them.
The shape of that graph is different for construction than it is for employment, medical malpractice, or commercial litigation. It is also different from project to project. A hospital expansion with phased turnover may produce a graph dominated by sequencing, access constraints, design revisions, and inspections. A condominium defect case may organize around assemblies, locations, specifications, installation history, testing, and repair. A subcontractor payment dispute may center on scope, approved changes, notices, pay applications, and retainage. The graph should reflect the litigation problem, not force every matter into the same schema. And it should change as productions arrive, depositions are taken, expert materials appear, and theories narrow.





