When Trimble announced its acquisition of Document Crunch in April 2026, the strategic pitch was legible but abstract: contract intelligence would become the “intelligent DNA” of the Trimble Construction One suite, providing a contractual rule set that the entire platform would operate against. On June 9, that pitch became concrete.
Document Crunch launched Project Assist — an agentic AI layer built on its CrunchAI technology that does not just analyse contracts but acts on them. Available immediately in the United States and Canada, Project Assist can automatically generate redlines, submittals, notices, and requests for information directly from a project’s contract documents.
That is a meaningful shift. Document Crunch’s original product was a risk analyser: it read contracts, flagged clauses, and surfaced risk scores for project managers and legal teams. Project Assist takes the next step. The analysis becomes the input, and the output is work — actual project deliverables generated based on what the contract requires.
What Project Assist Does
The agentic layer works by treating the contract as a source of obligations and then monitoring whether those obligations are being met. When it identifies a gap — a submittal that is due, a notice that should have been sent, an RFI that needs to be issued in response to a specification conflict — it generates the document rather than waiting for a human to produce it from scratch.
In practice, this collapses a workflow that currently involves a project administrator reviewing contract requirements, identifying what needs to be done, writing the relevant communication, formatting it for the recipient, and routing it through a review process. The first two steps in that chain — identifying the obligation and drafting the response — are now automated.
The product is positioned specifically around risk management at the project level rather than the pre-award phase. Document Crunch’s earlier tooling was primarily used before contracts were signed, to understand what a project team was agreeing to. Project Assist operates after execution — during the live project, where the real financial exposure accumulates.
The Agentic Framing
Trimble describes Project Assist as “construction’s first project-level AI risk intelligence platform.” The “first” designation is a marketing claim that cannot be independently verified, but the framing around agentic AI is worth taking at face value as a signal of direction.
The construction software industry has talked about AI agents — tools that do not just inform but act — for several years. Most of what is currently deployed in the market operates closer to the analysis end of that spectrum: tools that surface information, flag issues, and recommend actions, leaving the execution of those actions to humans.
Project Assist is designed to cross that line for a specific, bounded category of work: the generation of standard project documents. This is a sensible place to start. The outputs are templated, the inputs are defined by the contract, and the error modes are recoverable — if the system generates a notice that should not have been sent, the damage is limited and correctable. Compare that to the higher-stakes agentic use cases (autonomous scheduling changes, automated payment approvals) that carry much greater consequence if wrong.
The bounded scope is not a weakness. It is the engineering decision that makes deployment practical. Getting the easy things right creates the trust required to expand the system’s mandate.
What This Means for Trimble
Trimble SVP Mark Schwartz, in the April acquisition announcement, described Document Crunch as providing the contractual rule set that would serve as intelligent DNA for the entire TC1 platform. Project Assist is the first demonstration of what that means in product terms.
Trimble Construction One integrates project management, estimating, field operations, and financial management into a single suite. The addition of a contract-aware agentic layer means that the obligations contained in a signed contract can now propagate directly into the workflows of the people responsible for executing against them — without requiring a project administrator to manually translate contract language into action items.
That integration is what the acquisition was really about. A standalone contract analysis tool is useful. A contract intelligence layer that is embedded in the platform where superintendents manage their daily logs, where project managers track their submittals, and where finance teams reconcile cost codes — that is a fundamentally different kind of product.
The launch also signals timing: the June 9 release date is roughly two months after the April acquisition announcement, which suggests the integration roadmap was ready before the deal closed. Document Crunch was already part of Trimble’s venture portfolio and had an existing product integration with ProjectSight. Project Assist is not being built from scratch under Trimble’s ownership — it was developed while the acquisition was in progress.
Reading Between the Lines
The timing of the launch relative to Autodesk’s own AI expansion is worth noting. Autodesk’s recent acquisitions — Rhumbix in March and the $3.6 billion MaintainX deal announced in late May — are pushing that platform toward field data and building operations respectively. Trimble’s move with Project Assist pushes into the contract and compliance layer, which is a domain where Autodesk does not have a strong native offering.
The platforms that will matter in five years are probably not the ones that have the most features — they are the ones that own the most meaningful process steps in a project’s lifecycle. Contracts are one of the highest-leverage process steps that exist: they define the financial and legal structure of everything that follows. If Trimble can make its platform the system that turns contract language into project action, it controls a node that the rest of the software stack has to route through.
Project Assist is available now for US and Canadian customers on Trimble Construction One.
