Can AI Help a BIM Team Apply Company Standards Consistently?
Every company has standards for how it wants work done. Those standards do not always produce the same result on every project. Different people may read a rule differently, work from different revisions or make a local exception without documenting it for the rest of the team.
That is why Arcadis's expanded collaboration with Autodesk interests me. Arcadis says it is connecting its own knowledge, standards and guidance with Autodesk Assistant through Model Context Protocol capabilities. It also describes a separate research effort, See Through Walls, that uses AI and sensor data to infer hidden conditions in existing buildings. The announcement describes Arcadis's own initiatives; it does not announce a general-purpose standards checker for every firm.
A standard needs a project interpretation
In BIM work, the company standard is often only the starting point. An owner may have its own asset names and handover fields. A project specification may define equipment information differently. A BIM Execution Plan may assign a particular person or trade to make the final decision.
I discussed this problem in an earlier DDG article on trusted BIM data before coordination: a model can look coordinated while its names, classifications or source documents are out of date. Applying one office rule more consistently will help only if it is the correct rule for that project and stage.
An assistant could be useful when it shows the governing company standard alongside the applicable project requirement and points to the exact record behind a warning. For example, it might identify equipment with a missing asset property or drawings whose naming does not match the approved convention. The reviewer could then decide whether the item is an error, an approved exception or a conflict between requirements.
What the assistant should show
Before relying on an automated standards check, I would want to know:
Which version of the company standard did it use?
Which project specification, owner requirement or BIM Execution Plan rule overrides or supplements it?
What model element or document triggered the finding?
Who can accept an exception, and where is that decision recorded?
These questions matter as much as the speed of the check. Arcadis reports that one of its automation efforts reduced a quality-review process from five days to about half a day. That is a company-reported example, and it does not tell us how a different team's review would perform. The more transferable idea is to keep firm knowledge close to the work, with enough evidence for a professional to check the result.
When standards become easy to find and project exceptions are visible, newer team members have a clearer path to doing the work consistently. Experienced reviewers can spend more time resolving the real conflicts that no general rule can settle.
Comments