AI Engineers Cannot Fix a Broken Project Workflow
Updated: Aug 31
Putting technical people directly into project teams is one of the more credible construction AI strategies. It shortens the distance between a promising idea and the messy reality of RFIs, submittals, schedules, cost reports and field decisions.
Construction Dive’s discussion of Suffolk’s Jobsite of the Future program emphasizes consistent technology deployment before useful project data can be captured.
For DDG, that sequence matters. AI is effective when records use stable structures, teams follow the same workflow, and the system can distinguish current information from superseded information.
What weak data looks like to an AI system
RFIs with inconsistent subjects and undocumented closure logic.
Submittals linked to the wrong specification section or package.
Schedule activities that do not connect to locations or responsible trades.
Change records that separate cost, scope and supporting evidence.
Models and documents with unclear status or revision authority.
AI Engineers Need a Project-Process Partner
Embedding an AI specialist is not enough if that person can only build prototypes around bad inputs. Project leadership must be willing to standardize forms, remove duplicate channels and assign ownership for data quality.
The technical role should be paired with a project-process role. One builds the capability; the other makes the workflow usable and repeatable.
Ask better questions before buying another tool
Owners and contractors should ask where the source data lives, who validates it, how permissions are applied and what evidence supports an automated answer. Those questions reveal more about readiness than a vendor demonstration.
AI should reduce the time required to find and compare evidence. Accountable people must still own the project decision.
Source and further reading
The DDG perspective
At DDG, we would start by stabilizing the RFI, submittal, change, schedule, and revision workflows. Only then would we automate retrieval and comparison. An AI engineer can improve the capability, but the project team must still own the records and decisions.
If your team is trying to connect models, field information and accountable delivery, talk to Detail Design Group.
Comments