Commercial Retail Development — Five Cross-Discipline Conflicts Caught Pre-Construction

• 5 of 5 conflicts confirmed by the GC
• $55K-$120K rework caught pre-con
• $233K-$583K total project exposure

A-Z Builds approached Flikt.AI as a validation exercise rather than a purchase. The General Contractor wanted to know whether automated plan review could catch real coordination conflicts before they became expensive change orders, and whether the findings would hold up to scrutiny from people who build for a living.

That framing set the bar. Returning a long list was not the test; the findings had to be checkable. Flikt.AI works directly on the 2D PDF set with no BIM model required, reads every sheet against every other across all five disciplines on the job, and cites the sheets each finding came from. That made independent verification possible - the GC could take each flagged conflict back to the drawings and decide for himself whether it was real, rather than trusting a score. All five held up, and the GC also supplied severity calibration based on construction practice, which fed back into how findings are ranked.

Every conflict sat between two disciplines, and every sheet involved was correct on its own terms. There was nothing for a single-discipline reviewer to find.

Sign-off was organized by discipline, so a cross-discipline conflict had no owner. It became visible only when two sets were read against each other, which was nobody's assignment.

Underground scope is especially exposed: conduit, irrigation lines and existing utilities all occupy the same ground with different drawings and different reviewers.

Structural implications of late equipment changes were easy to miss - electrical plans showed modified equipment while the structural calculations index had no foundation design for the new loads.

Coastal wind loading added a code dimension: new site lighting structures were not accounted for in the wind load calculations for a high-wind zone.

Without a validation method, there was no way to know whether an automated review would produce real findings or noise - which is why the GC reviewed all five himself.

Design review was organized discipline by discipline. Each consultant checked and signed off on their own sheets, and the working assumption was that a correct set from every discipline would add up to a coordinated project.

Nobody in that process was assigned to read the civil survey against the architectural demolition plan, or the landscape planting plan against the electrical underground routing. That is precisely why these five conflicts survived. Every sheet involved was correct on its own terms - the electrical engineer routed the conduit correctly and the landscape architect placed the trees correctly. Neither drawing was wrong, and neither reviewer was looking at the other one. The conflict existed only when the sets were read against each other, which was nobody's assignment until the excavator was already on site.

Five conflicts detected on an 8-sheet set, and five confirmed accurate by the General Contractor - no false alarms, and no time spent chasing findings that turned out to be nothing. That verification rate is the number that mattered to A-Z Builds, because a review that returns noise costs more attention than it saves.

The rework identified in the detected conflicts alone ran $55K-$120K. Benchmarked against 14 ground-truth conflicts from independent GC field review, total exposure across the project ran $233K-$583K. Individual findings ranged from $4K-$12K for the wind load calculation gap on site lighting up to $25K-$50K for the demolition-versus-utilities overlap.

The efficiency is in when the work happens, not just whether it happens. A conduit rerouted around a root zone on paper is a redline. The same reroute after the trees are planted and the conduit is in the ground is excavation, replanting and schedule. Reading every discipline against every other before ground breaks is what moves that decision earlier.