The Review Bottleneck
Building departments are losing experienced plan reviewers faster than they can replace them. AI-assisted review addresses the labor shortage, cuts cycle times, and eliminates the inconsistency that makes the permitting process unpredictable.
By Will Maclean
In most cities, housing production moves at the speed of plan review. A permit application sits in a queue until a staff reviewer checks the drawings against the IBC, state amendments, local ordinances, and ADA requirements. For a mid-size commercial project, that review takes 20 to 40 hours of reviewer time. Departments handling hundreds of applications per month don't have that time to spare, and the queue grows.
The cost is not abstract. The National Association of Home Builders estimates the carrying cost of permit delays at roughly $6,538 per month per lot. For a multifamily project waiting eight weeks for a first review, that figure compounds across every unit in the pipeline. Multiply it across every jurisdiction running behind, and you begin to see the shape of the problem.
The Workforce Cliff
The labor shortage in plan review is structural, not cyclical. The Bureau of Labor Statistics projects building inspector employment to decline 5 percent through 2032, driven primarily by retirements. The reviewers who know the code — who carry decades of institutional knowledge about local amendments, historical enforcement patterns, and the practical implications of fire separation requirements — are leaving faster than departments can hire and train replacements.
This is not a hiring problem that better salaries alone can solve. The pipeline of qualified reviewers has narrowed because the work requires a combination of construction knowledge, code fluency, and regulatory judgment that takes years to develop. Meanwhile, the code itself adds complexity with each new edition cycle, raising the bar for entry while the pool of candidates shrinks.
The Consistency Problem
What often goes undiscussed is the inconsistency this creates alongside the delays. Two reviewers working from the same code will flag different issues on the same set of plans. One catches a non-compliant egress width under IBC 1005.1; the other misses it. One demands documentation for a fire-resistive assembly; the other doesn't ask.
Architects and contractors figure this out. In high-volume jurisdictions, experienced applicants learn which reviewers are strict on which sections. They route submissions accordingly, or they resubmit until an overloaded department approves. This is not an edge case. It is standard operating procedure in markets where the review queue is measured in months rather than weeks.
The downstream effect is that code compliance becomes partly a function of reviewer assignment rather than plan quality. That undermines the purpose of the review process entirely.
The 2024 IBC Compounds the Problem
The 2024 IBC adoption cycle is landing on departments already running behind. Several meaningful changes will require reviewers to retrain:
- Section 1010.1 revises door hardware requirements for egress in Group I and Group R occupancies.
- Section 903.2 updates when sprinkler systems are required in Group A occupancies.
- Section 1209 changes occupant load calculations for toilet and bathing facilities.
Most jurisdictions are still operating under 2018 or 2021 editions. Retraining a department on 2024 while simultaneously working through an existing backlog is not a theoretical challenge. It is the reality facing most building departments over the next two years.
What AI-Assisted Review Actually Does
AI-assisted plan review addresses the labor problem without requiring departments to find reviewers who don't exist. The mechanical work — checking dimensions against code minimums, verifying occupant loads, flagging missing required elements, cross-referencing occupancy classifications against fire separation requirements — gets handled automatically.
This is not a replacement for professional judgment. It is a reallocation of reviewer time toward the work where human expertise actually matters: evaluating alternative means of compliance, assessing complex assemblies, making judgment calls on ambiguous conditions. Automated pre-screening handles the checklist so reviewers can focus on the analysis.
The consistency gains may matter more than the speed gains. An automated system applies the same standard to every submission. It does not miss the egress width on plan sheet A3 because it is the forty-seventh application reviewed that week. It does not forget that Section 903.2 changed in the 2024 edition. It eliminates the variability that makes the permitting process unpredictable for applicants and indefensible for departments.
Departments that have implemented this kind of automated pre-screening report 30 to 50 percent reductions in first-review cycle times. That translates directly to months off project timelines and thousands of dollars off carrying costs per unit.
The Procurement Barrier
The adoption barrier is procurement, not capability. Government technology purchasing moves slowly, and building departments rarely have dedicated IT budgets. The approval process for new software can take longer than the implementation itself.
The departments getting ahead of the backlog problem are framing AI-assisted review as operational infrastructure rather than a technology initiative. The distinction matters for budget routing. A staffing solution that reduces overtime and prevents the need for additional FTEs moves through approval differently than a software purchase request. The framing reflects the reality: this is a workforce tool, not a technology experiment.
The departments that act on this now will absorb the 2024 code transition without falling further behind. The ones that wait will continue to lose experienced staff, extend their review queues, and watch the inconsistency problem deepen. The backlog does not resolve itself.