Paper Plans Are Dead. Digital Review Is Half-Built.

Building departments across the country have spent the last decade digitizing intake. Applicants upload PDFs instead of rolling out D-size sheets across a counter. The paper is gone. But the review process behind it has barely changed.

A reviewer opens a PDF on a monitor, scrolls through sheets, and manually checks dimensions, occupancy loads, egress widths, and fire separation distances against the applicable code. They type comments into a spreadsheet or a tracking system. They send correction letters. The applicant revises and resubmits. The cycle repeats.

This is not digital plan review. This is paper plan review performed on a screen.

The Digitization Gap

The distinction matters because it explains why permit timelines have not improved despite billions spent on e-permitting platforms. A 2024 survey by the International Code Council found that 78 percent of jurisdictions now accept electronic plan submissions. Yet average review times have not decreased. In many jurisdictions, they have increased.

The reason is straightforward: digitizing the submission format does not digitize the review. Accepting a PDF instead of a paper roll saves counter space. It does not save reviewer time. The cognitive work — reading drawings, locating dimensions, cross-referencing code sections, checking calculations — remains entirely manual.

E-permitting platforms solved the logistics problem. Plans no longer get lost in transit. Applicants can check status online. Fee payments are electronic. These are real improvements. But they address the wrapper, not the contents. The actual bottleneck — the hours a qualified reviewer spends reading drawings against code — is untouched.

What Digital Review Actually Requires

True digital plan review means the software reads the drawings. Not just displays them — interprets them. It identifies walls, doors, stairs, occupancy boundaries, and fire-rated assemblies as discrete objects with properties. It measures distances, checks clearances, and flags non-compliant conditions against the applicable code edition, including state and local amendments.

This is a fundamentally different capability than what e-permitting platforms provide. E-permitting is document management. Digital review is document understanding.

The technical requirements are significant:

• Drawing interpretation. Architectural plans are not standardized data. They are graphical representations with varying conventions, scales, and annotation styles. A door on one firm's drawings looks different from a door on another's. Extracting meaningful information requires models trained on the visual language of construction documents.
• Code mapping. The IBC alone spans roughly 700 pages. Add state amendments, local ordinances, and referenced standards like ICC A117.1 for accessibility, and the compliance surface expands to thousands of individual requirements. A digital review system must map drawn conditions to applicable code sections automatically.
• Amendment tracking. No two jurisdictions enforce exactly the same code. California amends the IBC through Title 24. Texas has its own accessibility standards. New York City maintains an entirely separate building code. A system that checks against the base IBC without accounting for jurisdictional amendments is worse than useless — it creates false confidence.

The Middle Ground Most Departments Occupy

Most building departments today sit in an uncomfortable middle position. They have digital intake but manual review. They have invested in the front end of the process while the back end — the part that actually determines whether a building is safe — runs on the same human-intensive model it always has.

This creates a specific set of problems:

Reviewer time is misallocated. Experienced plan reviewers spend a significant portion of their time on tasks that do not require their expertise: measuring corridor widths, counting plumbing fixtures, verifying that door hardware complies with Section 1010.1.1. These are important checks, but they are mechanical. They consume hours that reviewers could spend on judgment-intensive questions — means of egress design for unusual occupancies, fire protection strategies for mixed-use buildings, structural adequacy in seismic zones.

Inconsistency compounds. When two reviewers check the same plans, they will flag different issues. One catches a non-compliant guard height under Section 1015.3. The other misses it but flags an unrelated annotation. Without a systematic baseline check, the quality of review depends on which reviewer is assigned and how much time they have. This inconsistency is invisible at the department level but obvious to applicants who experience it submission after submission.

The queue never shrinks. Departments that digitize intake without digitizing review often see queues grow, not shrink. Digital submission lowers the friction of applying, which increases volume. But review capacity remains fixed. The result is a wider funnel pouring into the same narrow pipe.

From Display to Understanding

The transition from displaying plans to understanding them is where AI enters the picture — not as a replacement for reviewers, but as a first pass that handles the mechanical checks.

Consider what a reviewer does when checking egress compliance for a commercial project. They locate every exit, measure travel distances from the most remote point in each space, verify corridor widths, check door swing directions, confirm panic hardware on doors serving occupant loads above 50, and cross-reference all of it against the occupancy type and applicable code edition. For a single floor of a mid-size office building, this takes one to three hours.

An AI system trained on architectural drawings and building code requirements can perform these same dimensional and prescriptive checks in minutes. It does not get tired. It does not skip sections. It checks every door, every corridor, every stairway, every time.

The reviewer then receives a set of flagged conditions rather than a blank set of plans. Their job shifts from finding problems to evaluating them — confirming the flags, dismissing false positives, and focusing their expertise on the design questions that require human judgment.

The Permit Timeline Equation

Permit timelines are a function of three variables: submission volume, review capacity, and review efficiency. E-permitting addressed none of these directly. In some cases, by making submission easier, it increased volume without a corresponding increase in capacity.

Digital review — actual digital review, where software reads and checks drawings — acts on review efficiency. It does not add headcount. It multiplies the output of existing staff by handling the routine checks that consume the majority of review time.

For departments facing the structural workforce challenges documented elsewhere — declining inspector employment, an aging reviewer population, a shrinking pipeline of qualified candidates — this is not a nice-to-have efficiency tool. It is the only realistic path to maintaining review quality as institutional knowledge walks out the door.

The paper plans are already gone. The question now is whether departments will finish the digital transition they started, or continue reviewing PDFs the same way they reviewed paper — one page, one code section, one correction letter at a time.