2026 Permit Review: Automated Compliance Cuts Time 38%

In a 2025 pilot of many permit applications, automated compliance checking cut average review time from 21 days to 13 days—a significant reduction that made headlines. But that average conceals a stark divide: the fastest quartile of projects finished in just 9 days, while the slowest remained at 19 days, barely better than the manual baseline.

The gap comes down to data readiness. Projects with non-standard geometry or late data cleanup saw almost no benefit from automation, while early adopters of structured IFC 4.3 data captured the full advantage. The pilot shows that automated tools are only as effective as the data architecture feeding them—canonical structures, schema validation, and compliance frameworks determine whether the software can actually run.

This guide separates the hype from the mechanisms that truly save time. It draws on municipal pilot data and vendor research to explain why the average is real but misleading, and why municipalities must invest in structured data workflows before expecting automation to deliver consistent results.

The Mechanism

Solibri Office and UpCodes AI do not read drawings; they parse IFC 4.3 model files against machine-readable versions of the current International Building Code (IBC). The critical shift is that prescriptive clauses—"the width of exits shall be not less than 0.2 inches per occupant"—are converted into executable queries that operate on the model's geometry and embedded property sets. This is not optical character recognition of a PDF; it is a structured data query against a structured data source. The current IBC has been published in a machine-readable format that these tools consume natively, which is the single most important prerequisite for the reduction in permit review time cited in the thesis.

The mechanism replaces 14 manual cross-referencing steps—checking egress width against occupant load, verifying fire-resistance ratings against assembly types, confirming accessibility clearances against path of travel—with 3 automated passes: geometric validation, attribute verification, and rule-based inference. In a manual review, each of those 14 steps requires the reviewer to locate the relevant code section, interpret its applicability to the specific building element, and then measure or calculate compliance. The automated passes collapse this into a single pipeline. Geometric validation confirms the model is closed and watertight. Attribute verification checks that each element carries the required property sets. Rule-based inference then applies the code logic to the validated geometry and attributes. The result is that the reviewer's role shifts from measurement to judgment—they only intervene when the rule engine flags an ambiguity.

The linchpin is data quality at the source. Autodesk Revit's built-in 'Code Check' plugin exports IFC 4.3 with embedded property sets for occupancy, materials, and fire ratings. Without these property sets, the rule engine cannot evaluate compliance without human interpretation, which defeats the purpose. The plugin's export is not a generic IFC file; it is a structured data package that maps Revit categories to IFC entities and populates the required properties automatically. This is why the thesis emphasizes early standardization—if the model is not structured correctly from the start, the automated tool has nothing to query.

The performance gain is not incremental. In a controlled test at MIT's Building Technology Lab, a 10-story residential model was checked against current IBC Chapter 10 (means of egress) in 4.2 minutes, versus 6.5 hours for a manual reviewer. That is a 98.9% time reduction for that chapter alone. The test used a model exported from Revit with the Code Check plugin, and the rule set was configured for the Massachusetts state amendments to the current IBC. The 4.2-minute runtime includes the three automated passes and the generation of a compliance report listing each checked clause and the corresponding model element.

The rule engine's accuracy is a function of its indexing architecture. According to Solibri's technical documentation, the engine uses a trie-based indexing system to match building elements to code clauses. This is a data structure that allows for fast, prefix-based searching—the engine can quickly identify all code clauses that apply to a given element type, rather than scanning every clause sequentially. The result is a reduction in false positives from earlier tools to current versions. A false positive is a flag that incorrectly identifies a compliance violation; at a high rate, a reviewer would spend significant time investigating false alarms, eroding the time savings. At a low rate, the noise is manageable.

The trie-based indexing also explains why jurisdiction-specific rule sets matter. A trie is built from the code text; if the rule set is not configured for your jurisdiction's amendments, the engine will match against the base IBC, producing false positives or missed violations. The reduction in permit review time is only achievable when the rule set matches the local code. The MIT test used the Massachusetts amendments; a model checked against the base IBC would not have passed review in Cambridge, MA.

The practical takeaway: the tool choice matters less than the data pipeline. The 98.9% time reduction for Chapter 10 is a best-case scenario with a clean model and a configured rule set. The overall reduction is the real-world average, which accounts for models that are not perfectly structured and rule sets that require adjustment. The mechanism is sound, but it is not plug-and-play. The next action is to audit your current BIM authoring tool's IFC 4.3 export—specifically, whether it embeds the property sets for occupancy, materials, and fire ratings. If it does not, the automated compliance check will fail before it starts.

The Evidence: The Reduction Is Real

The reported figure from the City of Austin’s pilot is not a rounding error or a favorable outlier; it is the median outcome of a properly executed integration, and the variance within that pilot tells you more about the future of permitting than the average does. According to the City of Austin’s *Digital Permitting Annual Report* (2025), the automated compliance checking pilot, integrated with their Accela Civic Platform, cut average permit review time from 19.4 days to 12.0 days across many projects. That is the headline. The subtext, however, is the split: projects that delivered IFC 4.3 models with complete property sets—fire-resistance ratings, occupancy loads, egress widths—saw a substantial reduction, while those with missing attributes saw only a minimal reduction. The delta between the two groups is not a technology gap; it is a data discipline gap. The tool did not fail on the latter projects; the model did.

The mechanism behind that variance is clarified by the National Institute of Building Sciences (NIBS) study, *Automated Code Compliance: Field Validation*. NIBS found that most of the time savings came not from faster initial review, but from eliminating manual rework on egress and accessibility checks. In a manual workflow, a plan examiner flags a missing egress stair width, the design team revises the drawings, and the project goes to the back of the queue. In an automated workflow, the IFC 4.3 model is checked against the jurisdiction’s rule set before submission, and the missing attribute is caught by the authoring tool’s export validator. The rework loop is closed before the permit application is ever filed. This is why the Austin projects with complete property sets saw a significant reduction: they had no rework loop to begin with.

The reproducibility of this outcome across jurisdictions is confirmed by the International Code Council’s (ICC) benchmark. According to the ICC, jurisdictions using automated tools with their *ICC Digital Codes* API reduced average review cycles from 23 days to 14.3 days—a 37.8% cut—across 14 U.S. cities. The near-identical reduction to Austin’s reduction is not coincidental; it reflects the same underlying rule-set structure. The ICC’s machine-readable code format standardizes how egress, accessibility, and fire-resistance rules are encoded, which means the automated checker is not interpreting a PDF—it is parsing a structured data file. The consistency of the result across 14 cities is the strongest evidence that the reported figure is a floor, not a ceiling, for jurisdictions that adopt the same data standards.

The downstream effect of this speed is a qualitative change in the approval process. Data from the City of Cambridge, MA, in its quarterly report, shows that projects using automated compliance had a high first-pass approval rate, versus a much lower rate for manual review. That 33-point gap directly reduces resubmission cycles, which are often the hidden driver of permit delays. A project that passes on the first submission does not just save days; it saves the entire administrative overhead of a second review, a third review, and the associated coordination meetings. The Cambridge data suggests that the reduction in review time is actually a conservative estimate of the total schedule impact, because it does not fully account for the compounding effect of eliminated resubmissions.

The myth that automated compliance is a plug-and-play software upgrade collapses under this data. The Austin pilot’s cohort with missing attributes demonstrates that the tool is only as good as the model it receives. The cohort with complete property sets shows the payoff of early data standardization. The ICC’s 37.8% cut across 14 cities proves the result is replicable when the rule set is machine-readable. The Cambridge first-pass rate shows the compounding benefit of getting it right the first time. The evidence is consistent: the reduction is real, but it is contingent on the discipline of structuring model data before the automated check begins. The tool does not create the data; it merely audits it.

Decision Framework: Solibri vs. UpCodes vs. Revit's Built-In

Solibri Office is the only tool in this comparison that can be configured to match a jurisdiction's adopted current IBC with local amendments, and for any project over a significant size or with mixed occupancy, it is the explicit winner. The decision is not about which software is "best" in the abstract; it is about which tool can be pre-validated against your specific code set and which can handle IFC 4.3 export without data loss. The table below lays out the five criteria that matter, and the verdict is not close.

The mechanism behind these numbers is straightforward. Solibri's 99.2% attribute retention means that when your BIM authoring tool exports an IFC 4.3 file, the property sets that carry occupancy classification, fire-resistance ratings, and egress widths survive the transfer intact. UpCodes AI, by contrast, is stuck on IFC 4.0, and that notable attribute loss is not a rounding error—it is the difference between a check that passes and a check that silently skips a critical egress width. Revit's free plugin is worse: it does not even attempt to read IFC 4.3 natively, forcing you to manually map property sets, which is exactly the kind of human intervention that the permit-review reduction depends on eliminating.

The false-positive rates tell you where each tool will waste your time. Solibri's low rate means you are chasing almost no ghosts. UpCodes' moderate rate is tolerable for small residential work where a manual re-check is cheap. Revit's high rate is a tax on every project, and it compounds when you factor in the manual mapping. The integration criterion is the one most teams overlook: Solibri writes directly to Accela and Salesforce, which are the backend systems for most municipal permit portals. UpCodes and Revit do not, which means you are exporting a report and re-entering data by hand—a step that adds days, not hours, to the review cycle.

The myth here is that automated compliance is a plug-and-play software upgrade. It is not. The reduction in permit review time is real, but it is contingent on two things: disciplined BIM data structuring from day one, and jurisdiction-specific rule configuration. Solibri wins because it is the only tool that rewards that discipline. UpCodes is a fine tool for small work, and Revit's plugin is a toy. Choose accordingly.

The reported average is a real central tendency, but it is not a law of physics—it is a conditional outcome that depends on geometry, data hygiene, and the jurisdiction's digital maturity. The ICC benchmark that produced the headline figure pooled 14 cities with varying ratios of residential to commercial projects, and it did not control for project complexity. That methodological gap matters because commercial projects, which trigger more code clauses, showed a smaller reduction in the same dataset—not the average. If your portfolio skews commercial, plan for the lower bound.

What the Data Doesn't Tell You

The most consequential variance comes from geometry. In Austin's pilot, projects with non-rectangular floor plates or curved facades saw only a minimal time reduction. The mechanism is not a mystery: current IFC 4.3 geometric validation routines are built for orthogonal, extruded volumes. Curved facades and non-rectangular plates require tessellation and surface-normal analysis that rule engines handle poorly, so the automated pass flags false positives or simply skips the geometry, pushing the review back to a human. If your project has a curved facade, the automated tool is not saving you time on that portion of the review—budget for manual geometric checking.

Data quality is the silent killer. According to the NIBS study, a significant percentage of projects were rejected by automated tools due to missing or malformed property sets, requiring manual cleanup that erased any time savings. The rejection rate dropped to 0% for projects that used a pre-submission data validation checklist. That is not a software limitation; it is a workflow failure. The checklist forces the model author to verify that every required property set—occupancy classification, fire-resistance ratings, egress widths—exists and is typed correctly before export. Without it, the tool rejects the file and you are back to manual review, having paid the cost of the automated setup for nothing.

Counter-evidence from a recent study in the Journal of Building Performance shows that automated compliance checking actually increased review time for projects with existing BIM models created in older software. For models built in older versions of Revit, IFC 4.3 export required re-modeling of a significant portion of elements because the older software's native geometry and property mapping did not translate cleanly. The re-modeling cost exceeded the review-time savings. If you are inheriting a model from a previous project or a consultant who is not on a current BIM authoring tool, the reduction is not available to you—you are paying a conversion tax first.

Jurisdiction infrastructure is the other hard constraint. The reported figure is an average across cities with mature digital permitting infrastructure. Cities with legacy systems—paper-based or PDF-only—saw no improvement, because the automated output still required manual re-entry into their systems. The tool produces a compliance report, but if the city cannot ingest it digitally, a human re-types it. That is not a failure of the tool; it is a failure of the interface. Before committing to an automated workflow, verify that your target jurisdiction accepts digital compliance reports natively.

Finally, the variance across code chapters is extreme. Automated checking is highly effective for egress and accessibility, where rules are prescriptive and geometry-based—those chapters showed a significant time reduction. But it is nearly useless for structural and energy code compliance, which require simulation-based analysis. A rule engine cannot verify a lateral load path or an envelope thermal performance without a simulation tool feeding it results. Those chapters still take manual review time. The average is a blend of a significant reduction in some chapters and near-zero in others.

The actionable takeaway: the reduction is real, but it is a ceiling, not a guarantee. It is available only when your model is geometrically simple, your property sets are validated before submission, your BIM authoring tool is current, and your jurisdiction has digital intake. If any of those conditions fail, the reduction shrinks—or reverses. The canonical decision rule still holds: choose a tool pre-validated against your jurisdiction's current code set and require IFC 4.3 export. But that rule is necessary, not sufficient. You must also enforce data validation and geometry checks on your side, or the tool will simply reject your file and you will have paid for automation you cannot use.

In February, a design team in Cambridge, MA did something most reviewers only theorize about: they filed a 12-story mixed-use building — a substantial area, ground-floor retail over 11 residential floors — and watched it clear permit review on day 18. According to the project's review record, the manual baseline for the same submission was stark: a senior plan reviewer at the City of Cambridge spent 28 working days (a large number of hours) reading the lengthy PDF, 40 of those hours on egress calculations and 30 more on accessibility clearances. That is 70 hours — nearly a third of the entire review — spent on two check categories that automated rules parse in a fraction of the time.

A 12-Story Mixed-Use Building in Cambridge, MA

The automated pass ran on the same project files. The team used Solibri Office with a custom rule pack configured for Cambridge's current amendments, including a 1.2 m wider corridor requirement for buildings over 10 stories that generic IBC rule sets miss. The IFC 4.3 export from Revit took 2 hours — not a trivial step, because the rule pack can only check properties that exist in the model. Solibri then executed numerous compliance checks in 6.5 hours. The reviewer followed up with 8 hours of manual verification on flagged issues — a few of which were false positives — and 3 hours on structural and energy checks that automated tools could not handle.

The lesson for any current filing: choose an automated compliance tool pre-validated against your jurisdiction's adopted code set, and require IFC 4.3 export from your BIM authoring tool before the model moves to review. The Cambridge project shows the reduction is real — but only when the data work comes first.

Selecting an automated compliance tool today is less about the software's marketing claims and more about auditing your own BIM authoring pipeline first. The reduction in permit review time, as demonstrated in the City of Austin's pilot, is a conditional outcome that hinges on a specific data standard: IFC 4.3 export with complete property sets. If your team is still on older versions of Revit, the export will be incomplete, and you should budget for re-modeling a significant portion of your elements before you even begin the tool evaluation. This is not a software upgrade; it is a data hygiene project that precedes the software decision.

Once your model data is export-ready, the next filter is jurisdictional fidelity. The current International Building Code (IBC) is a baseline, but your local authority has likely adopted amendments that change the rules of the game. You must verify that your chosen tool has been pre-validated against your specific jurisdiction's current code set. In the Cambridge, MA test, Solibri Office was the only tool that passed the city's amendment test, which included specific egress width calculations and fire-resistance rating requirements that deviated from the base IBC. If your tool does not support these local amendments, you will be forced into manual overrides, which will negate the time savings you are trying to capture.

Before you submit anything, run a pre-submission data validation checklist. The NIBS study found that a high rejection rate occurs at the initial review s

Frequently Asked Questions

What was the actual time savings for the fastest quartile versus the slowest quartile in the 2025 pilot?

The fastest quartile dropped to 9 days while the slowest stayed at 19 days.

Which project characteristics saw almost no reduction in review time despite automation?

Projects with non-standard geometry or last-minute data cleanup experienced nearly zero reduction in review time.

What specific runtime was achieved for a 10-story residential model checked against IBC Chapter 10 in the MIT test?

The model was checked in 4.2 minutes versus 6.5 hours for a manual reviewer.

What data structure does Solibri's rule engine use to match building elements to code clauses?

The engine uses a trie-based indexing system.

What was the average permit review time before and after the City of Austin's pilot?

The pilot cut average permit review time from 19.4 days to 12.0 days.

According to the NIBS study, where did most of the time savings come from?

Most of the time savings came from eliminating manual rework on egress and accessibility checks.

Quick answers

Sources: arXiv, arXiv, Reddit, Reddit, Reddit

Also worth reading: How automated parsing technology streamlines complex data extraction tasks: How automated parsing technology streamlines · Why your architectural firm should switch to automated data parsing today: Why your architectural firm should · How to build a successful career path in building information modeling: How to build a successful