Reference Pillar

AI Plan Check: How AHJs and Permit Reviewers Are Using AI on Construction Drawings

A reference pillar on AI plan check: what it does on the AHJ side, how it differs from design-side AI plan review, which jurisdictions are running it today, what it catches reliably, and what plan reviewers still need to do themselves.

AI plan check and AI plan review get used interchangeably in marketing material, but the operating models are meaningfully different. AI plan review is the design-side tool — applied by the architect, engineer, or contractor before submitting drawings for permit. AI plan check is the AHJ-side tool — applied by the jurisdiction reviewing whatever submittal arrives, regardless of drawing convention or quality. The underlying technology overlaps; the workflow constraints, accountability, and risk profile do not.

This pillar covers the AHJ-side application: what it does, which jurisdictions are running it today, the legal and accountability framework around it, what it catches reliably, and how it integrates with existing permit-review workflows.

AI Plan Check vs. AI Plan Review: The Operating Model Difference

Same AI engines can serve both sides of the table, but the constraints differ. The table below summarizes the differences that matter for tool selection and workflow design:

DimensionAI Plan Check (AHJ side)AI Plan Review (design side)
Who runs itBuilding department, code consultant, third-party plan reviewerArchitect of Record, Engineer of Record, GC quality team
Drawing input qualityWhatever the applicant submits — any convention, any qualityTeam's own standardized drawing output
Accountability after AI flags an issuePlan reviewer decides whether to issue a correction noticeDesign team decides whether to revise before submitting
Code scopeAdopted code edition + local amendments (jurisdiction-specific)Anticipated code edition (design phase decision)
Workflow constraintEmbedded in permit-review system, public-records complianceEmbedded in the design QA process, no public-records requirement
Approval timeline impactCompresses discovery; full approval timeline still governed by AHJ processCatches issues before submittal — reduces correction cycles
Liability if AI misses an issueAHJ professional liability + sovereign immunity in some statesDesign professional liability under licensing-board rules

For the design-side application, see the AI plan review guide. For the broader umbrella topic, see AI for construction drawings.

Which AHJs Are Running AI Plan Check Today (2026)

AI plan check is still emerging as a category but pilots and production deployments are accelerating, particularly for high-volume residential permit types where the code rules are well-encoded and the throughput pressure on AHJs is highest. Notable deployments and pilots:

City of San Jose, CA
AI-assisted plan check pilot for ADUs and residential additions; uses computer vision and code-rule encoding to pre-screen submittals before a plan reviewer touches them.
City of Honolulu, HI
Integrated ICC Digital Codes AI-assisted compliance into permit review through an ICC partnership. AI screens for IBC and Honolulu-amended code compliance.
City of Austin, TX
SolarAPP+ deployment for instant residential solar permit approvals. AI checks the solar permit submittal against an encoded version of the local code; same-day approval for compliant submittals.
ICC + Digital Codes platform
ICC has invested in AI-assisted plan-check tooling that AHJs can adopt via the Digital Codes platform. Available to any subscribing AHJ; deployment varies by jurisdiction.
Singapore BCA CORENET-X
International leader — Singapore's Building and Construction Authority operates an AI-augmented BIM e-submission system for permit review. Frequently cited as the most mature AHJ-side AI deployment globally.
UK Building Control Bodies
Several UK BCBs have piloted AI plan check for Building Regulations compliance, particularly post-Grenfell on fire safety and Part B compliance.

For city-specific permit-review information, see Helonic's AI plan check by city hub with 130+ jurisdictions covered.

What AI Plan Check Catches Reliably

AI works reliably on code rules that are geometric or numeric — measurable on a drawing and comparable against a code-mandated minimum or maximum. The reliability table below reflects industry consensus on the state of the technology in 2026:

IBC Chapter 10 (Means of Egress)
Corridor width, travel distance, dead-end limits, common path of travel, exit capacity
IBC Chapter 11 + ADA Standards
Turning circles, accessible routes, clear floor space at fixtures, reach ranges, accessible entrance count
NFPA 13 (Sprinkler Systems)
Head spacing, obstruction rules, hydraulic remote area, hose stream allowance
NFPA 101 (Life Safety Code)
Occupant load calculations, exit access width, smoke compartment sizing
NEC (NFPA 70) Working Space
NEC 110.26 panel clearances, dedicated equipment space, working depth and width
ASHRAE 90.1 + IECC
Envelope U-value compliance, fenestration-to-wall ratio, lighting power density, equipment efficiency
Energy code compliance paths
Prescriptive vs. performance path verification, mandatory provisions checklist
Zoning + height/area
Building height and area calculations against zoning envelope, setback verification

Code rules requiring judgment — equivalency arguments, alternative methods of compliance under IBC 104.11, performance-based design submittals — remain the plan reviewer's domain. AI surfaces the data; the licensed reviewer makes the call.

What AI Plan Check Does Not Catch (and Why That Matters)

Three categories of issues remain firmly in the plan reviewer's domain. AHJs adopting AI plan check should not represent the tool as a complete replacement for plan review — doing so creates both quality and liability problems:

  • Code interpretation and equivalency. When a submittal proposes an alternative method of compliance under IBC 104.11, the plan reviewer has to evaluate the equivalency argument. AI cannot do this; it can only verify against the encoded prescriptive code.
  • Jurisdiction-specific amendments and local rules. Most US states and cities have substantial amendments to the model codes. An AI trained on IBC base text will miss every amendment unless the AHJ has explicitly loaded the adopted text. Plan reviewers know the local amendments by heart; AI does not, unless deliberately configured.
  • System-level life-safety reasoning. Code compliance is more than checking individual rules — it's reasoning about whether the building, as a system, achieves the intent of the life-safety code. AI checks rules in isolation; system-level evaluation remains human work.

How AI Plan Check Integrates Into the Permit Workflow

AI plan check delivers value at specific stages of the AHJ's permit workflow. Most successful deployments slot AI in at one or more of these four points rather than treating it as a single "turn it on and run" integration:

Submittal intake / completeness check
Before the submittal enters the queue, AI verifies completeness: required sheets present, key schedules included, applicable code edition referenced. Reduces back-and-forth on incomplete submittals.
First-pass pre-screening
Before the plan reviewer opens the submittal, AI runs a code-compliance sweep. Findings go into the reviewer's workbench with sheet references and code citations. Reviewer evaluates findings, accepts or rejects.
Specialist hand-off
When the plan reviewer hands off to a specialist (accessibility, fire, mechanical), the AI's pre-screen for that discipline gives the specialist a head start instead of starting cold.
Correction-notice drafting
When the reviewer issues a correction notice, AI can draft the citation language with the specific code reference and sheet number, saving reviewer time on the boilerplate.

Frequently Asked Questions

How is AI plan check different from AI plan review?

AI plan check is the AHJ side — applied by the jurisdiction reviewing a permit submittal. AI plan review is the design side — applied by the architect, engineer, or contractor before submitting. Same underlying AI engines can serve both, but the operating model differs: AI plan check handles whatever submittal arrives, while AI plan review runs against a team's own standardized drawing output.

Which AHJs are using AI plan check today?

San Jose (CA) for ADUs and residential additions, Honolulu (HI) integrated with ICC Digital Codes, Austin (TX) via SolarAPP+ for solar permits, multiple UK Building Control Bodies post-Grenfell, and Singapore's BCA CORENET-X (international leader). The category is still emerging in 2026 but adoption is accelerating, particularly for high-volume residential permit types.

Will AI replace plan reviewers at the AHJ?

No. AI plan check is a pre-screening tool, not an approval authority. The plan reviewer retains professional and legal responsibility for the permit decision under the AHJ's adopted code and state licensing-board rules. AI handles the consistent checks; the plan reviewer handles equivalency, alternative compliance methods, and code interpretation disputes.

What can AI plan check reliably check?

Geometric and numeric code rules: IBC Chapter 10 egress, IBC Chapter 11 + ADA accessibility, NFPA 13 sprinkler spacing, NEC 110.26 working space, ASHRAE 90.1 envelope and lighting power density, IECC energy compliance, and zoning height/area calculations. Less reliable on judgment rules and jurisdiction-specific amendments not encoded into the AI's reference codes.

How long does AI plan check take vs. manual plan check?

AI plan check runs in minutes per drawing set, vs. days to weeks for manual plan check (national median permit review time is 4-8 weeks for commercial projects per ICC data). AI does not shrink the entire AHJ timeline because procedural steps remain — but it compresses the discovery phase from days to minutes, freeing reviewer time for the judgment-required parts.

Is AI plan check legally defensible?

AI is not an AHJ-approved decision-maker. Permit approval remains the licensed plan reviewer's legal act under the AHJ's adopted code. AI plan check generates findings; the plan reviewer evaluates and acts on them. As long as the AHJ documents that AI is used as a pre-screening tool with human-in-the-loop approval, the workflow follows the same legal framework as manual plan check.

Related Resources

Run AI Plan Check on Your Jurisdiction's Submittals

Helonic runs AI plan check against IBC, NFPA, ADA, ASHRAE, NEC, IMC, and IPC — with jurisdictional amendments support, sheet-and-code-citation findings, and reviewer-feedback loops that keep noise under control. Used by both AHJ plan reviewers and design-side teams pre-screening before submittal.

Try AI Plan Check

Related guides

All guides