Building Permit Review: City of Scottsdale

The Arizona State University Artificial Intelligence Cloud Innovation Center (ASU CIC), powered by Amazon Web Services (AWS), collaborated with the City of Scottsdale to develop an AI-enabled document-intelligence solution capable of reviewing, extracting, and validating information from complex regulatory and engineering documents.

Building permit applications often require extensive documentation and verification. For example, applicants need to prepare multiple types of complex documents containing diagrams, charts, and text, while reviewers must verify hundreds of checklist items against laws and regulations. This collaboration focused on understanding how AI could streamline the review of pool and spa site plans documents that must conform to engineering, zoning, environmental, and safety standards with the goal of reducing manual review time and improving consistency across submissions. Through proving the AI-applicability to the pool approval use case, it paves the way for expanding the applicability of this technology to other permitting requirements and approvals. 

City of Scottsdale staff review hundreds of pool and spa construction plans every year. These plans vary widely in format, completeness, and technical detail. Some include drainage sheets, easements, or wall details; others contain gas line requirements, native plant inventories, ESL (Environmentally Sensitive Lands) zoning conditions, or multiple rounds of redline revisions.

Because designs come from many engineering firms, Scottsdale reviewers must manually verify:

• Zoning and parcel information
• Pool placement and utility conflicts
• Drainage and grading requirements
• Fence and barrier compliance
• Native plant protection (when applicable)
• Adherence to city, EPA, ADEQ, and Maricopa County standards

In some cases, only revised documents are submitted, while the original site plan iteration is missing. Revision naming conventions differ by contractor (e.g., “_1,” “-2,” “-4_3”), requiring reviewers to reconstruct the order of changes.

This manual effort is time-consuming, and Scottsdale sought a solution that could help classify document types, detect inconsistencies, identify missing components, and ultimately accelerate the review workflow while maintaining compliance and safety.

The CIC team conducted a detailed review of real permit packages provided by Scottsdale —including site plans, standard engineering sheets, drainage maps, wall calculations, authorization letters, and approved plan sets. These documents revealed significant variation across cases, including:

• Missing first-iteration plans
• Multiple approval stamps representing separate departmental checks
• Fence details appearing only in later revisions
• Inconsistent inclusion of natural gas line information
• Redline comments and revision histories embedded within PDFs

To design an AI-assisted review workflow, the team categorized each document type, mapped submission patterns, identified common sources of inconsistency, and outlined which elements could be automatically recognized through computer vision and language models.

AWS tools such as Amazon Textract, Amazon Bedrock, and Amazon Comprehend were leveraged to extract zoning labels, detect missing annotations, classify revision iterations, and highlight discrepancies (e.g., mismatched QS numbers or missing drainage sheets). This framework lays the foundation for a future system that could automate preliminary checks before a human reviewer performs a final evaluation.

This initiative demonstrates how government agencies can leverage generative AI and document-processing technologies to improve regulatory workflows. Water departments, planning agencies, and permitting offices nationwide face similar challenges: high document volume, inconsistent submissions, and labor-intensive manual checks.

By identifying the structure and variability of Scottsdale’s plan-review ecosystem, this project outlines a scalable pathway for:

• Reducing review time
• Standardizing document intake
• Improving reviewer accuracy
• Minimizing back-and-forth with applicants
• Increasing transparency and consistency across approvals

“This collaboration reflects Scottsdale’s commitment to innovation and service excellence. By working with ASU’s Cloud Innovation Center, we explored new ways to make permitting faster, clearer, and more consistent for our community. This initiative sets a precedent for how cities can embrace emerging technologies to improve customer experience and strengthen trust in public processes.”

City of Scottsdale

The analytical framework used here can be applied to a wide range of public-sector document sets, including:

• Building permits
• Engineering submittals
• Environmental compliance reports
• Zoning and land-use applications
• Utility inspections
• Water-quality compliance documentation