Seattle’s housing costs continue to rise alongside the cost of building new homes. While many factors contribute to these trends, the City can directly influence one critical driver: the construction permitting process.
Construction permit applications go through review by City staff to make sure they follow zoning, safety, and other requirements. Applications often take several cycles of review before they are approved; each review cycle adds a month or more to the process. Long review timelines add uncertainty and delay to housing projects, increasing costs and sometimes ending projects before they start.
The City of Seattle is pursuing many strategies to help make permitting faster and more predictable. Exploring the use of new AI tools is one of those strategies. Starting in March 2025 and in partnership with the Seattle Department of Construction and Inspections (SDCI), Innovation & Performance (IP) conducted a proof of value study with CivCheck to learn how their AI prescreening tool could impact the construction permit review process.
How Construction Permits Work Today
When someone applies for a construction permit, they submit a complex application containing multiple forms, documents, and technical drawings. That application goes through multiple stages of review, including:
- Intake screening: SDCI staff perform Zoning and Ordinance/Structural reviews of the application, with a focus on completeness: does this application have all the necessary documents for this project type? Are those documents complete?
- Plan review: multiple teams at SDCI and other departments do in-depth reviews of the application to ensure the design is compliant with the Land Use Code, Seattle Residential Code, and other building codes.
Both review stages usually take several cycles: first the applicant submits their materials, then the City teams review and either approve it or send it back with corrections. This cycle repeats until all review teams approve. Once an application is approved, the applicant pays fees and the City issues a construction permit.
How CivCheck Fits In
IP tested CivCheck as a pre-screening tool used by applicants. In this case, an applicant planning a small residential project—like a townhouse, single-family home, or accessory dwelling unit—runs their application through CivCheck before they submit their application to the City. CivCheck uses AI to analyze the files and run them through automated checks based on City requirements. The tool gives the applicant feedback on whether their application is complete and if their design is compliant. The applicant can then fix the errors in their application before submitting it to the City. Once the applicant submits their application, it still goes through the normal City review process.
“I’m hoping that this tool is something that will be available to [applicants] to…flag problems that you could fix today. I would rather wait a day or two to fix those problems [before submitting] and then potentially get through our first real review [from the City]… Time is everything.”
– Developer
Our hypothesis was that if many errors were caught by CivCheck and fixed up front, the application would need fewer review cycles and would be approved more quickly.
What We Did
We tested our hypothesis by engaging in a proof of value test with CivCheck. We tested CivCheck’s ability to automatically check both the completeness of an application and the compliance of a design, then modeled the impact on the permitting process.
CivCheck ran real project applications through their platform, SDCI staff reviewed the same applications, and then we compared results. We also interviewed 20 former applicants and more than a dozen staff members to hear what they thought about AI review of permits. This work began in March 2025 and concluded in October 2025; analysis of the results continued into early 2026.
What We Learned
The testing showed that CivCheck’s checks were accurate: application completeness checks were 87% accurate and design compliance checks were 92% accurate. Accuracy is likely to improve with more usage: it improved over the course of the test as CivCheck further calibrated the tool.
The checks were accurate, but when it comes to the impact on applicants’ experience, we found that check coverage—the ability for CivCheck to automatically detect all the different corrections a staff reviewer would find—matters as much as accuracy. Applicants would save the most time if pre-screening eliminates an entire review cycle. Removing a review cycle requires the pre-screening to flag every single correction up front. If the pre-screening only catches some of the corrections needed, then the applicant may end up going through the same number of review cycles as they would without pre-screening.
Based on this testing, we believe we can achieve higher coverage of completeness checks than compliance checks. Completeness checks look at a smaller set of requirements related to City processes and document standards. These checks are simpler to program into a pre-screening tool.
In contrast, checking compliance requires comparing the design to the thousands of requirements in Seattle’s building codes. Some of those requirements are conceptually complex and likely difficult to automate. While the test proved that many of the most commonly-failed compliance checks can be automated, the sheer quantity of checks required makes it harder to eliminate entire review cycles. Simplifying the City’s own codes and processes would make it easier to automate both completeness and compliance checks in the future.
What’s Next?
This pilot showed that tools such as CivCheck have real potential to save time for both permit applicants and City staff. It also demonstrated that the City has more work to do to make our requirements and processes clearer and less subjective.
Based on our findings, we recommend a production pilot for pre-screening application completeness. While we explore options for funding and logistical support to leverage new tools like CivCheck in our construction permitting process, we will continue working to clarify and streamline our permitting and decision-making processes.
Read the Full Report
The full report is available here: CivCheck AI Pre-Screening Pilot Report (PDF)