Search across all content
An AI platform that helps applicants prepare complete building permit submissions and helps plan reviewers identify which codes apply.
In Honolulu, building a new home, extending one, or carrying out major alterations requires a permit from the city, and securing one could take months, sometimes years. The root of the problem was not that reviews were slow. It was that too many applications arrived incomplete. “One of the biggest challenges that creates permitting delays,” said Dawn Takeuchi Apuna, Director of the Department of Planning and Permitting, “is submission of building permit applications that are incomplete and of poor quality, requiring corrections that result in additional review cycles that extend the overall building permit process.” In 2023, the pre-screening stage alone had a six-month waiting period.
Once an application cleared pre-screening, it moved to plan reviewers, who checked it against local building, zoning, and fire codes. This is specialist work, and each review took an average of 60 to 90 minutes. With experienced reviewers retiring and new staff needing years to build the same depth of knowledge, it became increasingly difficult to keep pace with demand.
It is a challenge shared by planning departments across the United States. Regulations change frequently, applications vary widely in quality, and much of the delay sits in the gap between what applicants submit and what reviewers need to begin their work.
Honolulu tackled the problem in stages, starting with what came through the door and then turning to how it was reviewed.
In 2022, the department introduced robotic process automation to handle pre-screening. When a submission arrives, an automated system checks it against a list of required information. If anything is missing, the application is returned before it reaches a human reviewer. That alone cut months from the earliest stage of the process.
With pre-screening working more efficiently, the bottleneck shifted to code compliance review. In 2023, the department began a pilot with CivCheck, a software company that builds AI-powered platforms tailored to each city's regulations. Following the pilot, CivCheck launched publicly on 8 December 2025.
The platform works on both sides of the permit process. For applicants, it guides them through preparing their submission before they send it to the department, checking plans and documents against Honolulu's requirements, from code references and electrical outlet placement to land use regulations. It can also process handwritten submissions. When an application is approved, the applicant receives a certificate of completion and an ID code to use when submitting it through the city's official system.
For plan reviewers, the platform identifies which regulations apply to a given project, highlights the relevant compliance points in the submitted plans, and indicates whether the plans meet the applicable codes. The final decision on every application remains with the human reviewer. CivCheck CEO Dheekshita Kumar has been clear about this, saying that “AI doesn't make decisions. It helps humans make better decisions faster by organising information, flagging relevant code sections and surfacing issues that still require professional judgement.”
The platform also helps address the loss of institutional knowledge as experienced reviewers retire. CivCheck works with a department's senior staff to build their expertise into the platform, so that newer reviewers can draw on it from their first day. The aim, in Kumar's words, is to “translate their knowledge onto the platform so that even the most entry-level plan reviewers can conduct the review to the same level of quality as your most experienced plan reviewers.”
CivCheck is currently available for residential permits, covering single-family homes, duplexes, accessory dwelling units, ohana units, and all residential new construction, additions, and alterations. It is free for applicants and optional at this stage, though the department plans to make it mandatory.
1. The pre-screening backlog dropped from six months to seven days
Before the department began introducing new tools, projects entering pre-screening faced waiting periods of up to six months. Robotic process automation, introduced in 2022, progressively reduced that by running applications against a checklist of required information before they reached a human reviewer. By the time CivCheck launched in December 2025, the remaining backlog was about five weeks. Within weeks, that was down to seven days, with approximately 174 projects in the queue. Davis Pitner, a public information specialist with the department, said the team was “confident that CivCheck will not only help [us] continue clearing it, but will also make the prescreen process faster and more efficient than before.”
2. Review times fell from over an hour to under twenty minutes per application
Before CivCheck, plan reviewers spent 60 to 90 minutes on each application, working through the relevant codes and checking compliance point by point. With the platform identifying which regulations apply and highlighting the relevant information in the plans, that dropped to 15 to 20 minutes. The reviewer still makes every compliance decision. The tool reduces the time it takes to reach that decision.
3. Applications are arriving more complete
Because CivCheck guides applicants through the requirements before they submit, applications are entering the system with fewer gaps. Fewer incomplete submissions means fewer rounds of corrections, which means less time in the cycle that caused much of the original delay. Director Apuna summed up the combined effect, sharing that “the result is higher quality applications, shorter permitting times, and better-informed applicants.”
1. It was important that the department addressed the intake problem before the review problem. There was no point in speeding up code compliance review if the applications arriving were not ready to be reviewed. Innovation director Abe Toma described applications that were often “not close to being something that we can even start reviewing.” Honolulu is clearer that the bottleneck first, to both help applicants prepare more complete submissions and help reviewers check code compliance faster.
2. Keeping human reviewers at the centre of every decision built trust in the tool. The AI identifies relevant regulations and surfaces potential compliance issues, but every decision on whether a project meets the code remains with the reviewer. Kumar described this as a deliberate choice that matters in public-sector work, with the tool “reducing the time it takes for the plan reviewer to make a decision on whether or not a project is compliant with a given regulation” rather than making that decision for them.
3. Embedding experienced reviewers' knowledge into the platform helped bridge the gap left by retirement. Many planning departments face the same challenge: senior staff retire, and the expertise they carry takes new hires years to develop. CivCheck worked with the department's experienced reviewers to “translate their knowledge onto the platform,” in Kumar's words, “so that even the most entry-level plan reviewers can conduct the review to the same level of quality as your most experienced plan reviewers.”
This case study was written with assistance from artificial intelligence.





Connect with 500,000+ public servants solving your hardest challenges.





Connect with 500,000+ public servants solving your hardest challenges.
Help public servants worldwide learn from your work, what worked, what flopped and what you'd do differently
Share your project
Log in or sign up to continue the conversation