In March, a brand new mild rail line connecting Seattle to downtown Bellevue opened, including to growth stress within the fast-growing suburb.
That stress is acute within the metropolis’s allowing workplace, the place workers members subject questions from allow candidates, interpret allowing codes and consider pre- and post-submittal requests.
Like many cities, Bellevue is exploring find out how to use AI to enhance public providers, stated Sabra Schneider, town’s CIO, in an interview with InformationWeek. Streamlining allowing is a sensible place to start. The challenge’s bold objectives: to cut back by 30% the 20,000 workers hours devoted annually to allowing and to chop the variety of resubmitted permits by half.
Sabra Schneider, CIO, metropolis of Bellevue, Wash.
Discovering a accomplice
Bellevue had quite a few choices to think about for its early AI forays, with startups clamoring to share their merchandise with metropolis governments, Schnieder stated. Govstream.ai emerged as a frontrunner based mostly on three components: it was an area startup, had historical past within the civic tech house and proposed a partnership that “gave Bellevue confidence it could be a superb pilot,” she stated.
A number of months of negotiating with Govstream.ai earlier than work formally started in April 2025 made it clear that town’s contract course of wanted to be up to date to work with AI startups.
“We knew with AI we have been going to need to suppose slightly bit in another way as a result of a lot of the distributors working within the house did not have 5 years of presidency expertise or, usually, the issues that you just’re searching for in a conventional RFP,” Schneider stated.
Adopting an ‘innovation decision’ to accomplice with AI startups
As a substitute, town adopted the Progressive Design Partnership Coverage in July 2025. The decision goals to assist town accomplice with native and nationwide startups via extra of a design partnership mannequin than a conventional RFP.
Govstream.ai and Bellevue labored collectively to determine the objectives of the allowing pilot challenge, which is damaged into the next phases:
-
Section one: The primary part of the challenge focuses on an inside information that solutions questions for town’s allowing workers.
“It acts as a chatbot for workers. It additionally acts as an e mail response device to reply routine questions, in order that workers are spending much less time wanting up routine issues or code and extra time responding to advanced questions,” Schneider stated. The objective is to make this device public-facing as properly.
-
Section two: The second part of the challenge, at the moment underway, will present real-time allow software steerage. “Ideally, that may shorten timelines for brand spanking new housing and in addition industrial house, principally permitting prospects to submit a cleaner packet on the primary shot round as a substitute of going backwards and forwards with workers,” Schneider stated.
-
Section three: This part goals to leverage AI to automate allowing software triage. The objective is to identification points and push permits ahead sooner. With AI dealing with that work on less complicated permits, the allow group may deal with extra nuanced, advanced points.
Preliminary outcomes and ongoing work
Progress to this point is encouraging; 198 customers in growth providers now have entry to the AI device. “We’re fairly optimistic about reaching that 30%,” Schneider stated.
phases two and three of the challenge, town is focusing on a 50% discount in software resubmissions. Resubmissions can sluggish software approval significantly. Every resubmission can take weeks, and candidates might undergo a number of cycles. Govstream.ai and town have a tentative plan to launch its software assistant in June.
AI adoption for civic CIOs: Classes discovered
Civic CIOs will be swept up within the pleasure of AI innovation similar to their friends within the personal sector, however after they innovate the outcomes should profit the general public. Schneider has stored that in thoughts as she and her group proceed to maneuver ahead with AI adoption. Among the many helpful classes she has discovered:
-
Neighborhood alignment. Civic CIOs have to think about if AI use instances are aligned with the communities they serve, in line with Schneider. With that in thoughts, Bellevue has actively sought neighborhood suggestions from stakeholders, together with companies, nonprofits and college students via an innovation discussion board.
From that public engagement, digital fairness for residents emerged as a definite precedence. Stakeholders additionally raised considerations about jobs and the equity and accuracy of the data town provides with the assistance of AI. This perception helps Bellevue because it thinks via its AI insurance policies past allowing and continues to search for methods to use the know-how to its providers.
-
Information administration. Schneider famous that she and her group have discovered quite a bit about knowledge readiness as they work on this challenge. “Bellevue needed to do a good quantity of knowledge cleanup to ensure that Govstream to ingest our code and ingest our guidelines,” she stated.
-
Human partnership/ area experience. Bellevue’s IT group performs an vital function within the allowing challenge, however the Govstream.ai group labored straight with a number of the metropolis’s allow techs to grasp find out how to ship a challenge that solved precise ache factors and will enhance over time. “Clearly, the folks closest to doing the work know essentially the most about find out how to do the work higher,” Schneider stated.
-
Traceability and transparency. Bringing allowing workers into the method introduced traceability to the forefront of the dialog. They needed to understand how the AI device supplied the solutions it did. Early on, Govstream.ai in-built traceability capabilities and suggestions loops to allow coaching of the techniques over time.
-
Peer networking. AI adoption could be a collaborative course of. CIOs at completely different organizations and in numerous jurisdictions can study from each other. The GovAI Coalition might be a spot for presidency CIOs to begin. It has templates and instruments that policymakers can reference and adapt as they set up AI governance and launch AI initiatives. “It is by authorities and for presidency. And that is an ideal group to assist get [you] began,” Schneider stated.
