A team Dev site for the Pre Permit AI App
The winning team of the City of Seattle PACT-Athon (Permitting Accountability and Customer Trust Hackathon) wanted to continue their work in creating this app that can help aim to create the solve for the issue of backlog of permitting that City of Seattle has explained exist
Hackathon #2: The PACT-athon (Permitting) The City of Seattle launched the PACT-athon (October 9) to solve a critical housing crisis where construction permitting times had doubled over the last decade due to endless review cycles. Our team attacked the "Garbage In, Garbage Out" root cause by building Pre-Assess, an AI engine that validates architectural plans against the Municipal Code before they ever reach a city reviewer. By pivoting to Google Gemini we successfully triangulated live GIS data with complex zoning laws to generate instant project specific compliance checklists. Securing 3rd Place we proved that an AI "Junior Responder" can filter out bad applications on day one validating the roadmap for our current commercial pilot.
Hackathon #3: The Customer Service Sprint (November 6) Hosted by the City’s Innovation & Performance team, this event challenged 18 teams to fix the "front door" of government by modernizing how residents submit and track service requests. The mandate was to use Open Data to "end the runaround" and build tools that make city services transparent and accessible. Our response, Rainy Day Friend, secured 2nd Place by replacing static forms with a conversational AI "Civic Translator." This solution directly supported the new PACT (Permitting Accountability and Customer Trust) initiative, proving that AI can simplify complex code for residents while generating real-time "trust metrics" for city leaders.
Hosted by the City of Seattle, AI House, and Ada Developers Academy, the PACT-Athon was a call to action for Permitting Accountability and Customer Trust. It wasn't just a competition; it was a rapid deployment environment where developers and policy experts convened to solve immediate civic challenges in 2 hour sprint.
We attacked the "Garbage In, Garbage Out" problem. Instead of just speeding up city reviewers, we built "Pre-Assess"—a tool to validate applications before submission. The goal was "Getting it Right from the Start," ensuring builders submit code-compliant plans on day one
Pivoting from standard tools to Anthropic Claude for better reasoning, we built a system that triangulates data: it takes an address, pulls live GIS/Parcel data, reads the specific Municipal Code (SMC), and generates a strict, project-specific compliance checklist
Securing 3rd Place, "Team Five Star" was recognized for delivering immediate, practical value. The win validated our "Junior Responder" concept—proving that an AI "gatekeeper" can successfully filter out bad applications before they clog the city's workflow, solving the root cause of the delay.
ChatGpt - Custom GPT Model with Actions
The Spin-Off: Automating the Architect's Workflow
During the PACT-athon, teammate and architect Kevin O'Leary identified a critical "out of scope" bottleneck. While the event focused on the City's permitting issues, architects struggle before submission, wasting hours manually scraping zoning data from scattered government maps. A Subject Matter Expert (SME) confirmed this manual "Goal 1" workflow was a major pain point that the main event didn't address.
The Solution: Seattle Parcel Extraction Agent Andrew Powers took the "2-hour team sprint" philosophy and applied it to a "2-day solo sprint." Using self-directed rapid prototyping, he built a custom tool to solve the specific pain point immediately after the event.
Input: A simple address or Parcel PIN.
Process: The agent bypasses manual map searches and connects directly to the City’s FeatureServer/2 API, triangulating 65 distinct zoning fields (setbacks, height limits, lot size).
Output: A structured JSON file that automatically populates the architect's AirTable, replacing hours of manual data entry with instant precision.
The Result: Proof of Proficiency This proactive build wasn't about a contract; it was a demonstration of technical competence. It proved that the hackathon's "Pre-Permit" engine could be adapted to private sector needs overnight. By delivering a functioning solution to an SME's problem in under 48 hours, we validated the ability to listen, build, and solve "Goal 1" challenges without waiting for permission.
Put in Address
2. Verfies address and Parcel ID
3. Pulls Parcel data from API
4. Allow for JSON Download