2026
PermitPilot
Making ADU permitting easier to understand
Role
Product Designer
Team
1 human + 1 agent swarm
Platform
Web
Tool Stack
Simtheory + Simlink / Claude / Next.js / Vercel / Supabase / Mapbox / Open City+County data via API
PermitPilot is a web app prototype that helps Bend homeowners answer a seemingly simple question:
Can I build an ADU on my property, and if so, what do I do next?
Accessory dwelling units can be one piece of the housing supply puzzle, but the path from curiosity to action is often difficult for a homeowner to navigate. Zoning rules, parcel data, development standards, application requirements, and payment steps are spread across systems that were not designed around a community member’s first question.
The Challenge
Permitting is not just a technical process. It is a confidence problem.
For many property owners, the first barrier is not construction cost or design complexity. It is simply understanding whether their idea is allowed. If that answer is hard to find, people may never take the next step. They may not call. They may not apply. They may not spend money on a feasibility conversation. The idea dies somewhere between interest and confusion.
That opacity also creates pressure on City staff, who spend time answering repetitive questions from residents who are not trying to be difficult. They are trying to make sense of a system that requires context they do not yet have.
The opportunity was to design a tool that could make that first step easier.
What I Built
PermitPilot answers a homeowner’s first question in a focused MVP: enter an address, get a clear eligibility answer, and understand the next steps in the permitting process.
Property + Zoning Data
Pulls real parcel and zoning information from City and County open data sources.
ADU Eligibility
Answers whether an ADU appears to be allowed on the property and surfaces the relevant development standards.
Permitting Roadmap
Explains what to submit, where to go, and how payment works.
The tool also includes an AI chat feature for follow-up questions, since permitting rarely stops at the first answer. The product structure keeps authoritative data at the center while using AI to help users navigate the surrounding complexity.
How I Approached It
I built the MVP in a five-hour window using Simtheory, an AI workspace where I worked with Claude through an agent-assisted workflow. Simlink connected the agents to my local machine, allowing them to create and modify the project files directly. I did not write code by hand; my role was to shape the product through scoping, decision-making, review, and iterative direction.
The most important part of the process happened before the build. I configured a Product Manager agent and started with a loose description of the product I wanted to create. Rather than asking the agent to generate immediately, I asked it to first walk me through a Socratic dialogue to clarify the user need, product scope, data sources, and technical approach.
That conversation defined the MVP. We identified the address-based flow, confirmed the opportunity to use open data APIs, separated deterministic data from AI-assisted guidance, and narrowed the interface to the three-card structure.
From there, agents scaffolded the app, connected the pieces, and deployed the prototype to Vercel. My role was to keep the product grounded: what problem are we solving, what should the tool answer confidently, where should it defer, and what would make the experience useful for a resident trying to take the next step?
Product Decisions
A few decisions shaped the MVP:
First, the product starts with an address because that is how property owners think about the problem. They are not starting with zoning terminology. They are starting with their lot.
Second, the interface separates the answer from the process. Eligibility and permitting steps are related, but they are different user needs. Combining them would make the experience feel more complete, but less clear.
Third, the AI layer is supportive rather than authoritative. The app uses structured data and development standards to ground the core answer, while chat is used for explanation, follow-up, and wayfinding.
Why It Matters
PermitPilot is a small prototype with a larger civic implication: reducing friction at the earliest stage of housing production.
If a property owner can understand their options faster, they are more likely to move from curiosity to action. If more property owners can do that, the community gains potential housing units without expanding the city’s footprint.
No single tool solves a housing problem. But the systems around housing are full of small moments where people lose momentum. PermitPilot explores what happens when one of those moments becomes easier to navigate.