City ordinance and rules search for normal people
Municipal ordinances are written in dense legal language and organized in ways that make sense to lawyers, not residents. When someone wants to know "How tall can my fence be in my side yard?" they shouldn't need to navigate hundreds of pages of legal text. Yet this is exactly what most cities require.
Civic Scout transforms city ordinances into a natural language search system. Residents ask questions in plain English and receive direct answers with citations to the relevant ordinances. The system uses embeddings to find the most relevant ordinance sections, then provides context to an AI language model that generates a clear answer grounded in the actual legal text. The platform is live at owatonna.civicscout.org for the city of Owatonna, Minnesota.
Software / Civic Infrastructure
2024
Production (Owatonna, MN)
Municipal ordinances are stored as unstructured, hard-to-parse HTML. They're organized by legal code sections, not by the questions residents actually ask. Finding an answer often requires searching multiple documents and interpreting legal language without context.
When residents don't have easy access to ordinances, they rely on hearsay and speculation. Neighborhood discussions about what's allowed become debates about who remembers the rules correctly, rather than references to authoritative sources.
Converting unstructured ordinance HTML into structured, searchable data is technically challenging. The data needs to be parsed, cleaned, and organized to contain the right context—enough to answer questions accurately, but not so much that it becomes noise.
The effectiveness of natural language search depends on high-quality embeddings. Each ordinance section must be encoded with enough context to match relevant questions, while avoiding overlap that would produce too many irrelevant results.
Ask questions in plain English and receive answers grounded in actual ordinance text
Every answer includes references to the specific ordinance sections used
Uses vector embeddings to find relevant ordinances based on semantic meaning
Language models synthesize clear answers from ordinance context
Custom pipeline converts unstructured ordinance HTML into searchable, structured data
Platform core can be adapted to other cities and ordinance systems
Interested in bringing natural language ordinance search to your city or contributing to the platform? Let's collaborate.
Start a Conversation →