How a Construction Firm Went From Days of Zoning Research to Minutes
A property development company in Los Angeles. Before every build, their team had to manually dig through city zoning records to figure out what they could and couldn't do with a property.
The Problem
Before the firm could break ground on any project, someone had to research the zoning rules for that specific property. That meant navigating ZIMAS (LA's Zone Information and Map Access System), cross-referencing parcel data, reading through municipal codes, and figuring out what building height limits, setback requirements, and use restrictions applied. For one property, this could take a full day. For a company evaluating multiple sites at once, it was a bottleneck that delayed every project by weeks. And getting it wrong meant compliance violations, costly redesigns, or worse — building something that had to come down.
Our Approach
Getting the Data Out of City Systems
LA County's zoning data lives in ZIMAS — a government system that's functional but not designed for speed or bulk lookups. We built an automated system that pulls property details, parcel information, and associated zoning regulations from ZIMAS and organizes it into a structured database that the AI can search instantly.
Teaching AI to Understand Zoning Law
Zoning regulations aren't simple. They reference other codes, have exceptions, and use language that even experienced developers sometimes misread. We trained the AI specifically on LA County zoning law so it can understand the nuance — not just keyword-match, but actually interpret what a regulation means for a specific property.
A Simple Interface Anyone Can Use
The team isn't made up of engineers — they're builders, architects, and project managers. The interface had to be as simple as sending a text message. Type a question about a property, get a clear answer with the citation. No training needed.
Results & Impact
The firm's planning phase went from weeks to days. Zoning research that used to require a full day per property now takes minutes. The team evaluates more sites, moves faster on the ones that work, and catches zoning issues before they become expensive problems. They've saved roughly $25,000 per project in research and outside consultant costs.
Technical Highlights
Technology Stack
AI Assistant (GPT)
Trained on LA County zoning law to answer property-specific questions in plain English with direct code citations
Automated Data Collection
Pulls property and zoning data from the city's ZIMAS system automatically — no manual lookups needed
Backend Infrastructure
Processes and structures zoning data so the AI can search and reference it instantly
Web Application
A clean chat interface where the team types questions and gets sourced answers — no training required
