Pascal Matta built software for tech giants like Google and T-Mobile before he ever poured a concrete foundation. When the Seattle resident wanted to add a deck to his home, he found that municipal permit codes were far more complicated to handle than the software code he had spent years writing. After struggling through a complex permitting process for a project he never even started, he decided to fix the problem himself. Matta launched Kolmo Construction and created a free online tool to decode the confusing regulations in King, Pierce, and Snohomish counties.
A tool for homeowners and builders
The site kolmo.io/permits functions as an un-gated database designed to answer questions that typically send homeowners into a spiral of municipal PDFs. It covers 457 verified rules across 80 jurisdictions and includes a Washington state baseline for 82 additional locations. The tool layers in a clear distinction between rules that have been directly verified by a city and those that have not.
Users enter a specific address to generate a lot-level breakdown. The system provides a 3D visualization of the property’s buildable envelope, height limits, and existing structures. It evaluates local zoning and building codes to show what can be built, such as whether a lot qualifies for two detached ADUs under Seattle code. The interface also offers instant answers for tasks ranging from plumbing work to tree removal to heat pump installation.
Matta said the information genuinely isn’t available anywhere else. He noted that simple permit questions vary wildly across neighboring cities. For example, a question might have a different answer in Bellevue, Kirkland, and Seattle, each buried in a different city’s PDF. Homeowners often discover these differences after they have signed a contract, which is when it becomes expensive.
Read Also: Amazon unveils Kindle remote and tactile upgrades
Technology applied to brick and mortar
The tool uses Google’s Gemini model with Retrieval-Augmented Generation (RAG). This system feeds the AI exact city rules so it can answer specific permit questions with citations rather than AI guesses. Matta also exposed the entire database over Model Context Protocol (MCP), allowing third-party AI assistants to query the rules directly.
To maintain accuracy, an automated weekly process re-checks every source page for updates. The results are posted to a public data-quality log. Matta noted that this system caught an error in his own fee calculator when Seattle updated its 2026 permit fee schedule. The tool costs him around $2,000 in LLM API tokens just to process Seattle’s data, but he has no plans to monetize it or put it behind a paywall. He simply wants people to use it.
Back on the job site, Matta’s software background gives Kolmo a distinct edge. He built internal AI tools that let him take voice notes and photos during a walkthrough. He can then generate a detailed project quote in under a minute while standing at a client’s front door. That overlap between code and physical construction is what keeps him energized. When asked whether he prefers building software or physical decks, Matta pointed to the sweet spot where his two careers meet.