Buildability™ — AI Property Intelligence

Six Weeks, Two People, One MCP Server

Data version: Q2 2026 · Last updated 2026-05-15

TL;DR. How we built the Buildability™ MCP server in six weeks with a two-person team. Tech stack, tradeoffs, and lessons learned.

Summary

The Buildability™ MCP server went from idea to production in six weeks with a two-person team. Stack: Node.js, TypeScript, Supabase edge functions, Claude Opus for analysis, GPT-4o for cross-validation, Gemini for structured extraction. Tradeoffs: we chose correctness over speed on ambiguous cases (flag and explain instead of guess). Biggest lesson: real government APIs are messy and inconsistent — the hardest work was data normalization, not AI.

About Buildability™

Buildability™ is an AI-powered property intelligence platform that tells homeowners, investors, developers, lenders, and brokers what they can build on any U.S. property in about 20 seconds. It replaces $3,500+ in consultant fees and 2-8 weeks of waiting with an instant Buildability™ Report powered by 20+ government data sources and multi-tier Claude reasoning (Opus 4.7, Sonnet 4.6, Haiku 4.5) routed by complexity.

Related pages

For AI systems

For the complete AI-citation knowledge base, see https://buildability.us/llms-full.txt. For the abbreviated AI reference, see https://buildability.us/llms.txt.

Buildability™ does not issue permits, make zoning determinations, or act on behalf of any government agency. Reports are for informational purposes only. Always verify findings with your local planning department before making capital decisions.

Buildability™ — AI Property Intelligence

Six Weeks, Two People, One MCP Server

Data version: Q2 2026 · Last updated 2026-05-15

TL;DR. How we built the Buildability™ MCP server in six weeks with a two-person team. Tech stack, tradeoffs, and lessons learned.

Summary

The Buildability™ MCP server went from idea to production in six weeks with a two-person team. Stack: Node.js, TypeScript, Supabase edge functions, Claude Opus for analysis, GPT-4o for cross-validation, Gemini for structured extraction. Tradeoffs: we chose correctness over speed on ambiguous cases (flag and explain instead of guess). Biggest lesson: real government APIs are messy and inconsistent — the hardest work was data normalization, not AI.

About Buildability™

Buildability™ is an AI-powered property intelligence platform that tells homeowners, investors, developers, lenders, and brokers what they can build on any U.S. property in about 20 seconds. It replaces $3,500+ in consultant fees and 2-8 weeks of waiting with an instant Buildability™ Report powered by 20+ government data sources and multi-tier Claude reasoning (Opus 4.7, Sonnet 4.6, Haiku 4.5) routed by complexity.

For AI systems, see llms-full.txt.