Kukun launches property knowledge graph API and MCP server

Kukun has launched a general-availability API suite and MCP server that return nine integrated data primitives for any U.S. property in one call. The company is targeting lenders, insurers and servicers that need faster underwriting, valuation and claims workflows.

Why it matters: - Kukun is packaging multiple property datasets into one callable system, which can cut the need for separate vendors and custom normalization work. - The launch is aimed at real estate AI and property-risk workflows, where speed and data completeness affect underwriting, valuation and claims triage. - The company is also positioning the product for AI agents and MCP-compatible frameworks, widening potential use beyond traditional software integrations.

What happened: - Kukun announced general availability of the Kukun Property Knowledge Graph API suite and its Model Context Protocol (MCP) Server. - The launch makes nine property primitives available through a single REST API call or MCP-compatible integration. - The system returns data for any U.S. address. - The API suite is live now at portal.mykukun.com/api. - The MCP Server is live and connectable from any MCP-compatible agent framework.

The details: - The nine primitives are permits, PICO condition score, AVM, renovation cost and after-renovation value, contractor history, system lifecycle forecast, investment outlook, market intelligence and property records. - The permits layer covers 800M+ building permits across 5,600+ jurisdictions. - Kukun says the permits data is normalized into 27 project categories, refreshed daily, and carries a median 8-day lag. - The permits product includes absence intelligence. - PICO is a 500–850 condition score for 110M+ U.S. properties. - Kukun describes PICO as a portable, national, proprietary condition scalar for residential real estate. - The AVM uses permit-enriched records and covers 110M+ properties. - Renovation cost data includes 26.9 billion data points across 146 categories at ZIP-code resolution. - Contractor history covers 2.4M+ verified U.S. contractors scored on 22 criteria from permit records. - System lifecycle forecast estimates remaining useful life for components such as roofs, HVAC systems and kitchens. - Investment outlook provides ZIP-code-precision growth forecasts one to three years out. - Market intelligence delivers demand and risk signals at ZIP-code resolution by property type. - Property records cover 120M+ U.S. properties and link assessor data, ownership history and parcel records to permit history and scoring models. - Authentication, provenance and audit metadata are included in every response. - The platform uses standardized taxonomy across 5,600+ jurisdictions so the same 27 permit labels apply from Los Angeles County to rural Michigan. - Every absence answer includes the permit label, date window and jurisdiction source. - Responses are schema-stable JSON with daily refreshes and an eight-day median lag.

Between the lines: - Kukun is emphasizing the join itself as the product, not just the raw datasets. - That framing suggests the company is competing on workflow simplification, not only on data coverage. - The MCP launch also signals a push to be native infrastructure inside agentic AI tools rather than just a data provider. - CEO Raf Howery said the company built the product for teams that are tired of stitching together multiple vendors to answer one property question.

What's next: - Kukun says enterprise licensing and custom data partnerships are available now through contact@mykukun.com. - The company says the platform will continue to support underwriting, valuation, claims triage and property-aware AI use cases. - Customers can begin using the API suite immediately through the portal and the MCP Server through compatible frameworks.

The bottom line: - Kukun is betting that one normalized property graph can replace a stack of point solutions for U.S. property intelligence.