Mireye is launching land search and parcel enrichment for AI agents that screen, underwrite, or acquire property in the United States. The San Francisco company, from Y Combinator's Summer 2026 batch, already exposes one API and MCP server to Claude, Cursor, and custom agents. Founder and CEO Ansh Chokshi ships each returned value with a source, URL, fetch time, and confidence, drawn from agencies such as USGS, FEMA, NOAA, and EPA, plus parcel records. Search looks across 160 million or more US parcels. An agent must include a state filter, and can narrow further by acreage, county, city, or polygon, then request fields such as owner and zoning on every match. The launch post says agents can filter and rerank using 550 data fields, then compare assets on structured records with sources. Search returns the first parcels that meet every filter, and the agent does the ranking. On the free plan a search query costs $0.20 no matter how many fields or matches come back. Enrich then resolves a shortlisted address or coordinate into parcel identity, ownership, and attributes, with confidence and source on each field. The new endpoints fit workflows Mireye already lists: data center siting, renewable screening, insurance underwriting, mortgage and title diligence, residential land, and commercial lending. An agent can filter a statewide set of parcels, then enrich the finalists with cited attributes. Fetch, the public coordinate endpoint, is now free. The launch post says it has already handled more than 4 million calls.