Mireye

AI agents like Claude or ChatGPT are frequently asked questions about physical locations, such as elevation, flood risk, or distance to infrastructure, but they only have general training knowledge with no real-time access, no source, and no timestamp. The page shows an example where Claude guesses an elevation range for Manhattan with no citation. Teams in sectors like data center siting, renewable energy, insurance underwriting, mortgage and title, land diligence, and commercial lending need precise, sourced answers to 'should I care about this place and why' questions, pulling from dozens of disparate federal and county datasets (USGS, FEMA, NOAA, USDA, EPA, and more). Doing this manually means stitching together multiple agencies' data sources, verifying currency, and handling edge cases like bad addresses, which costs time and introduces risk of citing stale or wrong data.

Mireye is an API and MCP server that lets AI agents ask natural-language questions about any US location and get back typed, cited answers with source name, source URL, fetched timestamp, and confidence level. It offers six endpoints: /v1/ask for natural-language queries, /v1/geocode to resolve addresses to coordinates with confidence scoring, /v1/lookup to resolve a single string into county, tract, district, elevation and flood status, /v1/fetch for bulk cited field retrieval across presets like flood_risk or solar_siting, /v1/proximity for drive-time and nearest-infrastructure analysis across the US and Canada, and /v1/field-requests to request new data fields that get matched against the catalog, built if missing, and made available to all future callers. Data is drawn from 96 authoritative sources, primarily US federal agencies plus county records and open datasets like Overture and Sentinel-2.