EarthOne is a fully hosted geospatial machine learning platform from EarthDaily Analytics, the company that absorbed Descartes Labs. Users search petabytes of Earth observation data in a central catalog, prototype AI models in a hosted Python development environment, and deploy them at scale with up to 1,000 parallel workers. The catalog accepts raster, vector and time-series data, connects to STAC-compliant repositories and open datasets such as Sentinel, Landsat and MODIS, and links directly to the EarthDaily Constellation. Pricing is handled through sales, with a demo and consultation on request.
Data scientists and applied scientists are the core audience, since they can build science-grade models with few lines of code and skip infrastructure work. Technology leaders get a shared environment with cost visibility and control, while heads of geospatial operations can centralize data across global teams and make it usable for stakeholders who are not specialists. Enterprises in agriculture, insurance, mining, forestry, defense and civil government are the typical buyers, and ISO 27001 certification and AWS hosting address security reviews.
Typical projects start with data discovery, then move to training a model on satellite time series, and finish with a scheduled production job. An insurer might score flood or wildfire exposure across a portfolio, an agribusiness could run crop identification over a growing season, and a mining team could set up monitoring and alerting orchestration over sites of interest. Companies can connect their own data sources to the catalog so proprietary imagery and public archives sit behind one interface. Dedicated support and science consulting are available for teams applying AI and ML to geospatial data for the first time.
