Wherobots

Market position

Positions itself as the commercial vendor behind Apache Sedona, described as the most widely deployed distributed spatial engine with 68M+ downloads, and contrasts its offering with single-machine/traditional databases and with general-purpose cloud engines it characterizes as having incomplete geospatial support. It also participated in standardization work, driving GEO type support in Apache Iceberg and Parquet [0][2][3].

Authorship of Apache Sedona and full compatibility with it (existing Sedona code runs unchanged), unified handling of vector and raster data in one engine, lakehouse-native design on Apache Iceberg/Parquet with open table formats and no lock-in, serverless planetary-scale compute with stated cost/performance advantages, and an agent-oriented interface layer (MCP server, CLI, agent skills, Spatial AI Coding Assistant) that lets LLM tooling operate on spatial data [0][2][3][4].

Technology

Built on Apache Sedona and stated to be 100% Sedona-compatible so existing code runs unchanged. The stack uses Apache Parquet, Apache Iceberg (via the Havasu spatial table format), Spark SQL and PyTorch. WherobotsDB abstracts coordinate systems, projections and spatial indexing; RasterFlow provides serverless inference over imagery with built-in models including Fields of the World (field boundary detection), Meta CHM v1 (canopy height), Tile2Net (urban infrastructure) and ChesapeakeRSC (rural road detection), plus bring-your-own PyTorch models via the Machine Learning Model (MLM) standard. A built-in STAC reader loads catalog items into Sedona DataFrames. Published performance claims include 3x faster at 46% lower cost, typically 5-20x faster than popular cloud engines, up to 20x faster than alternatives, and up to 7x query acceleration in WherobotsDB 2.0 [0][2][3][4][5].

Go-to-market

Product-led and developer-focused: free tier and free trial, free certification courses (Geospatial Data Engineering Associate, Wherobots Fundamentals), extensive documentation, notebooks and solution recipes, an O'Reilly book on Apache Sedona, and community channels (GitHub, Discord, YouTube, LinkedIn, Twitter). Distribution and co-selling run through cloud ecosystems — AWS Marketplace, AWS Glue Data Catalog federation, Databricks Unity Catalog — and through the Apache Sedona open-source community, plus placement in the Anthropic Connectors Directory [0][2][3][5][6].

Data engineers, data scientists and ML engineers, and AI/geospatial teams inside enterprises working with large vector and raster datasets. Named industry verticals include real estate, finance, retail, insurance, agriculture technology, climate, mobility, automotive, logistics and supply chain, and data-product companies [0][3][7].

Geography

Headquartered at 4250 Drinkwater Blvd., Suite 300, Scottsdale, Arizona 85251, USA, per a third-party company directory. The platform itself targets global/planetary-scale workloads and offers deployment in customer cloud accounts [7][0][3].

History

2017: first release of Sedona/GeoSpark. 2020: Sedona joins the Apache Software Foundation. June 2022: Wherobots founded. June 2023: $5.5M seed round led by Clear Ventures and Wing VC. October 2023: launch of SedonaDB and Wherobots Cloud as a serverless offering. 2024: Series A (announced November 2024 at $21.5M), WherobotsAI for Raster Inference, Spatial API/SDK integrations and AWS Marketplace availability. 2025: GEO support driven into Apache Iceberg and Parquet, SedonaDB and SpatialBench released by the Sedona community, WherobotsDB 2.0, Databricks Unity Catalog and Amazon S3 integrations, and the Wherobots MCP server. 2026: RasterFlow general availability, AWS Glue Data Catalog federation, expanded VS Code and agentic developer tooling, and inclusion of the Spatial AI Assistant in the Anthropic Connectors Directory [0][2][3][6].

Risks & controversies

The sources contain no reported controversies. Performance and cost figures (e.g., 3x faster at 46% lower cost, up to 20x faster, 5-20x faster) are vendor-published claims without independent verification in the material provided. Third-party data is limited: one directory lists an employee range of 1-10 and withholds revenue and funding details behind registration [3][0][7].