• oceans
• climate_events
• maritime_movement
• fleet_vehicles
• electric_grid
• wildfire_risk

Apache Sedona

Apache Parquet

Apache Iceberg

PyTorch

Planetary-scale spatial compute

Spatial joins in minutes, not days. 300+ spatial functions covering vector and raster data, with native Spark SQL for tabular operations. 3x faster at 46% lower cost. Built by the creators of Apache Sedona.

Perception-as-a-Service

Run satellite imagery analysis at any scale, from a city block to the entire globe. RasterFlow delivers production-grade inference without building computer vision pipelines from scratch.

AI that understands your spatial data

The Spatial AI Coding Assistant works inside VS Code and every major AI coding environment. Discover geospatial datasets, construct spatial queries, and run production jobs through natural language.

No new tools. No new languages.

Write the SQL and Python you already know. WherobotsDB handles the spatial complexity: coordinate systems, projections, and indexing.

Lakehouse-native with Apache Iceberg

Upserts, inserts, and transforms on physical-world data at any scale. Open table formats. No lock-in.

100% Apache Sedona compatible

68M+ downloads. The most widely deployed distributed spatial engine in the world. Your code runs on Wherobots with zero changes.

Data Engineers

Build production spatial pipelines on the data that drives your business. SQL and Python you know. Scale you couldn’t get anywhere else.

Data Scientists and ML Engineers

Explore physical-world data in notebooks, build geospatial machine learning models, and run satellite imagery analysis at any scale. All in one environment.

G. Bailey Stockdale

CEO Leaf Agriculture

Arka Pratim Das

Sr. Manager, Software Development, Amazon Maps

Danan Margason

Founder & CEO at Aarden.ai

John Powell

Sr. Geospatial Data Engineer, AddressCloud

Jerome Urbain

Head of Products at GeoPostcodes

Industries depend on spatial data 

Score millions of properties against every hazard layer. Spatial risk models run in minutes, not days, with the auditability NAIC compliance demands.

Industries depend on spatial data 

Monitor millions of acres daily with satellite imagery analysis. RasterFlow turns raw imagery into crop health scores, field boundaries, and change detection at continental scale.

Industries depend on spatial data 

Process global vector datasets, fuse GPS with road networks, and refresh maps at the speed your customers demand. Built for the scale of Overture Maps.

Wherobots is the AI Context Engine for the Physical World. It gives data engineers, data scientists, and AI developers a unified platform for geospatial analytics, satellite imagery analysis, and spatial AI at planetary scale. WherobotsDB handles distributed spatial compute with 300+ spatial functions covering vector and raster data, with native Spark SQL for tabular operations. RasterFlow runs satellite imagery analysis at any scale. The Spatial AI Coding Assistant connects to VS Code and every major AI coding environment. Built by the original creators of Apache Sedona (68M+ downloads), 100% code compatible across all spatial functions.

Yes. Wherobots connects to your data wherever it lives without requiring migration. For Amazon S3, WherobotsDB reads and writes directly to your own S3 buckets using cross-account IAM roles. For Databricks users, WherobotsDB connects to Unity Catalog and runs geospatial analytics on data already in your lakehouse. For teams using AWS Glue, WherobotsDB connects to AWS Glue Catalog as an Iceberg catalog, letting you manage and query your existing Iceberg tables from Wherobots Spark workloads. Apache Sedona workloads running in AWS Glue, Databricks, or EMR run on WherobotsDB with zero code changes. Spatial SQL and Python pipelines you have already written work without modification. For Wherobots-native storage, tables are stored in Havasu, Wherobots’ spatial table format built on Apache Iceberg, which adds native geometry and raster column types, ACID transactions, schema evolution, and time travel.

PostGIS runs on a single machine and cannot scale horizontally. Teams that outgrow PostGIS face a hard ceiling on geospatial data analysis at scale. WherobotsDB handles planetary-scale distributed geospatial analytics across billions of geometries. Google Earth Engine splits raster and vector into separate systems with separate APIs. WherobotsDB unifies both with 300+ spatial functions covering vector and raster data, with native Spark SQL for tabular operations. Databricks runs spatial joins using 90+ geometry and geography functions, but don’t extend beyond these to the completeness Wherobots offers. According to the Apache Sedona SpatialBench benchmark at Scale Factor 1000, WherobotsDB delivers up to 20x faster geospatial operations than traditional big data engines. In production, teams have seen processing times drop from 39 days to under one day.

The Spatial AI Coding Assistant is available on the Visual Studio Marketplace. Install it there to write and run spatial queries directly from VS Code. For Claude Code, OpenCode, and other agentic development tools, install the Wherobots MCP server. The MCP server gives your AI coding environment direct access to Wherobots, including the ability to design queries against your spatial data estate, understand datasets in S3 and Databricks Unity Catalog, and run those queries on WherobotsDB. Support for additional coding environments is expanding.

Wherobots was founded in June 2022 by Jia Yu and Mo Sarwat, the researchers who created Apache Sedona at Arizona State University. Apache Sedona is the most widely deployed open-source distributed spatial engine in the world, with 68M+ downloads. It runs on Apache Spark and Apache Flink and supports spatial SQL, Python, Java, and Scala. WherobotsDB extends the capabilities of Apache Sedona and is 100% code compatible, meaning any Apache Sedona workload runs on Wherobots with zero code changes. Wherobots adds managed cloud infrastructure, RasterFlow for satellite imagery analysis at any scale, and the Spatial AI Coding Assistant on top of the Apache Sedona foundation.

Apache Sedona is used in production by over 20,000 organizations for geospatial data processing within distributed compute environments as an add on library. It’s used within Databricks, AWS EMR, Microsoft Fabric, Google Dataproc, and open source self hosted Apache Spark and Flink (among others). Wherobots is a fully managed, provisioned, and performance optimized runtime for Apache Sedona that is anywhere from 3-20X faster and more computationally efficient. Furthermore, Wherobots Cloud offers serverless runtimes for both WherobotsDB (our optimized version of Sedona on Spark) as well as RasterFlow, which is an independent engine for running computer vision models at in batch processing on GPUs. Beyond these two spatial data processing engines, Wherobots Cloud provides a managed Apache Iceberg catalog that has provided type support for Geometry, Geography and Satellite imagery long before open source Iceberg has caught up, and we drove the innovation within the Iceberg and Delta lake communities to provide this type support. Wherobots offers additional evolving capabilities including: and MCP server and AI coding tools, mapping and visualization tools both within notebooks and as a standalone product, model management for computer vision and other models, and advanced data management. For a complete overview of our product capabilities review our product pages.

Get started with a Professional account with advanced features, or talk to an expert if you have questions about your use case. New Pro subscribers can get a 30-day free trial with $95 in consumption credits.