Geospatial workflows weren't built to be portable. Every mission starts with the same scripts, the same notebooks, the same lost time. Fabric makes your workflows portable — so the work you did last mission compounds into the next one.
Onboarding in small cohorts as capacity allows. Typically 2–5 new users per week.
Area of Interest
Define the spatial extent and temporal window for your analysis. Draw a bounding box, upload GeoJSON, or generate boundaries from a place name.
Data Sources
Acquire raw geospatial data from satellite, radar, elevation, weather, and vector sources. Fabric handles discovery, filtering, and download.
Preprocessing
Standardize and align all inputs to a common grid. Resolves differences in CRS, resolution, NoData values, and radiometric characteristics across sensors.
Analysis
Compute derived metrics from preprocessed data. Spectral indices, vegetation analysis, band math, and domain-specific calculations, all drawn from a task library.
Post-Processing
Convert continuous analysis outputs into actionable results. Apply thresholds, generate statistics, and classify rasters into discrete categories.
Output & Provenance
Format final outputs for distribution and cataloging. Every run produces a STAC catalog with full provenance, from raw source to final product.
The Problem
The number of Earth observation satellites is set to almost triple by 2030.[1]Novaspace: Earth Observation Satellite Systemsnova.space ↗ Spatial data from drones, sensors, and open datasets is following a similar curve. Data access isn't the bottleneck anymore.
The bottleneck is the middle stack: ingestion, harmonization, preprocessing. Even when datasets share common formats like GeoTIFF or STAC, combining them means resolving differences in sensor characteristics, spectral bands, spatial resolution, map projections, temporal coverage, and preprocessing pipelines. Today, analysts assemble these workflows by hand — writing scripts in Python, chaining tools in notebooks, rebuilding the same data preparation steps for every new question.
The hours are real — and Fabric removes them. That's the obvious win. The deeper one is harder to see: without portable workflows, knowledge doesn't compound. Every mission starts from zero. When an analyst leaves, a workflow dies with them. When a project ends, the lessons evaporate. The data is abundant — but the intelligence built on top of it isn't.
The Fabric Stack
Visual Workflow Builder · Invite-Only Beta
Build geospatial processing workflows visually. Load a study area, select your data sources, compose a Pattern step by step — reprojection, band math, masking, spectral indices, temporal aggregation — and watch the full execution graph update as you build. No code required. Export analysis-ready outputs directly to your own cloud storage.
→ Learn about StudioProcessing Engine · API Access
The runtime that executes every Pattern — the hardened transformation and harmonization system underneath Studio. Engine handles ingestion, sensor alignment, projection, grid resampling, and temporal stacking across heterogeneous sources. For teams that need to run Patterns programmatically or integrate with their own pipelines, Engine exposes a direct API. It's also the surface AI-built Patterns will call when that lands.
→ Learn about the EngineIntelligence Layer · In Testing
Describe the analysis you need in plain language. Iris translates your intent into a validated Pattern, selecting the right sensors, recommending bands and parameters, and generating a complete pipeline in seconds. Currently in testing inside Studio's validation suite — available to invited users as it matures.
→ Learn about IrisHow It Works
Step 01
Draw a bounding box, upload a GeoJSON, or drop a set of coordinates. Fabric locks your study area and uses it to scope every data request downstream.
Step 02
Choose from the full Earth observation stack: optical, SAR, elevation, weather, and more. Fabric handles discovery, filtering by date range, cloud cover, and normalization automatically.
Step 03
Compose a processing pipeline from Fabric's task library: reprojection, band math, masking, spectral indices, temporal aggregation. Every Pattern is portable — reusable across missions, shareable across teams, version-controlled like code.
Step 04
Fabric executes your Pattern against validated STAC items, streams results to your storage, and returns analysis-ready outputs in the format your team needs: GeoTIFF, COG, or JSON.
Results
60–80%
of analyst time spent preparing data instead of analyzing it — the problem Fabric was built to remove.
5–10×
reduction in data preparation time, per internal benchmarks across representative Patterns.
MVP Phase Data
Closed-beta users are validating efficiency, time savings, and data harmonization across dozens of Patterns with 7 sensor combinations in a GIS knowledge base of 10 domain priorities and counterfactuals.
Common Questions
Your scripts don't travel. A Pattern does. Every step is inspectable, editable, and portable across missions — your collection of .py files can't say the same. When a teammate picks up your Pattern six months from now, they'll see the full provenance of every transformation, not archaeology inside a notebook.
No. Every Fabric transformation is logged with line-by-line provenance. You can inspect, override, or replace any step. If you need something Fabric doesn't do yet, you can drop back to your own code and bring the output back into the Pattern. Fabric removes the plumbing — it does not replace your judgment.
Your data lives in your cloud storage. Fabric delivers Cloud-Optimized GeoTIFFs, reports, and STAC catalogs to the account you specify. No data tax, no vendor lock-in. Cancel and you walk away with every byte you came in with.
Studio is in invite-only beta and onboarding in small cohorts. Engine is live and battle-tested; we are pacing access intentionally so every early user gets a real support relationship, not a ticket number. If you want to test Fabric against your own data, request an invite — we'll be honest about whether your use case is one we can support well today.
Supported Data
Fabric connects to the sensors and datasets that power modern Earth observation. Each source is downloaded, validated, and prepared automatically — so your team works with analysis-ready data, not raw files.
Optical Imagery
Thirteen spectral bands at 10-meter resolution, including three dedicated red-edge bands that no other free sensor offers. A 290-kilometer swath and five-day revisit mean you can monitor entire regions at the pace change actually happens.
ESA / Copernicus Programme
Synthetic Aperture Radar
C-band radar that sees through cloud cover, smoke, and darkness. Dual-polarization (VV+VH) at five-meter resolution in IW mode. When optical satellites are blind (monsoon season, wildfire smoke, polar winter), Sentinel-1 is still collecting.
ESA / Copernicus Programme
Optical Archive
The longest continuous satellite record of Earth's surface, spanning over fifty years of 30-meter imagery dating back to 1972. Landsat 8 and 9 now fly in tandem, delivering an eight-day combined revisit. No other source lets you measure how a landscape has changed across half a century.
NASA / USGS
Digital Elevation Model
Collected over eleven days aboard the Space Shuttle Endeavour in February 2000, SRTM remains the most widely used global elevation dataset. 30-meter posting with roughly 16-meter absolute vertical accuracy, covering 80 percent of Earth's land mass between 60°N and 56°S.
NASA / NGA
Vector Data
The world's most detailed open map: buildings, roads, waterways, land use, and administrative boundaries, maintained by over ten million contributors. Updated continuously, with coverage that often surpasses commercial alternatives in rapidly developing regions.
OpenStreetMap Foundation
Climate Reanalysis · Coming Soon
ECMWF's fifth-generation reanalysis, providing hourly atmospheric, land, and oceanic variables on a 31-kilometer grid, reaching back to 1940. Over 240 parameters including temperature, precipitation, wind, and soil moisture. The gold standard for climate context in Earth observation workflows.
ECMWF / Copernicus Climate Change Service
Weather · Coming Soon
High-resolution weather forecasts and historical data aggregated from national weather services worldwide. Up to one-kilometer resolution for recent data, with harmonized quality-controlled output. When you need current conditions rather than a decades-long archive, Open-Meteo delivers without API keys or rate limits.
Open-Meteo (open source)
Agriculture · Coming Soon
The USDA's Cropland Data Layer: a 30-meter crop-specific land cover map updated annually for the contiguous United States. Over 130 crop categories classified from satellite imagery and ground truth data. The authoritative source for understanding what's planted where across American farmland.
USDA National Agricultural Statistics Service
Sub-Meter Imagery · Coming Soon
Centimeter-resolution multispectral imagery from sensors like MicaSense RedEdge and DJI Multispectral. Fabric normalizes raw drone captures to calibrated surface reflectance, so your field data and satellite data speak the same radiometric language.
MicaSense, DJI, and other sensors
Interoperability
STAC (the SpatioTemporal Asset Catalog specification) is the open standard that makes geospatial data discoverable and interoperable. Fabric works with any STAC-compatible catalog natively, which means new data sources plug in without custom integration. If it publishes a STAC endpoint, Fabric can ingest it.
STAC Community Specification
Request an invite. We onboard in small cohorts as capacity allows — typically 2–5 new users per week. You'll get a direct line to the team while we're small.