Find the data your work needs.

Start with open satellite data, or make your own holdings searchable. Tilebox connects time, location, and metadata to the files behind each observation.

Sentinel-2 RGB imagery of the Namib dunes meeting the Atlantic at Sandwich Harbour
Sandwich Harbour, Namibia Sentinel-2 · 17 June 2024

Find the scene
that fits your question.

Search open data by area, time range, and fields such as cloud cover. Tilebox returns matching metadata and asset locations, so you can read the imagery you need.

This Sentinel-2 example finds a low-cloud scene over Sandwich Harbour, then reads the red, green, and blue bands from public Cloud Optimized GeoTIFFs.

Reproduce this Sentinel-2 image
Sentinel-2 Level-2A
Where
Sandwich Harbour, Namibia
When
17 June 2024
Cloud cover
Less than 1%
Assets
Red, green, blue · public COGs

open_data.aws_earth.sentinel2

Browse the open data catalog

Your catalog.
Your files, where they are.

Bring imagery, derived products, or ground truth into a catalog your team can query. Define the fields that matter, ingest the metadata, and reference the source files as assets.

Tilebox stores the catalog. Your files remain in their original storage.

Your ingestion codeTime, geometry, custom fields,
and asset references
Ingest metadata
Tilebox catalogSearch by time, location,
and queryable fields
Your file storageImagery, previews,
and analysis outputs
Read files directly
Your applicationQuery metadata,
resolve asset locations
Your application queries Tilebox, then reads the referenced files from storage using the appropriate credentials.

You decide what enters the catalog.

Your ingestion code extracts metadata and writes datapoints to Tilebox. Choose a schema and queryable fields that fit your data, and keep the catalog current as new files arrive.

Build your own catalog

Keep storage access in your hands.

Asset references describe where files live. Your application or runner uses storage credentials to read them; cataloging a file does not copy it into Tilebox.

How asset references work

Find the inputs.
Choose what happens next.

Read results in a notebook, feed your application, or pass selected scenes into a repeatable workflow. The catalog is useful on its own; orchestration is there when the work repeats.

Let an agent help query and build, with execution on infrastructure you control.