TerraflowwDB · private beta

Geospatial isn’t special.

TerraflowwDB is a multimodal database: satellite imagery, AI models and your own tables, queried together, in the tools your analysts already use.

Request access

flood_risk.ipynbPython 3

[1]In · Salesforce

sites = salesforce.query("SELECT Site, City, Value FROM Insured_Site")

Output

	site_id	city	insured_value
0	S-1042	Houston	$4.2M
1	S-2210	Chennai	$1.8M
2	S-0877	Miami	$2.9M
3	S-0415	Rotterdam	$6.1M
4	S-1733	Denver	$3.3M

[2]TerraflowwDB

import terrafloww as tfw
db = tfw.connect()
db.register("sites", sites)
radar = db.read("esa.earth_observation.sentinel_1")
site_radar = db.read("sites").with_column("radar", radar)
flood_model = tfw.load_model("acme.models.flood_risk", device="cuda")
scored = site_radar["radar"].infer(flood_model, output="flood_risk")

Output

	site_id	city	insured_value	flood_risk
0	S-1042	Houston	$4.2M	0.82
1	S-2210	Chennai	$1.8M	0.71
2	S-0877	Miami	$2.9M	0.58
3	S-0415	Rotterdam	$6.1M	0.24
4	S-1733	Denver	$3.3M	0.05

[3]Out · Salesforce

salesforce.update("Insured_Site", scored)

Output

✓ 5 records updated in Salesforce

01The difference

Not another geospatial platform.

Space data joins the stack you already run.

No second platform
No specialist team
No files to download
No pipelines to maintain

02Performance

Space data, as fast as the rest of your data.

TerraflowwDB is built for imagery from the storage layer up, so a question over satellite data runs like any other query.

70×

faster satellite analytics than open-source tools

100×

faster AI model inference over satellite images

Crop health for 250 farm fields

Commercial satellite imagery, same cloud machine

Open-source stack9–10 min

TerraflowwDB8 s

03Data and models

Space‑derived data and scientific models, one query away.

Providers publish once. Every enterprise on Terrafloww can query it.

Publishing data or models? Talk to us.

Data

ESA Sentinel-1 radar
ESA Sentinel-2
Commercial 1 m imagery
ISRO rainfall
JRC surface water

Models

Flood risk
Field boundaries
Partner models and algorithms

04Terrafloww Console

Ask in business terms.

Describe what you need and get working Terrafloww SDK code, with the datasets and models that fit. The SDK is plain Python, so your AI agents can write it too.

Workbench

UsesESA Sentinel-2acme_bank.lending.farm_loans

import terrafloww as tfw

db = tfw.connect()
farm_loans = db.read("acme_bank.lending.farm_loans")

red, nir = tfw.raster_band("red"), tfw.raster_band("nir")
summer_greenness = farm_loans.zonal_stats(
    "esa.earth_observation.sentinel_2",
    values={"ndvi": (nir - red) / (nir + red)},
    datetime=("2025-06-01", "2025-08-31"),
)

RunCopy

05Your stack

Works with the stack you already run.

Snowflake

Databricks

SAP

Polars

PyTorch

PostgreSQL

Parquet

STAC · COG

Parquet · Iceberg · Delta Lake · STAC · COG · your data stays in your cloud

06Terrafloww Console

Ready for the whole enterprise.

Roles for every team and dataset, an audit log, API keys, and usage and credits for every query. Private by default.

Usage Metrics

Total Requests

171

Success rate 100.0%

Average Latency

10.2s

Per query

Assets Queried

124

Datasets and models

Data Transfer

29GB

Read where it lives

Global Footprint

85.4Kkm²

Total area coverage

Usage Events

API keyResourceStatusCredits

risk-analyticsJRC Global Surface WaterSuccess30.00

agri-lendingESA Sentinel-2Success30.00

platformFields of the World modelSuccess42.00

07Access

Join the private beta.

We’re onboarding enterprise data teams and data providers.

Request access [email protected]